The IEEE website will place cookies on your device in order to provide you with the best user experience. By using our website, you agree to the placement of these cookies. To learn more, please read our privacy policy.

Since industry leader Beacon Power filed for bankruptcy in 2011, flywheel has been out of sight of many people. Although the company recovered soon after-and other competitors have joined the market-flywheel has not yet fully returned to the mainstream. But a start-up company hopes to change this situation.

The flywheel battery stores electrical energy by converting electrical energy into kinetic energy by using a motor to rotate the rotor. Electric motors can also be used as generators; kinetic energy can be converted back to electrical energy when needed.

Although interest in flywheels soared in the late 1990s and 2000s, it also had drawbacks. These early flywheel batteries could not store energy for a long time. So flywheels were used more for short-term energy storage, such as providing backup power for five to ten minutes for data centers. Before the bankruptcy, Beacon Power focused on using flywheels as the frequency regulator of the power grid.

But Ben Jawdat, the founder and CEO of Revterra, a Texas-based flywheel startup, believes that his company has overcome these shortcomings, enabling flywheels to store long-term energy for renewable energy.

Revterra's technology benefits from developments in three key areas: rotor materials, motor generators and bearings. Improvements in metal alloys and composite materials have enhanced the strength of the rotor, allowing it to rotate more reliably at high speeds. The new generation of motor generators reduces system energy loss by switching its magnetic resistance (similar to the resistance in the magnetic circuit) to stop energy leakage and make power input and output more efficient when idling. 

But the most important technological development is in bearings, Jawdat said. Previous flywheel storage systems either used mechanical bearings, such as ball bearings, where the bearings were in physical contact with the rotor, or active magnetic bearings, which eliminated friction at the expense of complex and power-consuming control systems. Both of these options will eventually cause a large amount of mechanical energy of the rotor to be lost as waste heat. 

Revterra uses passive magnetic bearings, which can balance the rotor without external control and consume additional energy. By eliminating the energy consumption of the bearing itself, it further improves energy efficiency.  

Revterra's flywheel rotates in the vacuum chamber during magnetic levitation. The picture can be seen from the vacuum-sealed glass window: Revterra

The secret is to use high-temperature superconductors as bearings. This technique not only allows the bearing to lift a very heavy rotor — Revterra’s commercial-scale rotor weighs 7 tons — but it also reduces energy loss due to the bearing’s inherent ability to capture the magnetic field and hold the rotor in place. Revterra's 100 kWh flywheel system only loses 50 watts at idle speed. According to Jawdat, in contrast, many flywheels consume more than 1,000 watts of power. Therefore, if you charge the flywheel battery all the time and discharge it completely, you will only lose about 10% of your energy, he added.

Improvements in the manufacture of superconductors have made them more practical in commercial applications. Moreover, Jawdat said, Revterra's design requires only a small amount of superconducting material, which is maintained at a temperature of around -196 C or 77 K through an off-the-shelf cryocooler—it contains no refrigerant and does not require liquid nitrogen. At the same time, most of the flywheel system is kept at room temperature.

As the United States vigorously promotes renewable energy, the flywheel is now back. A key climate goal of the Biden administration is to make the United States a 100% clean energy economy and achieve net zero emissions by 2050. California — the fifth largest country if it is a country, then its place in the world economy — makes it a state law to achieve 100% renewable energy by 2045.

All renewable energy will also require grid storage. There are many competitors. However, Jawdat said that every leading grid-scale storage technology has flaws. Chemical batteries will degrade over time-the lithium-ion cobalt issue and other procurement challenges have not made the challenge of affordable grid-scale batteries easier. Another popular technology, compressed air energy storage, is cheaper than lithium-ion batteries, but the energy efficiency is very low-about 50%.

This is the market opportunity that Jawdat sees. Compared with lithium-ion batteries, flywheel batteries can basically be used permanently. "You can charge and discharge every day for 30 years, and your [Flywheel] battery will still have 100% capacity," Jawdat said. "For chemical batteries, you must replace them every five to ten years", which will increase the cost of long-term use.

With funding from the National Science Foundation, Revterra built and tested a working prototype 1 kW flywheel system. Jawdat and his team have been working on commercial-scale 100 kWh systems.

Jawdat said: "If we are to truly transition to low-emission renewable energy, we need a way to store a lot of energy without causing a lot of additional negative environmental impact." "I think there is a lot of work to be done in this regard. , [Flywheel] is very promising as a clean energy storage solution."

