Geli’s ‘Internet of Energy’ Software Gets Real-World Test in Texas Microgrid
Out at a Department of Energy-funded Texas microgrid project, California startup Geli is getting a chance to test its "Internet of Energy" software for behind-the-meter batteries — plus a whole lot of other intelligent, networked end loads.
Funded by a $2.15 million grant from ARPA-E's "NODES" program, the project will eventually grow to encompass up to 100 devices, including batteries, electric vehicles, controllable HVAC systems, and binary-switched devices such as water heaters and LED lighting controls. It's happening at Group NIRE's facility in Lubbock, Texas, which also houses a state-of-the-art microgrid project from utility Oncor, and ongoing battery testing by DNV GL, Geli's partner in the new ARPA-E project.
Its goal is to show that distributed energy resources (DERs) like these can be put to use for two main purposes, which can sometimes come into conflict. First, they must serve as "synthetic regulating reserves," ready to curtail or inject energy to meet system-wide grid operator commands. At the same time, they’ve got to manage those operations to ensure that whatever they're doing for the grid at large can also support the local distribution grid they're connected to — a challenge that's much more granular in scale.
That means that Geli's software has to allow these aggregated DERs to provide critical information to grid operators and local distribution utilities — but not give away too much information, CEO Ryan Wartena explained in a recent interview. Specifically, they've got to be able to protect the price and capability data they're presenting to those parties in order to enable a true market for DERs to develop.
"We encourage our partners to start with the assets behind the meter," he said. To date, Geli has primarily provided its software to manage DERs at individual sites. "But we recognize when we're doing demand-charge management with a battery, we’re only using that asset 10 to 20 percent of the time. We want to make the rest of that available for other grid services."
"But because of data privacy, you don't want to make your whole load profile" public, he said. "You only want to advertise what you have available and at what price."
To walk this line between providing too much information and not providing enough, Geli has set up a multi-layered approach. It starts at the device level by creating a "forward availability profile" and "forward operating profile" for each battery or controllable load involved.
"We've structured it in a way that, in a forward-availability profile, the assets say how much power and energy they have as a cost function. But that cost function can be determined by anything," whether it's the price being offered for grid services, or an order from the local utility to take action to support local distribution circuits.
All of these devices, in turn, are connected to a "micro-market server" — a hosted software platform that takes all of the bids and commands from these external grid players, and translates them into actions for each device to take.
"We've developed this lightweight, fast and flexible methodology to participate, and have the notification, authorization, operation, validation and measurement to make it happen," he said. "Different assets can subscribe to it and different operations can subscribe to it, and we can carry out the balancing right there."
DNV-GL's role in the project is to provide the “suite of industry-validated and -approved modeling tools, from the wholesale and ISO level, to the distribution scale and end user," Michael Kleinberg, the senior consultant who's leading the new project, said in a recent interview. In other words, DNV-GL will be coordinating the entire project, mimicking the signals that would be coming from grid operators and local utilities, and validating how Geli's coordination works in response.
These will include "various high-penetration renewables scenarios," he said, to mimic the effects that cloud cover and wind speeds can have on a grid with lots of solar and wind power. These can have system-wide effects, as with Texas's wind power fleet, or local effects, as with Hawaii's distribution voltage challenges in rooftop PV-rich neighborhoods.
"The Internet of Energy concept is built on the belief that future energy systems will [comprise] massive numbers of distributed energy assets, which need to be coordinated in a large-scale fashion," Kleinberg said. Of course, grid operators "aren't going to want to reach into homes and manage the dispatch of these assets," he said.
DNV-GL, which is ARPA-E's chosen accreditor for grid and microgrid storage systems, doesn't dispatch grid signals as a business, he noted. That makes it a good third-party arbiter of how well different software-communications-hardware platforms and integrations work under real-world scenarios.
That includes finding the right set of economic incentives and the proper way to communicate their call-and-response between different levels of the interaction, he added. "Ultimately, these devices need to be incentivized," he said. Geli will be testing out ideas for this in the Texas project — "They're going to be receiving payments based on performance. How can they distribute those payments to individual devices?"
Here's a graphic that lays out how the whole system will be put together.
This isn't the only attempt at a multi-party, multi-layer approach to integrating customer-sited DERs with grid needs. The term "distributed energy resource management systems" (DERMS) is generally used as a catch-all to describe this type of technology, at least when it's deployed by utilities or on the demand side. GTM Research predicts the U.S. DERMS market will more than double to reach $110 million by 2018.
At the customer side, we call this kind of technology a microgrid — although not all microgrids are set up to control lots of DERs, or to work smoothly with the grid. GTM Research predicts the U.S. microgrid market will grow from less than a gigawatt today to 2.8 gigawatts by 2020, with a smaller subset of more sophisticated microgrids seeking to enable DER integration at scale. Microgrid experts agree that the industry needs an open standards-based software and hardware approach — and importantly, the testing and certification regime they could enable — to grow to scale.
It's a bit too early to describe what a standards-based, interoperable microgrid or DERMS software platform might look like. It's also hard to draw boundaries between these kinds of software architectures, or those being put in place by big energy storage contenders such as AES Energy Storage or Younicos, or Tesla and SolarCity, or Stem and Green Charge Networks, to name a few.
Geli stands in a smaller group of pure-play software vendors, active in the energy storage space. California-based Greensmith is managing more utility-scale battery projects across the country, including 20 megawatts in PJM, 2 megawatts with Duke Energy, and fleets of smaller-scale batteries in California.
On the distributed front, Seattle-area startup 1Energy Systems has gathered some big grid and battery vendors and big utility Duke Energy behind the MESA standards group. This year, it plans to test the same kind of complex grid-price balancing tasks that Geli will be doing in Texas, working with utility Snohomish PUD.
Both Geli and 1Energy have also targeted devices beyond batteries, moving into devices like smart inverters for rooftop solar, or electric-vehicle charging systems in homes or parking lots. Wartena noted that the Texas project will be its first connecting HVAC systems and lights, for example. As DNV GL’s Kleinberg noted, "The effort is to have a standardized interface from the device to the operating system. One of our initiatives will be reaching out to the vendor community, so if they'd like to see their device tested on this platform, they can integrate it."