Episode 8: Analytics, Vegetation Strategy, and Reliability w/ National Grid’s Bert Stewart

Welcome to the Eighth Episode of Trees & Lines: Fresh Perspectives on Utility Vegetation Management!

National Grid’s Bert Stewart on Shifting to a Condition-Based Analytical VM Approach

Trees and Lines Podcast Transcript- Episode 8

A conversation with Bertram Stewart III (Manager, Reliability Analytics & Vegetation Strategy at National Grid) on technology and data analytics: The successful transition from cycle-based to condition-based vegetation management.


I’ve been focused on utility vegetation management pretty much my entire career. I started working for a tree contractor back in the mid-nineties to late nineties. Eight years ago, I had the opportunity to join National Grid. But one additional role that I did have really the privilege of having is reliability analytics. So not necessarily a vegetation management focus group, but it dovetails vegetation management in the utility world very, very well.

Of course, we know trees are typically our number one cause of outages. We invest millions and millions and millions of dollars in trying to mitigate that. So having that group tied in with my vegetation strategy team I thought it was a perfect fit and as a result I was able to learn far more about reliability metrics, those impacts, and how the vegetation piece can be an opportunity to improve those. It comes with just challenges, no doubt, but certainly it can be an opportunity to improve those metrics.

Shifting to a Condition-Based Analytical Approach to Managing Vegetation Assets

It’s a common question I get from colleagues across North America at different utility companies. I will say a time-based cycle approach is the best management practice in our business – there’s no question about that. At National Grid, we have been doing a form of that approach for over 40 years, consistently. In my prior role working for Central Vermont Public Service, Green Mountain Power, it was the same situation up there as well—time-based approach—for decades. What really made us pay attention and look at leveraging different tools to take a more condition-based approach leveraging technology to help us with that was that frankly, we had some common headwinds, not just with National Grid, but with a lot of utility companies out there.

As we all know about the vegetation management contractor service, they have really a significant difficulty in acquiring resources, talent, people to do this work. It’s getting harder and harder every day.

Early in my career, twenty years ago, I had the opportunity to work really close with our road tree workers from our contractors’ workforce. You’d see ten tree workers leave, then usually nine of those would come right back. A lot of them came back to the same exact contractor. Maybe they come back to a different contractor and change the color of their shirt or the color of their hat. The dynamic today has shifted so much today that if you have ten tree workers leave, you’d be lucky to see one of them come back to the industry.

Our contractors are investing more to keep the current talent they have. And they’re investing more to try to recruit more talent. So as a result, that comes at a premium and utility companies like Natural Grid are seeing those costs in the work that’s being, priced year-on-year. In a lot of cases, sometimes we’ll see double digit increases in a given year.

Certainly, it’s not a sustainable model. There are things that do suffer as a result. One of the things that suffered for us is our annual cyclical work plan, those miles that we prune. As a result, we found ourselves pulling work off the work plan, deferring work, kicking the can down the road per se. That certainly is not a sustainable strategy. We got to the point where we were deferring 10%-15% of our work plan. That was really one catalyst that led us to looking at more of a condition-based approach. On top of that, at the same time as all that was coming to a head, we had one of the worst reliability performance years at National Grid, specifically in our Massachusetts jurisdiction.

We have a duration target of about 150 minutes. It’s 147 minutes at the time. We more than doubled that in 2020. We had over 300 minutes of duration on average for the system, far exceeding our reliability targets, our DPU (data processing units) set by our department. As a result, our executives didn’t want to repeat that. So, we needed to figure out how can we right the ship. How can we figure out to do vegetation in a better manner leveraging data? We are a data rich company. All utilities are data rich. But how can you bring all that together and leverage that to really get better insights and make decisions that are actionable? There are a couple pieces there hopefully we can dive into on the data versus making it actionable, because that is a gap a lot of people struggle with. As far as the path that led National Grid to really make a fundamental change from a long history of time-based approach on our vegetation management to condition-based approach with insights, what led us there is largely those two causes.

Bringing the Team Aboard

When you look at like my team vegetation strategy, part of our role is the R&D arm of vegetation management. Based on our charge within the organization, we have to think outside the box, look at what’s out there, other opportunities that can really move the needle and the way we approach our strategies, the designs of our workplans. And ultimately, the execution of those work plans. Even though we are “the R&D” folks of vegetation strategy for National Grid, there’s certainly still some reluctance. Because it comes down to trust. Because we’ve been doing things a certain way for so long, we know they work.

