Winter weather poses unique challenges to power grids, especially regarding how vegetation might endanger power lines and other utility field assets. When planning how to prevent winter power outages, and how to mobilize to restore power after a winter storm, utilities should consider several season-specific factors such as the weight of snow and ice, and how wind moves tree limbs differently after the leaves have fallen.

Utilities must factor winter weather into their year-round vegetation management plans. "The approach is the same, but the risks are different," explained Allen Carey, Manager of Vegetation Management for Eversource Distribution in Connecticut. "In summer, trees still have their full leaf canopy, which makes them especially vulnerable to high wind conditions," Carey said. "The leaves act as sails. But in winter, when the branches are bare, the threats are more about the weight of ice and snow. Dead or unhealthy trees are less structurally sound. Limbs may fall more readily and trees may be uprooted. Roads may remain ice-covered for a long time, slowing outage response."

Increasingly, climate change is complicating winter weather outage prediction and prevention by blending warm- and cold-weather risks. For instance, during the first week of September 2020, temperatures around Denver, Colorado dropped as much as 70˚F in 24 hours as a surprisingly fast and early winter storm moved in, dropping six or more inches of snow across the metro area. It was still technically summer, so the leaves were green and thick on local trees. The wet, heavy snow brought down many large branches and many power lines in the process. Meteorologist Cory Reppenhagen of 9News Denver reported how climate change is a likely culprit behind such unseasonable winter storms, by causing sharp, sudden shifts in the jet stream. Also, by expanding the Earth's tropical zone, powerful and wet tropical cyclones are more likely to push farther into Northern latitudes.

Weather Data, Then and Now

Historically, utilities have relied mainly on past experience to plan how to manage vegetation to minimize weather-related outage risks, and to mobilize outage restoration resources. As weather-related outage risks keep shifting and growing for every season, more utilities are finding it helpful to combine highly accurate and granular weather data (forecasts and historical) with insight from artificial intelligence (AI) algorithms to fine-tune their planning and budgeting to keep the power on, even during the worst winter storms.

Furthermore, utilities traditionally have obtained weather data mostly from sources that are not optimal for predicting power system impacts. "Most federal government weather monitoring stations have been located near airports — but nobody lives at the airport, and most of the power that utilities supply is not consumed there," said Bryan Sacks, head of work and asset optimization solutions for IBM. "Utilities had to make interpolations to populated areas, which limited the accuracy of predictions."

In recent years, sources of weather data have become far more diverse: from highly granular and accurate satellite imagery to consumer weather apps and social media reports. Combining and analyzing all of these resources is beyond the capability of most utilities. But now, IBM is uniquely positioned to support utility vegetation management and outage prediction.

In 2016, IBM acquired The Weather Company, which included the world's largest repository of weather station data, satellite weather data, several popular weather apps and the data resources of The Weather Channel. This data is now analyzed in real-time by IBM's AI algorithms, leveraging vast cloud computing resources to support more effective modeling of both vegetation and weather. Automatically generated real-time and long-term insights can support utility planners with hyperlocal assessments of vegetation risks. "We can model what's happening, or what will likely happen, with vegetation and weather down to a 500-square meter grid segment," said Sacks.

Better Data, Better Results

One utility that is benefiting from rich weather and vegetation data and analytics is Énergie NB Power, serving New Brunswick, Canada. There, winter temperatures can plunge below -22˚F, causing extreme grid problems. For instance, in January 2017, a severe ice storm knocked out power for 133,000 of that utility's 400,000 customers. The damage required NB Power to install 600 new utility poles, 150 new transformers, and 52 kilometers of new distribution lines, at a cost of $30 million (Canadian).

Since then, NB Power has partnered with IBM to develop an AI-augmented system for outage prediction and resource optimization (OPRO). The resulting mathematical model can predict the effects of coming weather on the NB Power grid.

"Our legacy tools excel at managing outages after they've occurred," wrote Tony O'Hara, chief technology officer and vice president of engineering for NB Power. By contrast, "OPRO's machine learning looks three days ahead to forecast the location and intensity of coming weather. It helps to fill in specific details our operations people need to know. How severe will the impact be? Where should we position our resources? Which equipment should we reinforce? Will we need to bring in crews from outside the company?"

This technology investment quickly yielded financial and operational benefits. Prior to a Christmas 2017 winter storm, NB Power opted not to call in costly external outage restoration resources because IBM's projections of the storm consistently predicted minimal local impacts. "Fortunately, we were right," noted O'Hara. "Most of the storm passed us by and we quickly restored the few outages that did occur."

Less than a month later, in January 2018, Outage Prediction projections indicated that a forecasted storm would hit New Brunswick hard. Heeding this warning, NB Power expanded internal resources and contracted for additional crews, positioning them in expected trouble spots. That storm caused nearly 400 outages — but with excellent preparations, NB Power restored power to 90% of affected customers within 24 hours.

Does Your Utility Have the Data It Needs?

Utilities should assess where their vegetation and weather data is coming from, how it's being analyzed and the accuracy of past predictions of storm damage. Questions to consider include:

• How current and comprehensive are the weather and vegetation data we currently rely on for planning tree trimming and emergency response?
• In recent years, have we sufficiently accounted for season-specific conditions when planning year-round cyclic tree trimming?
• Which poles and transformers (not just power lines) might be at risk from ice and snow loading?
• When have we spent more than was needed to allocate restoration resources?
• How have the costs of weather-related outages created negative ripple effects across the finances of the enterprise?

Once the cost of incomplete, inaccurate or outdated weather and vegetation data are clarified, the value of both rich, high-quality data and advanced modeling and predictive algorithms becomes apparent — not just in winter, but year-round.