Insanely complex math, and never enough data -- that’s the conundrum in trying to model the ebb and flow ofsolarpower and energy storage on the grid edge.
Large-scale transmission systems are well modeled. But the majority of the grid below the substation provides little to work with beyond original engineering specs -- and hopefully, up-to-date maintenance and replacement records -- to turn into useful data for software platforms.
That makes it very difficult to capture or predict voltage anomalies, reactive power problems and other disruptions coming from customer-side energy assets like rooftop solar. It also makes supply-demand forecasting to optimize the interplay of solar power, energy storage and grid interconnection requirements almost impossible. Even so, with smart meters, grid sensors and advanced inverters starting to populate the grid edge, there are new tools that could make this micro-scale grid modeling a truly useful tool.
Power Analytics, a supplier of complex power flow modeling software for customers like the U.S. Navy and the FAA, believes its new platform is ready to take on the challenge. It’s called Paladin PV (PDF), the newest piece of the San Diego, Calif.-based company’s Paladin software suite. According to Kevin Meagher, Power Analytics’ CTO, it has the “potential to significantly change the landscape” for distributed solar’s integration into the grid.
Paladin PV stems from work Power Analytics has been doing for the Department of Energy’s SunShot program since 2011. The first part builds on the power flow modeling it’s done with the University of California, San Diego, which has a world-class microgrid including combined heat and power, solar PV and batteries.
The second part adds a key component of the solar-equipped grid: the inverters that convert DC-powered solar panels and batteries to grid-ready AC.
“We came up with a [method] to mathematically model an inverter in two ways,” said Meagher. "First as a real generator on the power line, and second to integrate the solar irradiance data” that informs just how much sunlight is being converted into the “fuel” that powers all that PV.
Here’s a diagram that represents the working parts of a Paladin PV system:
The first part of the equation is the “Total Sky Imager” system used by UCSD for its SunShot-backed high-penetration solar portal program. The camera and hemispherical mirror system captures cloud cover as it moves across the sky, with a panoramic aspect that can see what’s coming and predict just when it’s going to start shading that PV array.
“We’re talking about solar forecasting in less than fifteen minutes,” compared to more typical hourly forecasts from satellite data, he said. That localized advanced weather warning allows planning ahead for sudden drop-offs or surges in generation, and the effects that has on local grid conditions like voltage and power factor changes. It also could offer up to 98-percent-accurate forecasts of what a solar array is going to produce at any given point in time, according to a Power Analytics report.
From there, Paladin PV links that irradiance data to its models of how each PV inverter converts that solar “fuel” into electricity, Meagher said. That starts with generic inverter calculations, and is trued up by modeling the actual inverters being used in combination with the actual cable sizes, efficiency losses across different operating parameters, and other tricky real-world calculations, he said.
That includes things like voltage, current, and power factor conditions for the project developer and operator. It also includes a real-time tally of how those are affecting the utility at the grid interconnection. The final element is energy storage -- analyzing how batteries and inverters interact with the system, and figuring out how many of them are needed at any given moment of the day, projected out into weeks, months and years into the future. That enables the system to “come up with a size for the energy storage that’s a lot more precise,” said Meagher.
These aren’t simple tasks, as this generic inverter power flow model from Power Analytics DesignBase software indicates:
The goal is to provide a software representation of a real-world system, one that’s directly tied to the real-time operations for keeping that system in balance.
“That kind of accuracy is the basis for the range of planning that would be part of a study, and would be part of an interconnect agreement with a utility -- and when we’re in the real-time model, that same platform is operating,” Meagher said.
With other projects, the modeling for utility interconnect approval and the real-world monitoring and control schemes are separate affairs. That leaves quite a bit of uncertainty between what the model predicts should happen, and what’s actually going on, and lessens the value of that model as a real-world operating asset.
With Paladin PV, by contrast, “We have a very accurate mathematical expectation of what we should be seeing, moment by moment,” he said. “If we don’t see that -- if it’s off by a couple of percentage points -- we know where to look and why.”
This isn’t a concept unique to Power Analytics, although it's the kind of expertise that won that company a spot on Greentech Media's Grid Edge 20 list of companies to watch. We’ve been covering the innovations happening on the distributed, two-way grid edge, with startups like Spirae, Integral Analytics and Smarter Grid Solutions working alongside grid giants like Alstom, ABB General Electric, Hitachi, Siemens, Schneider Electric and Toshiba to sense, model and manage these new distributed systems.
Solar forecasting as a new technology market is on the radar of companies ranging from weather data providers to third-party PV aggregators like SolarCity and Sungevity, or solar metering providers like Locus Energy and Itron.
As far as deploying Paladin PV goes, Power Analytics is doing the software and consulting, and relying on design and implementation partners to build its automatic control capabilities, Meagher said. That involves partners like Colorado-based Homer Energy, a spinout of DOE’s National Renewable Energy Lab (NREL), which focuses on financial modeling to prove out the value of different combinations of assets for microgrid deployments.
But the kind of model-to-operations integration that Paladin PV is promising is also becoming more important for traditional large-scale solar projects, he said -- perhaps not today, but in the near future.
“If it’s a utility buying power from you, and you can’t show the ability to respond to cloud cover and ramp rate control, it significantly [decreases] the value of it on the PV side,” he said. That’s easier to contemplate in megawatt-plus PV installations, where the costs of boosting the system are matched by economic or interconnection incentives to do so.
Smaller projects are harder to pencil out, but “because of the drop-off of tax credits and stimulus funds, people are more interested in the performance of PV” on that scale as well, he said. Islands like Hawaii and Puerto Rico are already starting to demand some ability for solar farms to provide more stability, and to allow standby resources with warning to ramp up and down to meet their fluctuations when they do occur. California is driving smart inverter regulations that could unleash a range of voltage and reactive power support services from the everyday solar PV array.
At the same time, better modeling and more accurate forecasting could help push far more distributed rooftop PV into the grid than utilities think is possible today, said Meagher. “For years and years, everybody talked about 15 percent, which is kind of an arbitrary number,” as a limit for how much solar a local section of the grid could safely handle. That artificial limit has already been breached, however, with no dire effects -- “they’ve already gotten there in San Diego, a lot sooner than they expected,” he said.
The problem is, “they don’t have any data to tell them what’s going on,” he said. “The work that SunShot has done has been to greatly increase this” amount of data available to boost PV’s modeled, forecast and managed performance, along with the smart inverter integration to smooth it out locally.
“Maximum penetration could be 40, 50, 60 percent,” he said. “The more you know about it, the more you can manage it on the margins.”