by Jeff St. John
November 08, 2018

California’s big utilities see enormous promise for smart inverters in managing the grid edge. 

That value is only going to rise as California leads the country in requiring advanced inverter capabilities of all new solar installations, starting with simple autonomous functions, but eventually including real-time, two-way communications and control. 

But California’s investor-owned utilities also want the solar industry, regulators and customers to know that advanced inverters aren’t a one-size-fits-all solution to the grid’s problems. 

Sure, the simplest smart inverter functionalities — the so-called Phase 1 autonomous features required for all new California solar systems since September 2017 — can help reduce the local grid disruptions caused by increasing solar penetration. And the two-way communications capabilities required of all new solar installations starting in February 2019 will give utilities increased visibility into what’s happening. 

But what about the most advanced smart inverter capabilities, like tapping fleets of rooftop solar systems for real-time reactive power balancing, or aggregating inverter-connected household solar-battery systems to serve as a replacement for traditional grid upgrades? These still need a lot of work before they’re ready for prime time, from solving key communications and controls technology challenges, to building the regulatory structures to balance DER values with keeping the grid safe and affordable for all customers.  

These are some of the key findings from a new report, Enabling Smart Inverters for Distribution Grid Services, that summarizes what California’s investor-owned utilities Pacific Gas & Electric, Southern California Edison and San Diego Gas & Electric say they’ve learned from a multitude of smart inverter pilots over the past several years. 

“This experience has identified multiple factors for allowing smart inverter-enabled DERs to minimize potential grid impacts at higher DER penetration levels and, when cost-competitive, to provide distribution grid benefits such as deferral of utility investments, increased capacity, improved power quality, enhanced reliability, and greater resiliency,” the report states. But capturing these benefits will still require a lot of work from both smart inverter vendors and utilities. 

“I think, overall, we just want to be realistic,” Fedor Petrenko, PG&E’s grid innovation product manager, pointed out in an interview this week. “If these technologies are to be cost-competitive for electric customers, and we’re deferring infrastructure upgrades to deploy these technologies, we have to determine, is that technology cost-competitive?” 

PG&E has led much of California’s work on smart inverter testing through pilot projects funded by the state’s Electric Program Investment Charge program. These include its EPIC 2.03 A pilots in Central California’s Fresno County, as well as its EPIC 2.02 and EPIC 2.19 pilots in San Jose, which have already yielded lots of data on real-world challenges of smart inverter integration, and are expected to yield a final technical report early next year, Petrenko said.  

The new joint report also includes data from SCE and SDG&E’s smart inverter work, both in labs and in the field. And most critically, the new report also delves into the successes and failures of each pilot project, and uses those findings to lay out a set of recommendations for putting smart inverters to their most effective use. 

Early challenges: Customer acquisition, reliable communications, interoperability 

Pilot projects are meant to discover all the unforeseen challenges of deploying new technology in the real world — and California’s smart inverter pilot projects were no exception. Utilities encountered a host of problems, from customer enrollment struggles to technology failures, that will need to be addressed if smart inverters can be relied on as a grid asset.

PG&E’s EPIC 2.03A project in Fresno “encountered significant challenges in meeting customer acquisition objectives, leading to penetration targets ultimately not being met." SCE’s Integrated Grid Project EPIC 1 results showed “difficulty regarding customer participation in a program to monitor Smart Inverter-enabled DER data," according to the report.

And without enough customers participating, it’s hard to test whether smart inverters can have a significant impact on the circuits they’re connected to. “Field demonstrations by SDG&E and PG&E evaluated the aggregate effects of 400 kW and 62.5 kW of [smart inverter]-enabled [behind-the-meter] PV, which [accounted for] approximately 6% and <1% of the test feeders’ peak demand, respectively. Due to these low penetrations, aggregate smart inverter functions were observed to have little to no measurable impact on the distribution grid.” 

Issues of access will start to decline over the coming decade, as smart inverters proliferate in new solar projects across the state. But PG&E’s experience uncovered other potential pitfalls to customer acquisition, Petrenko noted. “When we were looking for another residential feeder to test in this current phase of EPIC 2.03A that we’re wrapping up now, because so many residential systems are financed in such a way that you can’t curtail the power output, or do anything but let it run 100 percent while the sun is shining, we couldn’t get any vendor to say, 'We’ll allow those assets to participate in your study,'” he said. 