Note: This story has been updated (5:30 PM, April 7th, US Eastern Time) to reflect the additional information and background provided by Revterra regarding superconductors and magnetic levitation in the flywheel storage industry. 

Intelligent image analysis algorithms provided by cameras carried by drones and ground vehicles can help power companies prevent forest fires

The Dixie fire in northern California in 2021 is suspected to be caused by Pacific Gas & Electric equipment. This is the second largest fire in California history.

The 2020 fire season in the United States is the worst in at least 70 years, with approximately 4 million hectares of land burned on the west coast alone. These West Coast fires killed at least 37 people, destroyed hundreds of buildings, caused nearly 20 billion U.S. dollars in damage, and filled the air with smoke that threatens the health of millions of people. This is done on the basis of the fire season in California that burned more than 700,000 hectares of land in 2018 and the wildfire season in Australia from 2019 to 2020 that burned nearly 18 million hectares of land.

Although some of these fires are caused by human negligence or arson, too many fires are caused and spread by power infrastructure and transmission lines. The California Forestry and Fire Department (Cal Fire) calculated that nearly 100,000 hectares of land burned in the California fire in 2018 was a problem with power infrastructure, including the devastating campfire, which destroyed most of the town of Paradise. In July of this year, Pacific Gas & Electric stated that a blown fuse on one of its utility poles may have caused the Dixie fire, which destroyed nearly 400,000 hectares of land.

Before these recent disasters, most people, even those living in vulnerable areas, did not think too much about the fire risk of power infrastructure. The power company regularly (if not particularly frequently) trims the trees and checks the wiring.

However, the frequency of these inspections has hardly changed over the years, although climate change has led to drier and hotter weather conditions leading to more intense wildfires. In addition, many key electrical components have exceeded their shelf life, including insulators, transformers, arresters and connectors that have been in use for more than 40 years. Many transmission towers (most of which have a construction life of 40 years) are entering the final decade.

There has also been little change in the way inspections are conducted.

Historically, it has always been the responsibility of frontline personnel to check the status of the power infrastructure. If you are lucky and have access, line workers will use bucket trucks. However, when the electrical structure is located in a backyard easement, on the side of a mountain, or out of the reach of a mechanical elevator, line workers must still tie their tools and start climbing. In remote areas, helicopters carry inspectors with optical zoom cameras, allowing them to inspect power lines from a distance. These remote inspections can cover more ground, but they cannot really replace careful observation.

Recently, power companies have started using drones to capture more information about their power lines and infrastructure more frequently. In addition to the zoom lens, some drones also add thermal sensors and lidar to the drone.

Thermal sensors absorb excess heat from electrical components such as insulators, conductors, and transformers. If ignored, these electronic components may generate sparks, or even worse, explode. Lidar can help with vegetation management, scanning the area around the line and collecting data, which the software will use later to create a 3D model of the area. This model allows power system managers to determine the exact distance between vegetation and power lines. This is important because when tree branches are too close to the power cord, they may cause a short circuit or spark from other malfunctioning electrical components.

Algorithms based on artificial intelligence can find areas where vegetation has invaded power lines and process tens of thousands of aerial images within a few days. Buzz Solutions

Combining any technology that allows more frequent and better inspections is good news. This means that, using the most advanced and traditional monitoring tools, major utility companies now capture more than 1 million images of the grid infrastructure and its surroundings every year.

Artificial intelligence is not only suitable for analyzing images. It can predict the future by looking at data patterns over time.

Now is the bad news. When all this visualized data is returned to the utility data center, it takes months for field technicians, engineers, and line workers to analyze it—up to six to eight months per inspection cycle. This prevents them from performing maintenance work on site. And it's too long: by the time of analysis, the data is out of date.

Now is the time for artificial intelligence to step in. It has already begun to do so. Artificial intelligence and machine learning have begun to be used to detect faults and breakages in power lines.

Several power companies, including Xcel Energy and Florida Power and Light, are testing artificial intelligence to detect electrical component problems on high- and low-voltage power lines. These power companies are enhancing their drone inspection procedures to increase the amount of data they collect (optical, thermal, and lidar), and expect artificial intelligence to make these data more immediately useful.

My organization, Buzz Solutions, is one of the companies that provide such AI tools to the power industry today. But we want to do more than just detect problems that have already occurred-we want to predict them before they happen. Imagine what the power company would do if it knew the location of equipment that was about to fail, allowed staff to enter and take preemptive maintenance measures, before the spark triggered the next large-scale wildfire.

