Utilities need cutting-edge distributed technology solutions to adapt to the accelerating changes happening on the edges of the grid -- and open-source programming and software-defined devices could be one answer. Certainly the smart grid industry is seeing more examples of how the computing revolution that has changed the server and smartphone industries is now beginning to provide models for converting yesterday’s proprietary, legacy grid technology landscape into a more internet-like IT architecture, and unlocking the possibility of innovations to come.
Last week we covered Duke Energy’s ongoing work with a coalition of vendors to enable a new level of grid device interoperability and flexibility via open-source software and distributed intelligence. This week, we’ll take a look at another example of this emerging grid model, featuring multi-purpose field computing devices and open standards-based, peer-to-peer protocols to linksolarsystems, batteries and distribution grids into a real-time, self-balancing network.
The test bed is Toronto Hydro, Canada’s largest municipal utility, which is facing significant challenges in incorporating its growing share of customer-owned solar PV and regional wind power into its constrained grid. Its partners on the project are National Instruments, the digital technology developer with $1.3 billion in revenues, and LocalGrid Technologies, a startup with networking and data management technologies built with grid edge needs in mind.
In April, the partners launched a project, backed by $2.5 million in Canadian government grants, aimed at creating a “next-generation energy management system." The hardware linchpin of the new system is National Instruments’ CompactRIO -- a ruggedized field computer with advanced analytics and processing power, and the flexibility to be reprogrammed to serve a variety of as-yet-undefined functions, that could go beyond most grid devices today.
LocalGrid’s contribution is the application of a standards-based peer-to-peer messaging protocol known as DDS, along with its own patent-pending “DataFabric” technology to manage multi-protocol data exchanges amongst disparate grid devices. In combination, they’re meant to allow a network of grid endpoints to manage local, real-time grid conditions on their own, as well as to communicate with Toronto Hydro’s central control platform and the rest of its grid infrastructure.
In other words, it’s a grid architecture built to do a multitude of edge-of-grid functions at speeds today’s centralized systems can’t attain, based on a set of software-defined parameters that can be adjusted over time to meet future needs. Here’s a breakdown of how the parts fit together, starting with NI’s CompactRIO technology.
The CompactRIO as Future-Forward Grid Edge Platform
Brett Burger, senior product marketing manager at National Instruments, described the CompactRIO as a “fully programmable, software-defined embedded controller” built for the same harsh outdoor conditions and instrumented to accept the same high-fidelity sensor measurements as a number of typical grid devices.
“We’re talking about relays, RTUs [remote terminal units], and PMUs [phasor measurement units], and all the things you need to automate to manage the grid,” he told me in an interview earlier this month. But unlike the vast majority of these purpose-built grid devices, “it’s fully programmable,” he said. “The user -- the engineer, the scientist -- gets to define its personality.”
The term of art here is a “software-defined” piece of computing hardware, akin to the smartphone hardware platforms built to run Google’s Android operating system. That’s in direct contrast to the computing hardware that goes into most of today’s grid equipment, which tends to be designed to do the specific jobs required of it at as low a cost as possible. That makes sense for grid vendors trying to capture large markets where cost competition forces them to avoid adding unnecessary computing power, he noted -- but "everyone's grids are so different now. [...] The market is moving toward a one-size-does-not-fit-all" model.
National Instruments, by comparison, has “always been the odd-job type company: if a device does not exist for the needs you have, we have the tools to let you build it yourself,” he said. In the case of the CompactRIO, that includes key integrated circuits, known as field programmable gate arrays (FPGAs), which permit the devices to “change personality through software,” as opposed to less sophisticated application-specific integrated circuits (ASICs), which, as the name implies, are “made for just one thing,” Burger said.
CompactRIO also comes with a variety of operating system flavors, he added. Many customers are using National Instrument’s LabView software platform to integrate transformer monitors, inverter control systems, solar panels and other grid devices. Others, including LocalGrid, are also using the Linux-based source code and operating system released for the CompactRIO last year, he said.
Much of this innovation is going on quietly, under the hood of broader vendor projects, Burger noted. For example, Lockheed Martin has built microgrid controllers on the CompactRIO, he said. Traditional grid devices could be designed around similar concepts and components, as could next-generation equipment such as advanced inverters or distributed grid control nodes, he said.
