Powering the future: Entergy's pioneering journey in grid innovation and resilience

SPEAKER_2 (0:05)

Welcome to the AiDASH Evolve 2024 podcast. AiDASH makes critical infrastructure industries climate resilient and sustainable with satellites and AI. Evolve 2024 in New Orleans, Louisiana was our inaugural event. We gathered industry leaders, innovators, and visionaries to explore how cutting edge technology and satellites and AI is safeguarding critical infrastructure and fostering biodiversity to ensure a sustainable future amidst climate change challenges. In this podcast series, we are recapping the entire event every speaker and session. You can find it all on our content hub at ai-dot-com-slash-evolve2024 or check the show notes for all pertinent links. This is the final episode of our podcast series. This one features the closing keynote for AiDASH Evolve 2024, delivered by Jason Chapman, SVP and Chief Technology and Business Services Officer at Entergy. Jason shared insights from Entergy's groundbreaking journey in adopting cutting-edge technologies. He talked about how Entergy utilizes AI to predict and prevent equipment failures, minimize system downtime, and keep customers informed. He explained how innovative AI-driven processes, including satellite-powered vegetation management and computer vision-powered asset inspections have supported Entergy by optimizing operations and reducing risks. With its suite of tools, Entergy is looking to revolutionize storm damage assessments by allowing crews to respond in hours instead of days and giving customers more accurate estimated restoration timelines. Here's Jason.

Jason Chapman (1:51)

So, moving on a little bit about Entergy. So Entergy is proudly headquartered here in New Orleans. And our 12,000 employees power life for more than 3 million customers across four states. Together, we generate roughly $12 billion of revenue on nearly $60 billion of assets. And we've been in business proudly for more than a century. So, our portfolio is quite large.

We serve five operating companies across those four states and have more than 106,000 circuit miles of distribution lines. Another 16,000 miles of high-voltage transmission lines and close to 24,000 megawatts of capacity. And approximately 70% of that capacity is either modern gas, nuclear, or renewables.

So moving on to really the spirit of discussion, which is really our AI journey at Entergy. Maybe just to set context before, what I'd really like to do is talk to you about some of the use cases that we're seeing. And before doing that, I want to talk a little bit about how we think about AI and really how we approach it from a discipline perspective. So ultimately, everything we do with AI is focused on improving outcomes for our key stakeholders. In other words, we're particularly going after and trying to solve problems that focus around four key areas, either improving our cost position, improving our customer experience, improving operational performance, or the organizational health. We do that by integrating technology solutions and most importantly, actually, we reimagine the business process. And I'm going to talk a little bit more about that as we go into this. But one of the key findings that we've had as we've walked through this journey is you never drop AI into an existing process. You've got to redesign the process from the ground up to really get maximized and get the most use out of the technology.

Just as context for us, we think of AI as an umbrella or a full portfolio of different technologies. And I won't go through every one of these, but that can include things such as automation, natural language processing, computer vision, robotics, any number of technologies that we bundle together as part of a rebuilt end-to-end process ultimately aimed at driving outcomes in those four spaces that I mentioned. So I thought it might be helpful to kind of walk you through an example of how we've used AI in the past to rebuild a process and ultimately to drive an outcome. So like a lot of utilities, one of the things that we've struggled with in the past is asset failures that cause unplanned downtime for our customers. The worst thing that can happen is when a customer's power goes out and it's not planned. Now, the process at the top is what historically has happened at the company. In this case, I'm going to use an example of where a transformer would fail. Once the transformer fails, the customer loses power. They're now in the dark. They're getting agitated. They call our contact center. Our contact center opens up a work order, we roll a truck. Meanwhile, the customer is there and waiting. They don't really have an estimate of when the power is going to come back on.