It's time to ask if artificial intelligence can be a modern version of the old smoke bear mascot of the US Forest Service: prevent wildfires before they happen.

Damage to the power line equipment due to overheating, corrosion or other problems may cause a fire. Buzz solution

We began to build our system using data collected by government agencies, non-profit organizations such as the Electric Power Research Institute (EPRI), power companies, and aerial inspection service providers that provide helicopter and drone surveillance rental services. Put together, this data set contains thousands of images of electrical components on power lines, including insulators, conductors, connectors, hardware, poles, and towers. It also includes a collection of images of damaged components, such as damaged insulators, corroded connectors, damaged conductors, rusty hardware structures, and broken utility poles.

We worked with EPRI and power companies to create guidelines and taxonomy for labeling image data. For example, what does a damaged insulator or corroded connector look like? What does a good insulator look like?

Then, we must unify the different data, even with images taken from the air and the ground with different types of camera sensors running at different angles and resolutions and shooting under various lighting conditions. We increased the contrast and brightness of some images, trying to put them in a cohesive range, we standardized the image resolution, and created a set of images of the same object taken from different angles. We must also adjust our algorithm to focus on the objects of interest in each image, such as insulators, instead of considering the entire image. We use machine learning algorithms running on artificial neural networks to make most of the adjustments.

Today, our AI algorithm can identify damage or failures involving insulators, connectors, dampers, poles, cross arms, and other structures, and highlight problem areas that require personal maintenance. For example, it can detect what we call flashover insulators-damage caused by overheating caused by excessive discharge. It can also find conductor wear (also caused by line overheating), connector corrosion, wooden poles and crossarm damage, and more.

The development of algorithms for analyzing power system equipment requires determining exactly what components are damaged from various angles under different lighting conditions. Here, the software flags problems with equipment used to reduce wind-induced vibration. Buzz Solutions

But one of the most important issues, especially in California, is for our AI to recognize when and where vegetation grows too close to high-voltage power lines, especially when combined with faulty components, which is a kind of fire country Dangerous combination.

Today, our system can process tens of thousands of images and find problems in a few hours and days, while manual analysis can take months. This is a huge help for utility companies trying to maintain power infrastructure.

But artificial intelligence is not just for analyzing images. It can predict the future by looking at data patterns over time. Artificial intelligence has done this to predict weather conditions, company development, and the likelihood of disease outbreaks, just to name a few.

We believe that artificial intelligence will be able to provide power companies with similar predictive tools, predict failures, and mark areas where these failures may cause wildfires. We are working with industry and utility partners to develop a system.

We are using historical data from power line inspections and historical weather conditions in related areas and providing them to our machine learning system. We ask our machine learning system to find patterns related to broken or damaged components, healthy components, and overgrown vegetation around the lines, as well as weather conditions related to all of them, and use these patterns to predict the future health of electricity Lines or electrical components and vegetation growth around them.

Buzz Solutions’ PowerAI software analyzes images of power infrastructure to find current problems and predict future problems

Now, our algorithm can predict the situation in the next 6 months. For example, there may be five insulators damaged in a specific area, and the vegetation near the line at that time is likely to overgrow. Combined, there is a risk of fire.

We are now using this predictive failure detection system in pilot projects of several major utility companies-one in New York, one in New England, and one in Canada. Since the pilot started in December 2019, we have analyzed approximately 3,500 electrical towers. We detected 5,500 faulty components that could cause power outages or sparks among approximately 19,000 healthy electrical components. (We have no data on repairs or replacements.)

Where do we go from here? In order to go beyond these pilots and deploy predictive artificial intelligence more widely, we will need a lot of data, which is collected in different regions over time. This requires cooperation with multiple power companies and their inspection, maintenance and vegetation management teams. Major US power companies have the budget and resources to collect data on such a large scale through drones and aviation-based inspection programs. But as the cost of drones drops, small utility companies can also collect more data. Making tools like ours widely available requires cooperation between large and small utility companies and drone and sensor technology providers.

Fast forward to October 2025. It is not difficult to imagine that the western United States will face another hot, dry and extremely dangerous fire season, during which a small spark may cause a huge disaster. People living in fire countries should take care to avoid any activities that may cause a fire. But now, they are much less worried about grid risks, because a few months ago, utility workers came to repair and replace faulty insulators, transformers and other electrical components, and trim trees, even those that have not yet reached the power line . Some people asked the workers why all the activities. "Oh," they were told, "Our artificial intelligence system indicates that the transformer next to this tree may produce sparks in the fall. We don't want this to happen."