The Role of AI in Electric Grid Management with with Buzz Solutions’ Vikhyat Chaudhry

Hello, and welcome to the latest edition of the Techstrong AI Leadership Series. I'm your host, Mike Baard. Today we're with Vic Chandry, who's CTO for Buzz Solutions, and we're gonna be talking about, well, how AI might be applied to making our electric grids more efficient and power distribution and all that good stuff.

Vic, welcome to Shah. Thank you, Mike. We really do appreciate your time and happy to be here.

We talk about the fact that AI needs power all the time, but I don't think we spend a whole lot of time talking about how AI might help us get more power and be more efficient and distribute power where it needs to be. So, explain to us, if you would, what is the opportunity here of using AI to kind of maybe solve a problem that's been around for a long time? That's correct, yeah, Mike.

So in terms of ai, as you can see with the rise of, you know, these generative AI models and LLMs and all the fancy stuff, there's a rise for more compute. So the data center load is increasing, so AI requires more electricity and power, and it's even forecasted, uh, the data load or the, uh, the, the load forecast would be increase double and triple in the next five years because of these data centers. But at the same time, AI and machine learning, uh, does provide opportunities to even help power utilities, power delivery systems, and the grid itself to become much more, uh, optimized.

So in a way, if we can use AI to optimize the flow of power or the flow of electrons to the, to the conductors, to the wires, that really helps, uh, the utilities kind of redirect power in ways that can serve the society in general. One of the ways we are using ABU solutions, um, one of the core capabilities of ours is computer vision ai, so visual analytics, visual, uh, intelligence, which basically emphasizes, uh, modernizing and then monitoring the grid itself. So any kind of, uh, inspection that the utilities are doing on the grid itself.

So the power lines, their substations, there are different kind of infrastructure assets. We are using all that visual data, so imagery and videos collected from drones or helicopters or fixed cameras. We are taking that and then analyzing it in a much more efficient, much more faster and cost effective manner.

For utilities, we are finding any kind of defective anomalies, you know, overheating in the lines, uh, transformers getting damaged, electrical equipment getting damaged or overheated, and even vegetation that's encroaching on the power lines that can cause a lot of problems, such as, you know, we've seen in cases like wildfires have sparked due to failed grid infrastructure. So we are providing all these insights back to utilities with our ai, and then they can use that effectively, make them actions and do a much more proactive maintenance so that the grid remains, uh, less stress strained due to climatic factors or external factors, but also internal factors do rising due to demands, uh, but also it makes the utilities much more proactive and then, uh, much more smarter on using such operations for their, for the power line and grid network systems. It also seems to me that the grid itself is highly distributed.

It's a physical piece of equipment everywhere. The utilities folks have enough people to manage all this, or is AI gonna help them level that playing field a little bit? No, that's a great question.

Um, the utilities are facing a lot of challenges in terms of resources itself and, uh, the, and the grid is expansion, it's expand, expanding. The grid is very distributed. There's more distributed energy resources coming online on the grid, for example, like renewable resources like solar, wind.

So that's adding a lot more power and strain on the grid itself. But managing the, and maintaining the grid is a challenge for utilities itself. So a lot of the field engineers, the inspectors, the linemen, uh, there is a workforce challenge and resource challenge that the utilities are facing.

A lot of this workforce is actually retiring, so there's a big gap in the workforce that is coming into these kind of roles. And that poses, uh, a problem where AI can be a solution where, you know, specifically for our example, we can help in resource optimization. So mundane tasks like looking at images for hours and for months, uh, finding out any kind of defects or anomalies on the grid, uh, is just such a mundane task.

Um, and it takes a lot of time. So instead of these specialized people and resources doing that, uh, why don't we give that to AI to kind of find trends and patterns and anomalies and defects, uh, in that data, and then all these specialized resources get that, uh, insights and those information that they can turn actions. So now the resources are more optimized.