The crew does show up, but again, they're slower to get there because this is a reactive situation. We've actually got to identify the crew and spin them out. They get there, they fix the transformer. Hours have passed by this time, and boom, without notification, the power goes back on. There's two bad outcomes that happened to that circumstance, and really two things that we're trying to solve for. The first and most importantly is we had a power outage. That impacted our customers. That was unplanned. Second thing, we had poor communication in that element. If you're going to have a power outage, you've got to be a really strong communicator, and constantly update your customer in terms of when they can expect power to be delivered back on. That allows them to plan for it accordingly and go about their day. So knowing that is the context of the problem. Our AI team started playing around with technology that would look at and monitor really the current transfer that was happening all the way down to our AMI residential meters. What they found in this analysis was that we could measure spikes in the current delivery that would ultimately, that were highly correlated to an upstream asset failure, in this case specifically a transformer failure. So in other words, when they measured a spike above or below a certain threshold, that was an indication that upstream of this, we had a transformer that was not working properly. Well that by itself was a really good milestone. And as they tested that, they were able to validate that, they were able to produce that outcome with a high degree of accuracy. But that by itself doesn't really allow us to action or to correct the first two problems that we had. So what we did was we built an entire end-to-end process around maximizing the value of that tool. So as that tool identifies a potential transformer that's going to go into potential failure, it runs an automation that does two things. The first thing that it does is it generates a work order. Second thing that it does, it generates a notification to the customer automatically and it says, hey, we've got maintenance that's scheduled maintenance, that's going to go on in your neighborhood during a specified period of time. You might be seeing a power outage in this window and power will be back up by that time. So we're maintaining communication with the customer. The work order that was generated meanwhile is rolling a truck. The truck ultimately gets out there, tests the transformer and either fixes it or addresses it right in the field. Once that's done, they update the system, the system generates a notification, and then the customer gets notified that their maintenance is done, they can move on. So there were two things that happened there. The first was that we avoided the power outage or in worst case, we minimized the time duration of that outage. But the second thing that we were able to do is we were able to stay in constant contact with the customer. So as I mentioned this, and this is a fairly basic example, but I wanted to illustrate the point of the power of the technology itself is extremely powerful, but to really gain full value, you've got to design an end-to-end process to take advantage of that. So real briefly, we've been working with AI really for several years now, and we've been able to generate a lot of use cases really across our entire business. This includes use cases in generation, power delivery, our customer channels, as well as our corporate environments. And I'll spend a few minutes going through some of the use cases in each of these. So I'm going to start in vegetation management. So very appropriate with this conference, vegetation management is something that almost every utility struggles with. You're struggling with that balance between affordability, in other words, how much could you actually budget with the risk that you're trying to manage and the operational outcome that you're trying to drive. That is, no matter how much money you have, it never feels like it's enough. In its context, our budget in this space is massive. I mean, you're talking about eight to nine figures on any given year, where we're focused on vegetation management. So seeing this as a massive opportunity for us to really do more, to do more, reduce the risk, drive the operational outcomes that we're looking for, we saw technology as a possible play in this space. So in 2020, actually starting in 2019, but in 2020, we went live and partnered with AiDASH to implement IVMS. And that was a very substantial move for us. It was a big bold bet on satellite-powered AI technology that ultimately allowed us to be much more intelligent with how we're managing veg. That had an immediate impact. So very next year, in a trend that's continued since that time, we've had our lowest vegetation CIs in our history. And that, as I mentioned, is a trend that's continued. Now, while we implement it, and this goes back to the process point that I was mentioning before, it is always our practice to redo the business process around the technology. And in this case, we took the time to digitize and automate planning and work execution processes. Now, that ended up in itself paying big dividends for us. In 2023, this entire service territory was really hit by a very severe drought, and it created exceptional fire risks that we had to deal with across our service territory. Because we had digitized the work processes, we were actually pretty well organized. In other words, we had our entire teams, roughly 300 to 350 people, all using the same platform to plan, prioritize, and execute the work. And what we were able to do from that is, we were very quickly able to go out and identify and mitigate more than 20,000 high-risk trees before they became an issue.