The largest smart grid demonstration project in the U.S., covering about a dozen utilities in five Northwest states, has shown how adding more information at all levels of the electricity system can improve efficiency and potentially lower costs.
But the project, which wrapped up data collection last summer and is now sharing results, also highlighted the challenge many industries face as they embrace the Internet of Things, deploying more sensors and connected devices that unleash a torrent of data: Ensuring that data is reliable is difficult. Drawing meaningful insights from it can be harder still.
The Pacific Northwest Smart Grid Demonstration Project, a five-year, $178 million effort funded by the federal government, utilities, and participating technology companies, included a host of component pieces from modern electricity meters to controllable thermostats to smart transformers. Perhaps most importantly, researchers designed and deployed a new signaling technology meant to allow these and other devices to rapidly respond to the needs of the broader electricity system.
This “transactive control” signal could help form a new market to incentivize new, more flexible technologies—potentially installed in individual customers’ homes and businesses—that would allow the electricity system to accommodate more renewable energy.
Good Data, Bad Data
At its essence, a smart grid provides more information to all electricity producers, distributors, and consumers, allowing better decisions and improved automated operations—all with the goal of reducing costs and improving efficiency.
“That’s our hope and the promise of the technologies,” said project director Ron Melton, on staff at the Batelle-operated Pacific Northwest National Laboratory, which headed the project along with the Bonneville Power Administration.
But producing and delivering high-quality, reliable information can be challenging. “One of the things we’ve learned on the project is that it is not easy to operate some of these technologies and some of this equipment in a way that produces good quality data all of the time,” he said, noting that this is not unique to the Pacific Northwest project.
“Data quality, data integrity is a problem right now,” Melton said. “We need to be developing tools and technologies for both the utilities and vendors to make sure that the data being generated by these information-enabled devices is good data, that they’re operating correctly, and the data is being captured and stored properly.”
Smart grid builders and operators face some of the same problems cropping up in many other industries, as new sensors and devices proliferate, each one throwing off streams of data that can pile up faster than they can be used. “The ability to effectively apply the data just from that point of view is a challenge in and of itself,” Melton said.
Utilities and their customers installed some $80 million in smart grid equipment as part of the demonstration project.
Spokane, WA-based investor owned utility Avista, for example, upgraded several aspects of the electricity distribution system in Pullman, WA, giving the utility instant alerts when problems occur, improving efficiency by using appropriately sized transformers, and installing thermostats the utility can remotely adjust (with customers’ consent).
Before the project, the University of Washington had only seven electricity meters monitoring flow into its main Seattle campus, which includes some 250 buildings and 14 million square feet of space. It now has about 200 meters that allow a much finer-grained view of where and how electricity is used. That, in turn, enables improved energy cost management and efficiency improvements, Melton said.
Portland General Electric, another investor-owned utility, built a Smart Power Center in Salem, including a 5-megawatt lithium-ion battery, connections to a local solar project, and new controls technologies to establish a micro-grid. Customers nearby can voluntarily reduce electricity usage to avoid times of peak power demand.
Of course, not every technology worked perfectly, and some things tried across the demonstration project’s 13 sites had to be taken out of commission, Melton said. But that’s part of the reason for demonstration projects like this one.
“Many of the utilities have learned a lot and are now able to progress in terms of making investments in additional equipment and technology with knowledge in hand,” Melton said.
Other improvements have been made behind the scenes, he said.
With the growing threat of cyber attacks on infrastructure such as the electricity grid, the project focused on security and interoperability. The project documented current best practices and encouraged utilities to adopt them. Melton said the larger utilities were already well on their way, but some of the smaller ones were confronting new security threats along with new technology, such as smart meters, which connect their customers’ sites directly to utility back-office systems.
“We upped the game for many of the participants and improved their cyber-security awareness and capabilities,” he said.
At the heart of the project was a set of two-way signals meant to communicate the expected balance of supply and demand on the electricity grid in the immediate future, allowing utilities and other players to respond accordingly. One of these is an incentive signal designed to indicate the price of delivering power to a certain location at a certain time.
The information in the “transactive control” signals is key to enabling demand response, in which utilities or even individual customers adjust their electricity usage in real time to keep the broader system in balance—a crucial, ongoing challenge to maintaining reliability. This balance is achieved through careful scheduling and management of power plants—the supply side of the equation—with substantial reserves in place in case of unanticipated changes.
Greater control of the demand side of the equation gives electricity system operators more flexibility to handle the unexpected, and to integrate more renewable energy from wind and solar plants, whose output can change suddenly with a turn in the weather (though forecasting has come a long way, making this variability less severe than it once was.)
As part of the smart grid project, a computer simulation took data from independent power producers, the Bonneville Power Administration (BPA), and other sources to create a signal that reflected different scenarios on the Northwest grid. These simulations included a rapid increase in wind power generation from an approaching storm front, or an unexpected loss of power from the Columbia Generating Station, the region’s sole remaining nuclear plant.
The goal was to see whether the incentive signal behaved appropriately, going down in value as power becomes abundant, and going up when it grows scarce. Utilities could respond to the signal by increasing electricity demand—charging up batteries with the low-cost wind power, say—or reducing electricity demand temporarily by turning down customer thermostats, for example—to help maintain that balance.
“We were able to validate that the transactive control technique basically works,” Melton said.The signal provided the appropriate information for several scenarios. In others, such as anticipated hot or cold days, when energy use spikes for air conditioning or heating, the signal didn’t change. That’s also expected behavior because the BPA, which operates the regional transmission grid, anticipates those conditions and plans accordingly. “There was no real need to take short-term actions,” he said.
For the most part, demand response technologies focus on quickly reducing demand when power prices go high, indicating a shortage of energy. There are fewer opportunities to quickly increase demand to soak up excess, low-cost power on the system. “This is a challenge for the industry going forward to rethink some of the relationships and business models to be able to be flexible in either direction,” Melton said.
As part of the project, IBM did a simulation in which 30 percent of demand on the electricity system could respond to the incentive signals. In that scenario, the Pacific Northwest could expect a 4 percent reduction of peak power costs.
But reaching 30 percent penetration of demand-response technologies in actuality would require lots more investment in smart transformers, controllable thermostats, and other equipment. Another challenge will be articulating what that 4 percent savings means in dollar terms to entities like BPA and its utility customers, and agreeing on a financial arrangement to recognize the benefit and make the technologies a viable investment for the utilities and their customers.
“This is true anywhere you want to deploy these new technologies,” Melton said, referencing similar efforts in California and New York aimed at creating markets to integrate renewable energy and provide electricity customers more choice. “At the end of the day, in all of these discussions, we have to be able to clearly identify the value streams and what they’re worth to all parties so we can figure out the finance side of this. We have not done that yet.”
While the smart grid demonstration stopped collecting data last August, Melton said there are efforts afoot to keep some parts of the transactive control signals alive in the region for further research locally.