How AI and Machine Learning Are Helping Utilities Digitize

Utilities are drilling into machine learning and advanced analytics to help improve forecasting for energy consumption, damaging storms and equipment maintenance, especially as the volume of data generated by sensors on meters and pipelines continues to grow, industry analysts said.

This “goldmine of data” includes distribution management data, geospatial data, emergency management data, workforce management data and asset data, said Matt Schnugg, senior director of data and analytics, ML/AI at GE Power Digital. Currently, most of that data stays with a utility’s historian or is archived. 

To get a sense of scale, one large utility that partners with GE shares roughly 14 petabytes of data from its grid operations, in some cases funneling GE 200-300 gigabytes a day. 

“The opportunity for us is to say there’s a lot of value inherent in that data,” Schnugg said, especially when it’s combined with third-party data from weather satellites, different market segments and geographic regions, for example. Utilities “can be more prescient in their ability to operate.” 

Analytics in Demand

With the increasing uncertainty and higher likelihood of severe weather events, GE released a new Storm Readiness analytic this month that creates models based on weather forecasts, outage history, crew response and geographic data. 

“We know anecdotally, that even if grid operators are right about [predicting] the location but wrong by a day, it can be up to a million dollars in operating cost from overtime and ensuring the area is blanketed by crews and equipment,” Schnugg said. “If we can influence that event one time over the course of the year, it is more than enough justification of investment in the license.” 

GE’s other new tools focus on improving the integrity of network data and stabilizing power generation. 

Brian Hurst, vice president and chief analytics officer for Exelon Utilities, an early adopter of GE’s grid analytics says, “we are just beginning to scratch the surface on the value of analytics.” 

“When it comes to storm restoration,” Hurst said, GE’s tool “will enable the utilities to become more surgical in prepositioning crews in advance of weather events — saving time, money, improving customer satisfaction and enhancing safety for employees.” Exelon serves 10 million customers through its six utility companies in the United States.

Additional solutions include an analytics tool from Siemens that helps identify locations to make creative, cost-effective investments (known as Non-Wires Alternatives) in aging or new infrastructure, often with an eye toward environmental benefits.

Ericsson offers Smart Metering as a service. The communications giant based in Sweden,

estimated that the global number of devices managed by utilities — meters, grid sensors, actuators and electrical appliances, for example — is projected to reach 1.53 billion in 2020. 

Plutoshift, a Palo Alto startup, focuses on IIoT analytics for hydropower and industrial waterworks to prevent leaks, improve flow rates and monitor water quality.