Then there were the technical snafus encountered during the pilots. If they’re to be relied upon as replacements for traditional grid investments, “DERs should be readily available to provide distribution services with a comparable level of certainty that a traditional ‘wires’ upgrade provides today,” the report noted. But it also noted that “scenarios such as a communications outage that prevents the DERs from receiving commands or relaying data to the utility or aggregator must be considered.” 

That finding is based on real-world experience. In PG&E’s EPIC 2.03A Location 2 demo, the vendor’s in-field technology “routinely failed to recover from temporary satellite and cellular communications outages, requiring a manual reset to restore visibility and control of smart inverter-enabled PV systems,” the report noted.

As Petrenko explained it, “if there was a comms outage, not only would we lose the data for the duration, the aggregator system would lock up when it couldn’t call back to the cloud” — a situation that has since been resolved, he noted. 

SDG&E’s SI Demo C experience and PG&E’s EPIC 2.19C/EPIC 2.03A Location 1 demos showed that smart inverter communications reliability “was well below the average communication reliability for supervisory control and data acquisition (SCADA)-enabled devices, such as line reclosers,” the report noted.

And in PG&E’s EPIC 2.19C project, which involved solar systems, smart inverters and behind-the-meter batteries managed by Enphase and SolarCity (now Tesla), “PG&E found that communication between the storage aggregators and individual smart inverter-enabled storage assets was an ongoing challenge.”

This was not a finding unique to PG&E: “In multiple...residential demonstrations, communication to residential smart inverters via customers’ home internet in combination with ZigBee was not always reliable,” a not-uncommon problem with the low-power wireless technology used in many smart meter-to-home energy device communications. 

The issue of real-time communications isn’t as important when it comes to autonomous inverter features, or simpler grid applications that rely on DERs responding in preprogrammed ways to predictable events like system peaks. But it becomes critical when it comes to meeting real-time local grid challenges, Petrenko noted. 

“The robustness of the communications will have to follow the use case you’re implementing,” he said. In some cases, utilities may wish to rely on probabilistic methods to dispatch DERs with the understanding that some of them may be unable to respond at the time. But as the report noted, “any communication reliability and performance standards that emerge should also factor in the use case: Are the DER assets primarily operating autonomously (with infrequent remote settings changes), or do they need to be available on-demand for active control cases such as capacity or reliability?” 

Beyond communications, the smart inverter pilots have also revealed challenges in technical interoperability between different smart inverters and DERs and the utility systems that monitor and manage them.

According to the report, “recent...lab testing found that one manufacturer’s new smart inverter unit failed to initialize, another stopped functioning upon executing the latest Rule 21 firmware update, and a third shut down unexpectedly under normal operating conditions. [Utilities'] field experience also shows that many installers have not been able to properly set all smart inverter parameters to comply with Phase 1 SIWG requirements that came into effect in September 2017.” 

Even when the technology worked properly, differences in interpretation could cause unforeseen problems, Petrenko noted. In PG&E’s San Jose project, for example, the two participating system vendors chose different ways to respond when the utility asked batteries to inject more power than they has available at the time, he said — one vendor would dispatch the battery to discharge until it couldn’t anymore, while the other would simply choose not to respond. 

This is more of a procedural than a technical challenge, but it’s an example of one of the many potential miscommunication challenges that can arise when new standards are applied. In this case, the IEEE 2030.5 Smart Energy Profile 2.0 standards are being implemented for the first time. “That kind of process and procedure around how utilities communicate to aggregators needs some ironing out, especially if there are lots of aggregators you need to deal with,” Petrenko said.

This issue came to the forefront in PG&E’s distributed energy resource management system tests in San Jose, where the utility and vendors were forced to come up with extensions to IEEE 2030.5 to handle use cases that weren’t designed into the standards yet, such as bidding DERs into a test-scale distribution capacity market. “The way the aggregation solution using the 2030.5 protocol was set up, sometimes the data would arrive in a way that made it challenging for us to match it to certain assets,” he said. “It’s something that utilities figured out and worked through – but it’s not so simple, especially when it’s an aggregation where you might have hundreds if not thousands of inverters.” 