They're the, they're, uh, they're time effective and they can be used in a much more efficient and optimized manner to do maintenance out in the field. Mm-hmm. It also seems to me that a lot of the equipment is aging and replacing all of it overnight would be cost prohibitive.

So how will AI kind of help us figure out maybe where the most chronic issues are? And maybe that's where we replace the gear than others and, uh, 'cause not all aspects of the grid are equally, uh, shall we say warm. Yeah, uh, that's a great point.

Uh, the, the aging, we have seen examples where there's a lot of components that are decades old and they might have been, um, out of their shelf life as well that are still deployed on the system and the network. So there is a big need to map all of the network from the physical space to a digital world, which the utilities are heavily investing in, is to building those digital, um, uh, twins, uh, digital transformation efforts and providing, uh, putting a lot of resources and efforts into that, but also figuring it out where these kind of assets, the, these equipment are located. And what is the condition of that?

Uh, first of all, to figure out when they would be, they would fail and cause problems on the network, but also due to supply chain issues that are happening. There's, you know, we have seen transformers that are backlogged by, uh, you know, three to five years, uh, that are needed. So there's supply chain issues adding more problems, uh, for the utilities as well.

And that's where the, these kind of AI solutions, which are more on the visual side, uh, which we are delivering, helps utilities to not only just inventory their assets and equipment that is on the grid, so we can tell the utilities how many transformers are located in a certain distribution feed line or transmission cord or how many insulators, but on top of that, we tell them what is the condition of that? Are they, uh, are they degrading? Do they have any kind of, uh, anomalies or defects on that?

Do they have cracks, um, you know, broken insulators, broken conductors? So now they're getting information about what is the condition of that, and that impacts the shelf life of the equipment itself, but it also helps utilities to send out crews or maintenance people in the field and take actions accordingly on their, um, on the grid, uh, equipment itself. So now they know what kind of asset, what kind of equipment is located at the assets, where it's located, and what is the condition of that.

Now they can take a, a pre preemptive and proactive measure on how to repair it or replace it ahead of time. Mm-hmm. And to that point, how predictive can we get with AI and can we see things coming sooner?

Because I think a lot of the times historically, at least we act like we're surprised, but I got a feeling that's a, some folks out there and knew something earlier, but they just didn't have a way to kinda share that. But maybe if we have ai, everybody can see what the issues are gonna be sooner. Yeah, that's correct.

So in terms of AI predictability, there's, again, I would, I always say that the world of AI is changing every six months now there's new techniques, new innovative solutions coming out, and we try to leverage all of them. So right now, what we are doing is we are detecting for utilities. We tell them, uh, where their problems are happening on the grid and the network, what kind of problems they are and what, and what can they do about it.

What we are trying to move towards, uh, where the utilities can leverage a lot of, um, uh, you know, benefits is, is the predictive capabilities. So since we are collecting a lot of visual data historically as well, so it it not only geographically, but over time, we are collecting all this data. We are looking into feeding into a predictive analytics or predictive asset management system.

So what it does is now we, we, using time series data, we are adding, uh, climatic, uh, you know, uh, features on that, you know, what is the humidity, temperature, pressure, wind patterns of the certain location. We are also looking at, um, you know, load patterns, uh, within the conductor or the line itself. So what's happening within the line, uh, what's happening outside the line and what's happening on the equipment, and trying to feed that to a system that can check out these trends and basically find out areas that will require much more prioritization or much more, uh, you know, emphasis on maintenance.

So for example, let's say we take an area where there's, uh, inspection being done over and over again, and utilities want to find out what is the likelihood of, you know, let's say a five transformers going bad. So this system would be able to forecast and predict into the future, uh, based on the data that's collected, that let's say in this specific area, there's like 90% likelihood that five, five transformers would go bad because, you know, let's say there were 50 electric vehicles added in that neighborhood, uh, you know, because of that. So now the utilities can basically, uh, prioritize that area for maintenance and do much more inspection maintenance in that area instead of looking at, you know, thousands and thousands of miles of, uh, of distribution lines.

