Wildfire season is well underway, but mitigation, prevention, and detection work is a year-round exercise.
In a recent media call highlighting Portland General Electric (PGE)’s wildfire mitigation strategies, the utility underscored two facts: it started its wildfire mitigation programs before being legally required to do so, and the utility has not experienced a catastrophic wildfire to date.
As is the case with other Western utilities, PGE is interested in high-tech solutions. Here’s a look at how PGE is approaching the issue.
Prevention & mitigation
PGE breaks down its wildfire mitigation strategy into a hierarchy of priorities. The long term is entirely about system hardening. For the short term, PGE focuses on improving risk management and situational awareness capabilities and improving detection of high-risk conditions and potential ignitions. Finally, in the immediate term, PGE can mitigate the risk of wildfire ignition in high-risk areas through planned Public Safety Power Shutoffs (PSPS) as a method of last resort.
PGE has issued two PSPSs so far: once in 2020 during a wildfire in the Mt. Hood National Forest, and in 2022 when wildfires prompted the utility to shut off power in 10 PSPS areas. With wildfires and “excessive heat” warnings popping up in several other Western states recently, some utilities have their work cut out for them.
PGE’s high risk fire zones (HRFZs) are constantly changing – some areas are expanded, and some are removed as investments in mitigation are materialized. The utility said 2% of its customers live in a HRFZ, 9% of its overhead lines are within HRFZs, and 4% of its customers have experienced at least one PSPS.
In the longer term, PGE has been focusing on system hardening. To date, PGE has :
- Installed 44 distributed protective devices
- Installed 4 circuits protected by non-expulsion fuses
- Installed 948 ductile iron poles
- Mesh-wrapped 1313 poles
- Reconductored 11.5 line miles
In 2024, PGE plans to complete the installation of 37 protective devices and 2 circuits with fire safe fuses, in addition to mesh wrapping 1200 more poles. Going into next year, PGE plans to finish reconductoring 15 more line miles, and converting 17 more line miles to underground lines.
PGE also completes thousands of pole inspections per year, checking for hardware that is damaged, broken, missing, or loose; in addition to damaged poles, inadequate conductor clearance, and abandoned facilities.
For vegetation management, PGE will have conducted an estimated 50,000 tree trims and 16,000 tree removals this year, in addition to 1065 vegetation patrols in circuit miles. This matters, PGE said, because 85% of PGE’s recorded ignitions have resulted from vegetation contact. Other factors that highlight the importance of vegetation management include increasing drought in the Portland area, increasing tree mortality from bark beetles and other pests, and “significant” tree fuel in HRFZs.
AI-enabled detection
Most of these efforts so far have been designed to mitigate or prevent wildfires. But no system is perfect, and that’s where detection methods come in. In a media call, PGE highlighted its business partnerships with two companies that are assisting in its detection efforts.
PGE’s wildfire detection methods include early fault detection, weather stations, and AI cameras. To date, PGE said it has achieved 100% HRFZ camera detection and weather station coverage, in addition to installing 6 distribution circuits with early fault detection sensors. This year, it plans to complete the installation of 2 more AI-equipped cameras, 5 more weather stations, and 2 more circuits with early fault detection.
The first partnership PGE highlighted was with Pano AI, a company that uses deep learning AI and computer vision to automatically detect, verify and classify wildfire events in real time. There are four main elements to Pano AI’s solution: high resolution cameras, AI detection intelligence, human intelligence in the form of 24/7 analysts, and a utility-focused platform that shows a triangulated location of a incident and alerts emergency responders.
Pano AI’s hardware does best when installed at the right height, said Arvind Satyam, the COO and co-founder of Pano AI, and the cameras can be installed on existing infrastructure like electric poles to meet this goal. The AI algorithm is constantly looking for smoke, and when it thinks it has found some, human analysts are alerted. The analysts can remove false positives like dust clouds or chimney smoke, but if an actual wildfire has been detected, they then start to consider the right size of response.
However, no matter how high they’re installed, cameras can’t see everything. When a camera’s view is obstructed, Pano AI can utilize infrared satellite feeds to provide additional data, seeing where the camera can’t. These two systems work hand in hand and compliment each other: when the infrared satellite picks up a heat signature, it can switch over to a ground camera feed to determine whether a fire has actually started. On the other hand, satellites can be used when a camera’s position doesn’t provide the needed visibility.
It’s an idea that’s catching on in other Western states. Last month, remote utility cameras powered by artificial intelligence helped identify the early makings of a wildfire before it spread in Colorado.
Cameras installed by Xcel Energy and CORE Electric Cooperative detected wildfire smoke caused by a lightning strike on June 16. Within 30 minutes of the identification, emergency services dispatched two helicopters to the fire when it was .25 acres. The helicopters dropped 17,765 gallons of water on the fire over the next six hours, limiting the wildfire to three acres.
The utilities said their wildfire camera system, also provided by Pano AI, identified the wildfire “simultaneously” with the U.S. Forest Service.
In Alaska, the Anchorage Fire Department and Chugach Electric have adopted a new automated monitoring system for the Anchorage Hillside, Alaska’s News Source reported. So far, the fire department has installed 25 FIRE-SPY systems from GRIDWIDE – an “always on” monitoring solution installed on overhead transformers. The system monitors and detects flames, smoke, gas, temperature, and humidity. When an abnormal condition is detected, an automated alert is sent to operators and first responders.
Further West, Hawaii’s electric utility, which is still recovering from wildfires that killed more than 200 people and damaged more than 2,000 buildings last year, is deploying high-resolution cameras equipped with artificial intelligence for early wildfire detection.
Hawaiian Electric is installing two high-resolution cameras at 78 locations across the five islands. The cameras will provide the utility with a full 360-degree view of areas with elevated wildfire risk and can be accessed by the public.
Early fault detection
In addition to the AI-enabled camera system offered by Pano, PGE also uses smart Internet of Things (IoT) sensors from IND Technology. These sensors continuously monitor electrical networks to feed actionable data to utility personnel.
The system remotely and proactively pinpoints common wildfire ignition sources to plus or minus 30 feet, said Andrew Ball, Head of North American Operations at IND Technology. These devices operate before, during, and after weather events, and do not require an outage install. Data is collected on site, sent to the cloud via LTE, then visualized for operators via a web portal.
The system is meant to make utilities aware of faults during blue sky events – the faults detected here are not as immediately obvious as smoke rising from a downed power line. Additionally, the tech is cheaper and faster to install than conducting vegetation management or undergrounding, said Dan Nunez, PGE’s manager of Wildfire Planning and Analytics.
CORRECTION: A previous version of this article incorrectly listed the accuracy of IND Technology’s solution as plus or minus 3 feet. This has been corrected to plus or minus 30 feet.
