Applications of Artificial Intelligence (AI) in the Electric Utility Sector
image credit: Dreamstime.com
- Nov 29, 2019 5:00 pm GMTNov 29, 2019 5:00 pm GMT
- 1795 views
This item is part of the Special Issue - 2019-10 - Artificial Intelligence, click here for more
Michael Albrecht, Utility Industry Researcher-Writer, interviews Kevin Chandra (KC), Engineer, Electric Service Delivery, Design & Standards, at Austin Energy, on Applications of Artificial Intelligence (AI) in the Electric Utility Sector
MA: Welcome Kevin. We're all on a learning curve with the implementation of artificial intelligence, algorithms, big data, and their application in the utility space. You mentioned grid automation as a particular area of interest at Austin Energy, so, let’s dive in there.
KC: Absolutely. We have a couple of cool AI-related projects going on right now at Austin Energy involving grid modernization and automation surrounding what we call Volt/VAR optimization and conservation voltage reduction (CVR). Essentially, we put devices out not only at our substation for our load tap changing transformers, but also downstream on our distribution grid on capacitor banks or voltage regulators.
With these devices and software tools installed we can help optimize and stabilize voltage on our grid. The reason we're doing this is ANSI C84.1 has the five-percent requirement for providing voltage to customers. So, we're finding more ways of trying to not only stay within the lower portion of boundaries, this allows us to provide good voltage and find ways to save on energy for customers.
CVR allows us to take parts of our grid where voltage is on the higher end of the tolerance band and safely lower them down within the lower tolerance bands. The slightly lowered voltage allows circuits to pull less power allowing for conservation. That's done automatically, as these devices sense changes in the grid. This is an example of great AI-enabled automation that we are testing on our system.
MA: Can you speak to how AI and its algorithms contribute to this activity?
KC: The algorithms can be voltage-based, so that when they're tying in all of these operations together, it's checking those initial parameters we set to make sure everything checks good, and then it makes adjustments as needed. This is an initial pilot phase, so we’re approaching it conservatively. We're slowly trying to say, ‘OK, we can push this to this end limit in terms of actually receiving data.’ Not only do we receive it from these devices, like our capacitors or tap change controllers, but we also receive data from the meters. Moving into the age of smart meters, we are receiving more data than we used to 20 years ago or even 10 years ago, and are finding more ways to take that massive data bank and use it for balancing operations with grid voltage.
MA: You said that the algorithms are voltage-based. Does Austin Energy customize the algorithms to perform certain functions, or are these off-the-shelf algorithms in an AI software bundle?
KC : Here's a good example. At the substation level, some tap changes are operating a lot more than they previously used to, and that's obviously caused by these new pilot programs that we have for conservation logic reduction or VAR optimization. We can potentially make adjustments and say, ‘well, let's not overstress equipment.’ We can change those boundaries on when they will activate. That is us analyzing and then customizing based on initial results.
MA: Other utilities, especially small and medium sized co-ops and munis, are weighing whether to implement an AI system. They'll be weighing out the pros and cons, asking, how is this going to benefit us and our customers? Will it be worth the effort and expense?
KC: From a very simple standpoint, if we're able to reduce things like losses from excess voltage, that automatically is increasing efficiency in power delivery. This requires little customer engagement or incentives, yet produces energy savings on our circuits. That's better for the community as a whole.
Also, from a customer standpoint, if they're able to consume a little bit less power than they used to consume, that's a little bit less money they have to pay. If we can save money on the delivery side, those savings can potentially be passed on to the customer.
From an operator standpoint, we're reducing system losses. We're able to reduce our peak demand and reduce our energy consumption using conservation voltage reduction. For utilities, that can equate to a device we don't have to add, or another upgrade we can avoid to our infrastructure that might have been needed to support a new higher peak demand. That's an immediate payback from reduced system loss and increased grid efficiency.
AI implementation is one of those low hanging fruit categories where, yes, there's some cost involved in data processing and communication between these tools. But once we're able to connect these voltage regulating devices or reactive power devices together, we're able to automate and achieve these limits while still meet operating constraints when it comes to voltage or load. I see that as a good win for the utility and its customers.
MA: Excellent. That's a great note to end on. Thanks for your time, Kevin.