Perhaps maybe we get a little bit comfortable with it. But they’ve worked. The key is recognizing that because we’ve done it that way for so long, does that really make it the right way to do it? Nonetheless, even within my organization and my team, people are suspect of technology and whether it’s right. Now we look at our operations group that works with our contractors in the field. It’s a big lift to get them on board to say, “This is the way we’re going to go. This is the change we’re going to make.” It takes time. And a lot of that time is exposing our staff to the technology and to the outputs, then validate those outputs.

We’ve been using our condition-based approach now for over two years with about a year and a half of actual execution in the field. We still validate. As a matter of fact, we’re going out next week in the field to do more validation on the data and the outputs that we get from our vendor. It does take time; It does take effort. But it’s needed. If you don’t trust the information, if you don’t trust the outputs, you’re not going to go very far with it. So that is key.

About Tolerance Limits and Scheduling

Like any part of your program, you make your own business rules, based on the constraints that you are working with. And as a result, you could take technology and the output at face value, and it could push it out eight, nine, ten years.

As a company we said, “No, we don’t want any of our circuits going longer than seven years”. So, we’re managing some of that uncertainty or long-term risk by limiting the length that some of this work can go. Not to say that some of these circuits can’t go six, seven years. Some can, but nonetheless, we just weren’t comfortable to extend that out however far the model will dictate. We did want to set rules on that. And there are certainly other business rules that we put in place as well within the model that we’re using to make sure that we’re capturing synergies and efficiencies.

Historically we would bundle several circuits together as a package and we put the bid to our contractors or award that to our contractors. Part of the optimization approach is you don’t necessarily bundle those together anymore. They could be individual circuits, but there’s still a risk in certain geographic areas that’s not really all that efficient. You think about some of the smaller circuits in the city of Worcester, Massachusetts, for example. These could be a tenth of a mile long. These circuits don’t have a lot of exposure. They’re short. The concern is we don’t want to forget about these. Because to do those coupled with other circuits in the area still does make sense. So, we end up designing business rules within the model. If we’re going to identify a circuit on this street, we want to incorporate these other smaller segments so we can get some sort of economies of scale as well when the contractor does go in there and do that. We don’t want them in there for a day or two, and then they have to travel an hour away to do their next job. That certainly doesn’t make sense as well. There are certainly some efficiencies that we recognized that we needed to put in place.

Surprises Within the Analytical Approach and Feeding the Model

We’ve certainly had several surprises. Some of them in a good way. But as far as an output that may not have been what we expected, a couple things that we’ve dealt with and we’ve worked with identifying: overhang.

Tree density is straightforward, especially if you’re using satellite. But then when you get the measurements around the wire that space in the vegetation encroachment within that space. The overhang was tough. And what was interesting was it wasn’t consistent. In certain geographic areas—that could have been due to terrain issues—the technology had a little bit more difficult time identifying overhang at a high level of accuracy. So that’s information that we feed back to the vendor. They feed back within the model, and we continue to do validation around that as well.

Those changes that are being done with the vendor result in better accuracy. It goes back to being accurate, making sure that our resources/stakeholders internally can trust the data. That’s key. Those are things that we found on the onset that were a little surprising that they couldn’t really decipher the overhang and that encroachment of overhang. Because they come from above. it’s really important to get that right.

Using Technology to Understand Vegetation Encroachment

When we talked about the path that led us to where we are today, part of that was deferring the work because we couldn’t afford to deliver the work plan as designed. One year, 2018 or ‘19, we actually leveraged our data science group to help us make a decision on what work should we include and what work can we defer or push down the road. And we did the process: We leveraged all our data, our GIS data, our outage data, and the history of our vegetation management inputs. The one thing we lacked in doing that process internally was the near real-time condition of vegetation adjacent to our overhead conductors.

We work with AI Dash on leveraging the optimized approach—ingesting all that information that I just mentioned. They capture the near real-time vegetation conditions around our assets. That really is the game changer. Whether you use satellite, LiDAR or Geiger-mode LiDAR or any other remote technology that’s available within our industry, that’s the piece that really turns the ship in the right direction: Understanding what that vegetation encroachment looks like. A lot of folks don’t understand what the density of vegetation is on their system.

