For the past year or so, giant U.S. utility Duke Energy has been engaged in a potentially revolutionary project: getting some of the world’s biggest smart grid technology vendors to open up their systems, akin to how iOS and Android have opened the world’s smartphones.
Now Duke is recruiting new utility and vendor members as it publishes its open-source code manifesto.
The work started at Duke’s McAlpine pilot project in Charlotte, North Carolina, where the utility and its partners formed the “coalition of the willing” -- Accenture, Alstom, Ambient, Echelon, S&C Electric and Verizon -- to connect smart meters, solar inverters, battery controllers, distribution grid control systems and network nodes, all with open-source adapters that render them capable of “talking” to one another and make split-second grid balancing decisions.
At this January’s DistribuTECH conference, Duke and its coalition partners did their first off-site demo, showing how voltage sensors, fast-reacting grid support, and battery and inverter energy injection could smooth out a mockup of a set of grid-edge scenarios -- say, clouds passing over a big solar PV array -- in ways that centrally controlled systems simply couldn’t manage.
This kind of grid-edge, virtual power plant capability typically requires a large amount of expensive and complicated work to pull off, with a lot of back and forth between different vendors sharing software development, data modeling and physical communications integration tasks.
But for Duke’s project, since each vendor has “unlocked” its device-to-network interfaces to interact with open-source-based code, the time to integrate has dropped from weeks to days, “all at an incremental cost of about $50” per new integration task, said Raiford Smith, Duke’s director of smart grid emerging technology. These order-of-magnitude improvements have allowed Duke and its partners to push ever more complex solutions to grid-edge challenges at its McAlpine test site, he said -- just the kind of speed and flexibility required to handle grid problems that are just beginning to emerge.
For solar smoothing, “We were able to pull the data from the previous day, tweak the code, recompile it, and recode it, all in a single day,” he said by way of example. When new meters came in for integration, “It took us about a week to unlock a meter -- and then all the meters from that vendor were unlocked.” In other words, “That’s rapid prototyping, all at a very low cost.”
Now Duke and its partners want to invite a new round of participants to the open source revolution. In the next month or so, they intend to publish their open-source code for the field message bus and device adapters they’ve developed. Next to come will be a hardware reference design to help jump-start development work, as well as active outreach to a whole host of grid vendors.
Many massive steps remain between today’s mostly proprietary, limited-flexibility grid devices and a future of open computing grid platforms, of course. But at the end of the road could lie a transformation similar to the one that has come to smartphones over the past decade, Smith said -- a once-thick array of proprietary contenders, fallen before the might of Apple iOS and Android, and the hardware makers who have been able to adapt and thrive in the new landscape.
Both opportunity and threat are implied in that comparison. “This effort isn’t a subversive attempt to get vendors in trouble and blow up their market,” Smith said. Instead, it’s more of a friendly nudge, along with a helping hand, to remind grid vendors that utilities are facing unprecedented challenges on the grid edge -- and that “if traditional utility business models have to change, traditional vendor business models also have to change.”
Opening the grid edge to the open-source (r)evolution?
This is exactly the kind of nascent, yet potentially disruptive technology development that Greentech Media will be covering at our Grid Edge Live 2014 conference in San Diego this June. What’s more, rather than pertaining to a specific industry vertical such as smart metering, distributed generation, demand response or energy storage, the move toward open-source computing underscores them all. That’s because every device in the smart grid has a computer at its heart -- and that computer can either be built to serve the needs of the past, or of the future.
Up until recently, most grid devices have been built to provide a specific set of grid sensing, control and automation functions at the lowest possible cost -- all for a utility sector that’s relatively tiny compared to mass-market mobile or computing markets. Further, utilities replace their grid technology products every few decades, not every few years, as is common in the smartphone realm. And because the utility industry is dominated by legacy, proprietary systems, there’s less pressure to adapt to new open-standards-based paradigms.
None of this drives grid vendors to open up their technology platforms. To be sure, the latest smart grid devices are likely to have a smartphone-equivalent level of processing power, memory, and networking capabilities. But they’re not necessarily designed to support functions that weren’t part of their initial use case -- or to expose their inner IT workings to open-source modification by third parties, even trusted ones like Duke.