Time, location and effectiveness: The variables of smart inverter value

Solving these technical challenges will be critical as smart inverters grow to make up an increasing portion of the state’s distributed solar fleet. 

PG&E forecasts that by 2020, smart inverters will control half of the PV interconnected to its grid, and that the number of smart inverter solar systems will climb to 1 million by 2025. Across the state, smart inverter-enabled solar will grow to make up roughly half of the total by 2022, and become the norm by mid-decade. 

“If you look at the smart inverter penetration graph, you see that overall megawatt capacity is ramping up really quickly in California,” Petrenko said. “That means that on some circuits, we may start to see organically growing high penetration of PV with smart inverters,” capable of performing a range of functions 

The first benefit of this organically growing smart inverter base is in their “phase 1” capabilities — seven specific autonomous functions, such as not tripping off during minor voltage excursions and thus making them worse, and automatically adjusting power factor in case of emergencies. “What we’ve found is that smart inverter functions can definitely mitigate the voltage rise on the secondary, and let the utilities look into the low-voltage side” of their grids with more granular data, Petrenko said. 

That can help utilities better assess true hosting capacity on circuits, and help customers avoid paying for distribution upgrades for DER interconnection, for example. But importantly, these are all problems that tend to be driven by increased solar penetration on circuits, meaning that “autonomous functions do add value, but primarily in avoiding the issues that DERs caused in the first place,” he said. 

Whether or not smart inverters can be aggregated in large enough amounts to affect voltages on the primary circuits, where utilities operate their traditional voltage control schemes, is something being tested as part of PG&E’s extended pilot in Fresno. Unlike its San Jose pilot, this rural circuit is well-suited to this use case. “It’s a long feeder, it has a lot of motor loads like agricultural pumps and stuff like that that can kick on simultaneously and contribute to harmonics issues, and because it’s not a robust urban feeder, it’s more prone to those issues," said Petrenko.

This brings up the issue of location. Not all smart inverters will happen to be located on circuits that need their help, or can find a justification for paying them for more advanced capabilities. California’s Distribution Resources Plan and Integrated Distributed Energy Resources proceedings are working on ways to create values for the costs and benefits of DERs, and then to build them into the state’s multibillion-dollar distribution grid investment plans. These efforts will be critical to informing what values could be assigned to existing smart inverters, he said. 

Besides being in the right place, existing DERs will have to be able to respond at the right times to make a difference at the distribution grid level. The behind-the-meter battery systems in California today are largely being run to reduce customers’ demand charges, or to store on-site generated solar power to discharge for energy price arbitrage — and neither of those economic imperatives aligns with the needs on the distribution circuits they’re connected to, noted Petrenko. 

Time-of-use rates and specialized DER tariffs could help change this circumstance, he noted. “You can achieve some of those things with rate architecture that considers when the output of the DER is going to be most valued to the grid. If the rates are designed correctly, that’s a simple, elegant way to do it.”

“On the other hand, there are a lot of grid needs that can’t be anticipated using rule-of-thumb rate design,” he added. “That’s where active control use cases will be really critical for more critical services like grid resiliency and reliability.”

And, of course, utilities have their own work to do before they’re able to effectively make use of these smart inverters, as the report notes. “First, utilities will need new modeling and distribution power flow capabilities to better forecast the operations of and impacts from smart inverter-enabled DERs, in order to utilize the full benefits of smart inverter functionality,” something PG&E is doing in its EPIC projects and SCE is exploring in its EPIC 1 Integrated Grid Project.

In the longer term, “Operational capabilities and systems that can automatically analyze grid conditions, determine optimized solutions, and communicate signals to aggregators and DER assets are needed to enhance the value of DERs to the grid operator and planner,” the report noted. California’s utilities will need to complete their advanced distribution management system software deployments before they can start “safely and reliably accounting for DERs in distribution grid operations, and laying the foundation for active DER management to enable distribution grid services,” the report states.

These efforts are still several years out for the state’s investor-owned utilities, but Petrenko said, “I think these value streams are really interesting, because it’s there that the more forward-looking smart grid features start to come out. But it’s going to require a lot of upgrading of utility infrastructure.”