Understanding Tree Density

We recently acquired Western Power Delivery in the U.K. and when you acquire another utility company, you start benchmarking. “Hey, what are you doing that’s really cool? Or not?” And you try to make comparisons and you ask the question, “What’s your tree density?”

“Oh, we don’t know.” They know how many spans have vegetation that requires work in the next four to five years, but they don’t have an idea of trees per a mile. They look at it entirely different way. And a lot of utilities in North America don’t necessarily understand their tree density and how that relates to impacts from our liability standpoint.

One thing that really has enlightened us is having that insight on the vegetation and how it’s changing as well out there.

Factors Considered in Determining Priorities

It really comes down to criticality. And when you’re looking at criticality, the tree conditions are certainly a major part of that. But that’s not the only part. It’s also about how has that circuit—or part of your system—performed in recent history outages. Then you look at the assets that are out there, the miles of overhead bare wire … it bundled cable, like Hendrix-type construction, or is it open three-phase bare wire, the critical customers that have service on those circuits. We feed that into the model as well. We understand where our critical customers are: Our hospitals, shelters that may be occupied and utilized during major events, police and fire, and first responder customers. All that plays into the criticality score of each individual circuit. And that criticality score can change year-on-year based on the way the wind blows and the way it changes the integrity and health of your vegetation.

The Cultural Shift from Time-Based to Condition-Based

The key is getting the right stakeholders to the table. I won’t say that bringing those stakeholders in at an early stage makes things easier – it doesn’t. Because they challenge every step of the way, and rightfully so. I challenge even in my position on some of the items that we’re working with and some of the outputs that we’re seeing. I think it’s appropriate. The key is bringing the right people to the table, making sure that they have a voice, making sure that they have a decision. They’re part of the decision. At the end of the day, if you can’t get everybody in agreement, it’s not going to pass muster and it’s not going to be utilized in the field.

Managing Up

That is probably where we have the most support. At National Grid we are all-in on digital solutions. This company wants to leverage digital solutions, not just for vegetation management, for many aspects of our business.

We have three or four significant digital evolution products going right now in addition to our vegetation management optimization product, and the company is all-in. We [recently] had an executive showcase in Boston where we had our CEO from the UK over and all those direct reports, and we had the opportunity to show them what VMO is, where we’ve been and where we’re at today. The takeaway was they want to see us leverage the technology and digital solutions to make us more nimble.  Not just for ourselves, but on behalf of our customers.

At the end of the day, that’s what it’s all about. The point is that National Grid is all-in on digital technology. We look at things as a “buy, build or hybrid” approach when it comes to technology. And they’re not afraid to do either. It’s all about making the best decision for the objective and the solution you’re trying to achieve.

The Difference Between Data and Actionable Data

We partner with AI Dash, but I like to talk with their competitors. I’ve had different levels of relationship with their competitors throughout the years from remote sensing offerings to software. And I like to see the evolution of how all that’s working across those different vendors. Because a lot of vendors out there have great analytics, they have beautiful dashboards that you can utilize, that you can apply to your system. I’ll dive into these different dashboards with different vendors, whether it’s GE, IBM or AI Dash, any of them.

I’ll look at their information. And after I review everything and ask my questions in probe, one of my final questions is: “Okay, I pushed the button to get a plan. Where’s the execute button?” How do I get an extraction out of that wonderful analytics software database to give to our vegetation operations group, or hand to a tree crew and say, “This is what you’ve got to do. This is what’s prioritized.” Some have that. A lot of them don’t. And that’s the piece that really holds back our industry on making that shift to leveraging the technology to help us optimize our approach. It’s about how you build trust. And then how you make it actionable. Where’s the button I push to get my work plan?

Thanks for listening to another episode of Trees and Lines, sponsored by Iapetus Infrastructure Services

We welcome you to listen to the episode of Trees and Lines on SpotifyAppleGoogleAmazonYouTube or your favorite podcast platform. If you like the show, please give us a five-star rating.

If you have any questions or if you have ideas for future episodes, please contact us at

If you have questions about efficient vegetation management solutions, please contact Iapetus Infrastructure Services. Our team of experts will support you through a wide range of vegetation and contractor management challenges. Don’t forget to leave a review if you enjoyed this week’s episode and follow our LinkedIn to stay updated on the latest news.

Leave a Comment

Your email address will not be published. Required fields are marked *