The U.S. has experienced one of the most impactful hurricane seasons in recent memory. And, to make matters worse, it’s not over yet.
Millions of customers in the Southeast have had their first brush with climate change, with many losing power for days on end. Debates over the reliability of the century-old power grid are no longer reserved for sleepy utility boardrooms. The abstractness of a challenge for future generations has given way to an urgent, and present, imperative.
Hurricane Beryl knocked out power to 2 million customers in CenterPoint Energy’s territory in Houston. Nearly as many were impacted on Duke Energy’s system in the Carolinas by what began as Hurricane Helene, including this reporter’s community of Asheville. Hurricane Milton battered the Gulf Coast of Florida, and Duke Energy, with a second major storm in as many weeks.
In the wake of seemingly every major storm, POWERGRID International traffic spikes. Specifically, years-old content related to the process and costs of undergrounding powerlines. But in each of these storms — and the vast majority of the ones that came before them — trees, not flimsy power poles, are the primary driver of outages.
Utilities across the U.S. are recognizing the importance of strategic vegetation management, and not just the ones sitting in Hurricane Alley. Data analytics and artificial intelligence, meanwhile, are bringing advanced strategies to a historically localized utility chore.
But don’t be fooled: vegetation management, while less sexy than the advanced grid tools praised as key enablers of the energy transition, is no throwaway task. In overall utility expenses, vegetation management is typically the largest line item, with electric utilities spending between $6-$8 billion per year on maintenance for their overhead powerlines, according to the research firm Accenture.
Avangrid, the parent company of a handful of electric and gas utilities in the Northeast, faced an ongoing reliability challenge. Much of its service territories are rural and blanketed with trees. One of its companies, New York State Electric and Gas, maintains more than 38,000 miles of overhead lines in Upstate New York in some of the densest forested areas in the Northeast.
Avangrid recognized that data would be at the heart of the solution. The company scaled up an in-house data science and analytics team that now supports reliability and resilience efforts for its electric operations in Connecticut, New York, and Maine. The group develops advanced analytics, models, and AI technology that allows operations personnel to make informed decisions and targeted investments, including vegetation management.
Led by Mark Waclawiak, Avangrid’s senior manager of Operational Performance, the group aimed to quantify the resilience and reliability benefits of individual tree-trimming activities. Using data analytics, they target the trees that produce the greatest benefits at the lowest cost on specific circuit segments based on a unified risk profile. In conjunction with a recent regulatory approval that allowed NYSEG to transition to a six-year vegetation management cycle, Avangrid has realized “double-digit benefits year-over-year” in NYSEG’s territory, Waclawiak said.
“That’s something that we’re really excited about because we can go to our regulators and customers and say the money we’re investing in vegetation programs is making a real, measurable, quantifiable difference,” Waclawiak told POWERGRID. “We don’t want to be in a position where we’re using analytics or we’re investing and we say, ‘Trust us, it’s beneficial.'”
For decades, utilities have accepted cycle-based vegetation management programs as the gold standard for prudent trimming and maintenance. A shift in recent years has led some utilities, like Avangrid’s Northeast neighbor National Grid, to adopt a condition-based model.
Under cycle-based programs, utilities routinely trim trees and floors along power line corridors every handful of years— typically a five or six-year cadence. Crews visit circuits when it’s their turn, not necessarily when they need maintenance due to reliability risks or vegetation growth.
In 2020, National Grid suffered significant reliability declines for its operations in Massachusetts. A dozen or so storms doubled customer outage minutes within the territory, and the individual storms weren’t widespread enough to qualify for exclusions from National Grid’s overall reliability metrics.
Bert Stewart, the manager of National Grid’s vegetation management strategy, was charged with finding a solution. On top of being “nickel and dimed” by minor storms, National Grid was facing double-digit labor cost increases from its vegetation management contractors. The utility found itself “kicking the can down the road” on some vegetation management projects, which further threatened reliability.
Stewart searched for answers in the market. He had conversations with IBM, Leidos, GE, and Hitachi before being drawn to a newcomer in the space — a startup, AiDash — that provided actionable insights in addition to the advanced data analytics delivered by its peers.
In just six months, National Grid had full access to AiDash’s platform as part of a pilot program. The goal of the pilot was to develop an actionable vegetation management plan for National Grid’s distribution system in Massachusetts for the upcoming year while meeting the reliability baseline and reducing workload.
National Grid measures vegetation management effectiveness by comparing the three-year performance average of a circuit to the year after work is completed. Traditional cycle-based vegetation management would produce an 8 to 11 percent reduction in customer interruptions the following year, but outage minutes remained relatively stable.
By switching to condition-based vegetation management, and incorporating AiDash’s data analysis, National Grid is realizing a 30 percent reduction in customers interrupted and a 55 percent reduction in outage minutes. In the first few years of the program, the utility also saved around $2 million a year by reducing the overall workload.
It’s a small sample size, and Stewart expects some diminishing returns as the program matures and reliability gains level off. But he expects condition-based vegetation management to continue to outperform cycle-based programs long into the future.
“Extraordinary numbers,” Stewart told POWERGRID. “I’m over the moon about them. I love them. I think that’s a huge step change in our program and results.”
Abhishek Singh, AiDash’s co-founder and CEO, hadn’t heard the phrase “vegetation management” before 2019. Born in India, Singh was living in Central California, where infrastructure owned by utility Pacific Gas & Electric ignited the deadliest wildfire in California’s history.
Having built enterprise software companies in the past, Singh and his co-founders were looking to launch a new AI venture but needed a direction. Through their research, they realized that many wildfires are caused by vegetation interacting with the power grid. And, more importantly, for any startup, utilities were spending billions on trimming trees with no software tool dedicated to vegetation management on the market.
AiDash signed Entergy as its first customer days after incorporating in 2019. The company made its formal debut at DISTRIBUTECH, North America’s largest utility event, in 2020. National Grid would become its second customer, and later an investor, through its investing arm, National Grid Partners. Duke Energy and Edison International supported later rounds. AiDash now boasts more than 135 utility customers globally, including around 100 in the U.S.
Singh recognized early on that utilities were looking for actionable insights more than enhanced data. AiDash uses AI to analyze satellite imagery along with other data about a utility’s power grid with very few inputs from the utility.
“We decided that we would build vegetation management software. It’s a use case with multi-billion annual spending (by utilities), it should have software of its own,” Singh told POWERGRID.
Some utilities are building advanced vegetation management tools themselves.
FirstEnergy, which serves 6 million customers in Ohio, Pennsylvania, New Jersey, West Virginia, Maryland, and New York, recently launched an AI-driven vegetation management platform to analyze data on soil, weather, roadway, historical outages, and aerial photos. The Advanced Vegetation Analytics Tool’s prediction model can calculate the likelihood of trees damaging a circuit, which supplements the utility’s four-year pruning cycle.
“By using this data tool to study the type of vegetation conditions in each area and the proximity to our power lines, we can better recommend the right type of equipment crews and contractors will need to bring with them to trim the first time they go out,” said Tyler Woody, FirstEnergy’s general manager of distribution vegetation management operations.
The ongoing challenge for Avangrid, National Grid, First Energy, and utilities across the country will always be that you can’t cut every tree. It’s simply not feasible, especially with rising contractor and labor costs. But software allows them to cut “the right miles at the right time,” National Grid’s Stewart said.
“The return on investment is in the form of reliability improvement. That’s our business case today. And that’s how it evolved—from doing fewer miles to doing the right miles and achieving significant reliability gains,” Stewart said.