I don't know if you're doing this or not, but are you also maybe playing around with digital twin technologies so I can model the grid and do a lot of what if simulations? We are, I think that's the plan where we want to move towards. Right now, as I said, we are, we are on the detection side and we are giving actionable insights, but we really want to take in different kind of data sets into our platform and different data sources so that we can build these digital twins, uh, for the utilities.

And one great thing about taking the physical aspect of the grid to a digital world is, is basically creating these simulations and, and checking out various scenarios. One example is, let's say I want to add 50 electric vehicles in a neighborhood. How would that impact the transformer for that neighborhood?

Would it, you know, overload it, would it cause sparking those kind of things? And how would that impact the equipments itself that are deployed on the network? So we are looking to put more emphasis on that as we are going forward.

Right now, what we are delivering to utilities is, uh, is a system that basically tells them the health of their system currently, but the value is what can we do to tell them what would happen in the future for their system? A lot of local, state, federal governments are usually involved in anything to do with utilities. Um, is there something that they could be doing to help maybe drive adoption of these technologies to make the grid more robust?

Because frankly we're all counting on it. Yeah. Um, I think one of the things I I say is a lot of times regulation has to catch up with technology.

Technology is always leading the charge over there. And I think the more education we can provide, uh, to the regulators, to the energy commissions, um, uh, the public utility commissions, uh, about the presence of this technology and how successfully it has been deployed with utilities, either through pilots or full scale deployments that we are doing with utilities like Dominion Energy, New York Power Authority, which are some of our biggest customers. The more education we can provide to the, to the regulatory commissions, the, the better it is for utilities to kind of incorporate newer technologies like, uh, like AI or IOT sensors or even drone inspections at a larger scale.

'cause now then they have the backing from the regulatory commissions, but also, uh, general public as well, that now they can test out these technologies even more. I like to say that utilities are facing lot of complicated issues on the network. The grid is not the same as it was like 50 years ago.

Uh, it's a whole different world of the grid. The electricity and power is flowing, bidirectionally, uh, now you have distributed energy resources on the grid. You have electric vehicles, everything's getting electrified.

The low demands are increasing exponentially. So utilities have to find innovative solutions to tackle this. One of them is, is how we can leverage machine learning data sense and ai and the utilities really need backing from, from the regulators for that.

And you didn't mention the data centers we're building for ai, but that's one other source of, uh, consumption that's rather large. Um, is it your sense that the grids are gonna be able to handle that if we're smart about it, or is that gonna kind of tip us over the top and we just need a new structure altogether? Yeah, I think if we are smart about that and we, we make the grid more smarter and we make the grid modernized, um, then we would be able to tackle those, those problems and tackle those challenges for utilities.

As you know, the grid has, is, is a century old, uh, the grid was the biggest, you know, invention of the 20th century. I like to say AI is, is the biggest invention of, of the 21st century. And you can actually actually call AI the electricity of, uh, of the 21st century.

Um, the, the big thing for, for the utilities would be to figure out how can they use these innovative solutions that are tested out in the field and how can they deploy successfully. The the grid I would say is, is think of it like a highway. It's getting, uh, congested because there's distributed energy resources like solar utility scale, solar farms, wind, wind farms, hydro hydropower, uh, electric vehicles, data centers that are driving a lot of power demand on the grid, both, you know, downstream and upstream.

Uh, in order to not have this highway or the grid get congested and stress and strain, there is an emphasis on continuous monitoring, frequent monitoring and inspection, and making sure that the grid is much smarter when it's dealing with these kind of bidirectional flow of power. All folks, you heard it here, electricity. We take it for granted, but looks like we need some advanced technologies to make sure it's gonna be around when we need it.

Hey Vic, thanks for being on the show. Thank you, Mike. Really appreciate it.

All right. And thank you for all watching the latest episode of the Techstrong AI video series. You can find this episode and others on our website.

We can invite to check them all out. Until then, we'll see you next time.