Photo credit: Duke Energy
It’s been just over a year since utility giant Duke Energy announced it was teaming up with Amazon Web Services to build cloud-based smart grid software. And while the first year of any major project is mostly planning and designing, Duke said it is already reaping the benefits of faster simulations.
Predicting exactly how much energy the Southeastern United States will need in future years requires running hundreds of millions of power flow calculations. That’s a process that used to take North Carolina-based Duke Energy around six weeks using traditional IT hardware.
But just over a year into a partnership with cloud computing giant AWS, the utility is running those same simulations in the cloud in a matter of hours. In the next few years, it expects that timeline to narrow to around 15 minutes.
Duke Energy brought AWS on board in late 2022 to move 12 intelligent grid apps to the cloud and help run them at scale — apps that process and analyze everything from electricity demand and energy efficiency to rooftop solar and electric vehicle adoption.
“We wanted to be able to do things in a very responsive and quick manner,” chief information officer Bonnie Titone told Latitude Media. “We have our own tools and we know how to forecast and plan and clean our data, but now we need to deliver them in a manner where they can respond almost instantaneously.”
Something like predicting future circuit usage takes “thousands and thousands of data elements,” she said. “We couldn’t do that all by backhauling it into our brick-and-mortar data center and then modeling it and spitting it out, because it would have taken too long.”
The process of building cloud infrastructure that meets the unique needs of utilities is a long one, Titone said, in part because the cloud is best suited for “generic services.” Cloud companies excel at building products that can be used across customers and industries, but the partnership with AWS entails creating a customized service that can manage and analyze data from a bidirectional grid.
“That’s the secret sauce, if you will,” Titone said. This utility-specific type of framework also enables customers to make decisions about selling power back to the grid as well as about where to build new load (like data centers and EV charging stations).
While the end goal is to bring all of Duke’s tools into the cloud, the utility’s power flow prediction tool is the closest to full internal deployment, Titone said. To date, Duke has run over 10 million simulations using the power flow service AWS created for them. Eventually, those simulations will help teams decide where to invest in substations and generation, among other things.
It’s still too early in the process — which Titone said will likely stretch beyond the five year period set out in the initial partnership — for Duke to have much more information about its impact, but the primary challenge has so far been cultural rather than technical.
“Think about Duke, we’re a 100-plus year old utility,” she said. “Think about AWS, a relatively new cloud services company. And guess what, we operate differently!”
There’s a maturity that has to develop across both companies to blend those two perspectives, Titone said.
Utilities aren’t known for their digital innovation, but Duke is bringing decades of experience in the energy sector; meanwhile, AWS brings the “ability to fail fast and try new things, see what sticks and works,” she added.
Year two of the partnership is focused on delivering beta services, and getting them into the hands of various Duke Energy stakeholders, Titone said. And by year three, Duke expects to be churning out tools at a steady clip, once they develop a pattern for production.
The utility doesn’t currently have plans to license its tools in any form, but in the long-term, the utility sector as a whole may benefit from the technology developed through the partnership. The impact of bringing Duke’s technology into the cloud, Titone explained, is developing a process and framework “that any utility can ultimately use.”
“They can’t use the tool that Duke built, but they can use that service to do the same thing for themselves, because everyone’s trying to solve this exact same problem.”