New standards, such as those coming out of the Smart Grid Interoperability Panel (SGIP), are helping to define different ways that grid devices should be able to work together. But as Erich Gunther, chairman and CTO of grid technology consultancy EnerNex, noted, “Historically, vendors told their utilities, 'Hey, here’s my product; figure out what to do with it.'"
So far, “Duke is the largest utility pressuring vendors to expose their systems for interoperability. That’s a big deal,” Gunther said. The catch for Duke and its partners is to stick with what he called “well-defined points of interoperability -- that means not only well-defined, but standards-based, and secure,” he said.
“We want to encourage the vendor community to expose those interfaces, to help us create the internet of things, as Cisco would call it, using industry standards, such as those the SGIP is documenting and promoting,” he said. “What we don't want to see is vendors who haven’t bothered to adopt any standards,” he explained. “We’ve found in other industries that that ends up costing money over the long run.”
To be sure, Duke’s partners don’t have products on the market yet. They’re using open-source adapters for the time being. Duke itself has relied on off-the-shelf hardware unsuitable to grid deployment -- first, the nonprofit-built kit computer Raspberry Pi, and now the Odroid platform from Korean open-source hardware developer Hardkernel -- to do its R&D work so far. Certainly new certification and testing regimes will be required to bring commercial-grade systems into play.
Competing architectures: iOS or Android as a model?
In the meantime, grid vendors and technology providers are enabling their smart meters, grid routers, advanced inverters and energy management devices as mini-computers, capable of analyzing data and making decisions in a more decentralized way. In fact, DistribuTECH saw quite a few announcements on this front.
Take Silver Spring Network’s new SilverLink Sensor Network platform, which allows third parties to write RESTful apps to run on its networked devices, or Cisco’s new IOx platform, which opens its IPv6-based grid routers up to Linux-based applications. Platforms like these are being tapped by a growing list of partners, ranging from grid device vendors like Alstom and Itron to data management and analytics providers like OSIsoft and AutoGrid.
But even these aren’t quite the same as what Duke is doing, according to Jason Handley, technology development manager for Duke’s emerging tech group. That’s because both of those platforms rely on either Silver Spring or Cisco’s underlying hardware, and require software developers to make use of application programming interfaces (APIs) to tap that hardware’s capabilities, he said.
The analogy from the mobile world is Apple’s iOS, he added -- a platform that’s open to applications development, but closed to tinkering with the operating system. Duke is “pleased that Cisco has come out with this IOx platform, because they see the ability to open up in part, at least,” and Duke is working with Cisco and other vendors developing more open systems like these, Handley said.
At the same time, an Android-style approach is independent of the underlying hardware, in the sense that it exposes that hardware to be “reprogrammed” with the open-source-based operating system of choice. While that’s a big step for vendors to contemplate, the mere existence of Duke’s coalition indicates that it’s not beyond the realm of possibility for more vendors to make this significant change.
“We believe that we need to kick-start something, and Duke has taken a stand in saying, 'This is how we’re going to do it,'” said Handley. Since the DistribuTECH unveiling, “The interest level has already gone through the roof with these traditional utility manufacturers, [which are saying] 'We believe this has some place in our business,'” he said. At the very least, “We’re not getting ‘Nos’ anymore, which is interesting.”
Meanwhile, Duke is seeing a lot of new entrants into the market, such as Texas Instruments and National Instruments. Those companies recognize that "if we start to push computational power to the edge, we’re going to have a new marketplace,” he said. “We’re testing a few products now that look really promising -- they’re almost like slotted card devices,” able to adapt to new communications and applications in ways today’s more limited, less flexible grid electronics aren’t built to handle.
That opens up opportunity for new “software-defined” forms of traditional grid devices -- relays, reclosers, cap bank controllers, RTUs, IEDs, PMUs, smart meters, you name it -- that could represent a brand-new grid architecture. We will have more on the subject of vendors creating the building blocks of this open computing grid architecture. Stay tuned.