Digital twins for utilities are proliferating — but challenges remain

As utilities battle with aging infrastructure, extreme weather events, and capacity constraints, the number of AI-powered technologies supporting them is growing, with new solutions rolling onto the market all the time. 

The utility asset management company Sharper Shape’s newly released Asset Insights is the latest. The digital twin software uses artificial intelligence and machine learning to detect damage in utilities’ infrastructure components. (A digital twin is a virtual replica of physical devices, such as an utility’s assets.) It aims to make inspection and maintenance of utilities’ sprawling assets both simpler and cheaper. 

The software, launched this week, analyzes thousands of images, detecting damages that would take hours of manual labor for humans to catch. These tend to be small things — rust, leaks, and cracked insulators — that compromise the safety and reliability of utilities’ assets if not repaired promptly. Sharper Shape said the tool can identify at least 40 different distribution and transmission components. 

The company is part of a growing number of solutions providers leveraging AI to monitor the health and safety of utilities’ assets. The group includes companies like Buzz Solutions, Arkion, mPrest, and Looq AI, among others.

Matt Casey, who leads the Latitude Intelligence research team, says there are three big reasons why we’re seeing the emergence of so many AI-enabled solutions focused on asset management: utilities prioritizing reliability and safety as growing demand and aging infrastructure strain systems; the fact that many of these solutions offer straightforward returns on investment — both in terms of operational cost savings and progress towards key reliability goals; and growing data access that is enabling vendors to build more sophisticated tools. The latter point is crucial, as an unprecedented amount of different types of data — from satellites and drones, as well as the ability to model and replicate systems via AI — has been unlocked in recent years.

“We’re not necessarily going to see a market explode overnight, but we’re starting to see the stars align for vendors developing AI-enabled asset management solutions. Asset and system health and reliability are a top priority for utilities right now, and this focus and demand is coming at a time when vendors are gaining access to new, more granular data, and have the ability to combine and model it like never before given advancements in AI and machine learning.” Casey led Latitude Intelligence’s research into the market for AI for utilities earlier this year.

At the moment, leveraging AI and ML to detect and manage potentially dangerous situations is the number one utility use case of AI, according to a recent report by Itron, a utility tech company. Other key uses include siting and permitting, grid planning, and grid operations.

But what are the hurdles?

As essential as AI is becoming for utilities — and nearly everyone in the sector agrees on that point, according to Itron’s survey of 600 utility executives — its adoption comes with challenges. A lack of expertise, for instance, is currently the top concern of utilities looking to embrace AI, according to Itron’s report. 

“Deploying AI in a utility context isn’t a flip of a switch; in some cases it can be as complex as designing a system from the ground up,” the report says. “This demands technical expertise to design, train, test and implement such a system.” 

And cost is the second major barrier. According to research by consulting firm Gartner, “global AI software spending in the power and utilities market is forecast to increase 19.9% in 2024 to $9.8 billion and reach $17.8 billion by 2027.” 

In an article published a few days before Sharper Shape’s release of its latest software, Kristy McDermott, the company’s vice president of sales, notes that things like data compatibility and system integration are also big hurdles. 

“Many utility companies operate on outdated platforms that are not readily compatible with the latest AI software, requiring extensive customization and sometimes complete system overhauls,” she wrote. 

Data quality can also be an issue, as AI requires high-quality data to operate at its best, and “utilities often have vast stores of unstructured or inconsistent data” that needs to be reorganized, she added. 

These implementation roadblocks come alongside the real major challenge AI poses for the grid: a dizzying increase in electricity demand driven in great part by power-hungry data centers. Research by Goldman Sachs predicts U.S. utilities will need to invest at least $50 billion in new generation capacity just to meet the demands of data centers.