Data analytics in action on the grid
"The intelligent use of data is really what is going to make the change and create the value out of [smart grid] projects. We've all built things and been able to demonstrate that they look good, but the key is getting sustainable benefits.
"That is the transformation element of the projects themselves."
--Glenn Pritchard, technology lead for PECO's SGSM project.
While the mainstream media continues to focus on concerns about data privacy, leading utilities are focusing on using the new data smart meters are providing to increase reliability, lower operational costs, and much more.
In last week's Utility Analytics Institute (UAI) webcast, Grid Analytics, Issues, Trends & Drivers, both PECO and CenterPoint Energy shared the ways in which they are already using the new data available to them. CenterPoint has more than two million AMI meters deployed, while PECO is using AMR data at the moment, and begins deploying AMI meters this week.
"We interrogate our meters three times a day, and we bring back 15-minute intervals from each one of those meters, which is technically 96 intervals a day. So our database is over 20 terabytes," Mary Rich, CenterPoint Energy's smart grid systems manager, told attendees.
"Now, the reason that's important is because if you're starting your project, you don't understand the magnitude of the data that coming back into your system for the analytics. Because there's a lot of data that these meters can return, no matter what meter you're installing," she said.
The big question, she said, is what to do with all of this data. "How do you use this data, and how do you present this data to someone that could actually use this data? Those are really, really tough questions that most utilities are facing right now if they're deploying smart meters," Rich said.
She said her utility quickly found that there was so much more they could do with the data than they had initially anticipated. A full-blown data analytics platform tool now gives access to the data across the company for a variety of different uses.
"Diversion was one of the big things that kept coming up," she said. "We wanted to point out diversion; we wanted to be able to identify diversion very quickly." A lot of analytics has to go on behind the scenes, replacing the meter reader eyeballs in the field, and out-of-kilter monthly reads. "If you look at a disconnect and a tilt on a meter, which are both alerts that you get back into your database, you have to look and see if you have a service order there, or if you have an outage in there or something like that," Rich said. "If you can't pinpoint those things, what happened with that meter? There are a lot of variables in that."
Once the meter connection has been restored, and data begins coming back in, the 15-minute data can be analyzed to detect whether the meter has been tampered with or not.
Other areas in which analytics are assisting the company are outage identification, meter status and health checks, planning optimization, unbilled revenue, "left in hot" meters, data cleansing, reporting asset management and transformer load management.
Transformer load management was an issue Pritchard delved into, as well.
"For distribution operations, PECO has been playing with our data for some time now, and I say 'playing' lovingly. We've actually seen some real, tangible benefits from this," Pritchard said. "Analytics have helped to identify overloaded transformers. The same analogy can be used for cables and other devices throughout the grid.
"By using the usage at the end points and marrying that up with the usage at the origination of the circuit, at the substation, you can create virtual models throughout the power grid with this kind of data."
Looking at the end points is particularly useful. "Even looking down at the customer level, what customers are contributing to some of the peak load conditions? Are they candidates for load control or demand response programs? Or maybe a candidate to realign their load to a different circuit, different facilities, to better operate your system overall," Pritchard said.
Pritchard also shared some lessons learned, including:
- Garbage in equals garbage out. Make sure you have a reasonably accurate customer-transformer model. Analytics can help clean that up.
- Don't be afraid to experiment, develop hypotheses and test them.
- Start small in manageable sets of data, and scale up with your successes.
- The business needs to actively engage in defining how this data can be used and processed.
- Several utilities already have AMR/AMI systems, so there is experience to learn from.
No discussions yet. Start a discussion below.