“I think we’ll see a mix of devices like these,” he said. Grid vendors will “still see a need for their fixed-functionality, ultra-low-cost devices…but they’ll also merge into their portfolios these intelligent, software-defined devices that will give them a market advantage over their international competitors.”
LocalGrid’s DataFabric for Peer-to-Peer Edge Intelligence
As for LocalGrid, it had National Instruments' hardware in mind since its parent company, Prolucid Technologies, was founded in 2007. “One of the first things we did was go down to our regional NI office and say, 'We want to be a partner,'” Bob Leigh, CEO of the Mississauga, Ontario-based company, told me in an interview last week.
The idea was to create embedded software and applications for specific industries, which in Prolucid’s case ended up being the medical sector, and in the case of LocalGrid, the utility sector. The startup has been working with Toronto Hydro since 2012 to “develop an embedded software solution that’s hardware-agnostic, to solve the problem of interoperability for utilities,” he said.
Toronto Hydro has a history of developing internal integration solutions to its grid IT needs, from transformer health monitoring to outage detection. It’s also piloting a community energy storage battery system that could be put to use in balancing its increasing supply of on-again, off-again renewable energy.
But traditional back-office, enterprise service bus integration methods can’t be relied on to do real-time distribution grid applications, like using batteries to balance solar intermittency. What’s needed is the ability to share the data in the field, via what Duke and its partners have labeled a “field service bus,” and what LocalGrid calls its “intelligent node” network.
The first challenge is to create a messaging protocol that can tie multiple devices together in as close to real time as possible. LocalGrid chose a technology known as Data Distribution Service (DDS), a secured messaging protocol originally developed by the U.S. Navy to connect shipboard IT systems in a real-time, peer-to-peer network, Leigh said.
One of the key problems that DDS helps solve is the translation of multiple protocols used by different devices on the grid, such as Modbus for solar or battery inverters and DNP3 or IEC 61850 for SCADA-connected devices, he noted. Unlike other protocols that require a “broker” device to do that translating, DDS “can achieve that sub-second messaging rate, even as you’re translating protocols on either side,” Leigh said.
The other key part of LocalGrid’s system is its DataFabric software, which consolidates an array of data points into a common data model, and allows for each software-defined endpoint to “aggregate and change resolutions of data on the fly,” he said. That’s important for condensing large amounts of grid device data into smaller sets that can be traded back and forth at the speeds needed to do real-time grid state calculations, as the first phase of the Toronto Hydro project will be doing, or even implement control decisions, as the partners are contemplating for future phases.
Because each CompactRIO endpoint is inherently flexible, LocalGrid can provide “protocol conversion which we can update dynamically over the air, analytics that we can update to the system, and run multiple applications on the same device,” he said. This is similar in intent to the kind of field-distributed computing capability that Silver Spring Network’s new SilverLink Sensor Network platform and Cisco’s new IOx platform are opening up to partners, but it’s pretty far ahead of the capabilities of the vast majority of today’s grid edge devices.
In fact, in terms of technology that allows interoperability without a lot of expensive and complex pre-integration work, “The existing players do not have solutions that will do this job,” Leigh said. “They’re not fast enough, they’re not open enough, or they don’t have solutions that are cost-effective enough in the distribution space.”
So far, LocalGrid has connected four sites with a combination of solar PV and wind turbine inverters and metering hardware, and is now in the midst of its second phase of developing custom algorithms for tasks such as detecting faults and forecasting solar and wind generation and loads on distribution circuits, Leigh said. These aren’t necessarily huge challenges for Toronto Hydro’s existing IT infrastructure at pilot scale, “But if we were to multiply that across the network, it’s just not feasible to get all that data to be analyzed into a back-end system,” he said.
As for how to expand LocalGrid’s software capabilities to a broader set of grid endpoints, Leigh cited Cisco’s IOx-enabled grid routers as potential future partners. Other big grid vendors like General Electric, ABB and Siemens “are at different stages starting to open up their systems,” he said. “The question that still has to be worked out is how much third-party development can take place on their new systems.”
That’s the same question that Duke has been asking the grid vendor community, via its plans to open its source code and hardware adapter reference designs to the public. The past half-decade has seen open-source grid systems emerge from simulation software and data management tools to a few real-world grid applications, albeit still in the experimental stage. Perhaps the next half-decade will see the open grid edge platform attain real-world status.