BETA
This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

As It Undergoes Transformation, U.S. Power Grid Embraces AI

Following
This article is more than 2 years old.

The U.S. electric power grid, once viewed as the most stable and secure power system in the world, is in the midst of undergoing a systemic transformation as grid operators are being forced to adapt to multiple challenges. Among these challenges are an anticipated surge in electricity demand, brought on by the proliferation of new electric vehicles and the digital systems needed to support the “Internet of Things,” and new generation coming on line from renewable sources — with their inherent intermittency issues — along with electricity storage capacity from next-generation rechargeable batteries.

With its ability to process large volumes of data to come up with solutions to complex problems, artificial intelligence (AI) technology is helping to guide this grid transformation. Industry experts agree that AI will be necessary to form the basis of the smart grid of the future.

“There are two things we look at. We need to decarbonize the planet,” said Colin Parris, chief technology officer at General Electric division GE Digital. “At the same time, you want to manage how you retire your coal [generation plants] and transform the gas [generation plants’] capabilities.”

In addition, steps must be taken to shore up the resiliency of the grid, to prevent significant power interruptions caused by cataclysmic weather events. The vulnerability of the electric grid to severe disruptions was demonstrated in two different states for widely disparate reasons within the past year.

In Texas, the deaths of more than 100 people have been blamed on power outages resulting from a severe winter storm in February. Last August, a record-breaking heatwave in California led grid operators to institute rolling blackouts that affected 800,000 homes and businesses across the state.

Stabilizing the grid

“We have to stabilize the grid,” Parris said. Unlike in the past, when the bulk of power was generated by stable sources such as coal or natural gas, the grid of today increasingly includes intermittent sources such as wind and solar, which don’t provide power when the wind isn’t blowing or the sun isn’t shining. This increasingly requires grid operators to employ advanced AI technology to ensure the grid remains in balance.

“People can’t depend on one smooth flow of electricity or one flow of demand inside of the utility. It’s going to be fluctuating,” Parris said. “That is all run by software: the grid software, the transmission software, distribution software, market-management software.”

AI software is also being deployed to accommodate the addition of battery storage onto the grid as that technology becomes more mature and is commercialized. Stem, a Silicon Valley-based battery storage company, uses AI technology to determine the most optimal and economic times to recharge the batteries and to release the energy onto the grid.

“When you think about it, the battery doesn’t do anything by itself, so you need the intelligence to understand how best to use that battery,” Larsh Johnson, Stem chief technical officer, said.

The AI software helps the operators forecast what individual customers’ load patterns are going to be, when they’re going to be consuming power and what the cost of power will be during different times of day. “For our customer class, it’s quite dynamic. It changes from hour to hour, sometimes even every few minutes,” Johnson said.

Balancing power supply and demand

AI technology is playing an increasing vital role in managing the electric grid to ensure that there is power available when and where it’s needed. This is going to become more important in the future as the demand for electricity is expected to rise, with consumers increasingly purchasing smart devices able to transmit and receive data via the Internet, commonly known as the Internet of Things.

In addition, the adoption of electric vehicles (EV), which is expected to ramp up dramatically in the coming years, will have a significant effect on electricity demand and the timing of that demand. For example, power demand in a given area could spike as suburban commuters all return from the office in the evening and plug in their EVs to recharge for the night.  

“Taking into account consumer behavior to ensure that supply matches demand as much as possible is a very large puzzle,” said Steve Kwan, director of power generation for Beyond Limits, an AI technology company.

“This is a perfect application of artificial intelligence, because you can take into account many variables and be able to provide a recommendation in a very timely manner to support the changing needs of the consumer on a 15-minute basis,” he said. “Using traditional physics-based modeling is inefficient or too slow.”

AI to help in carbon reduction

AI is also helping grid operators reduce their overall carbon footprint in several ways. One way involves the use of cognitive computing, based on AI and signal processing technologies, and neural networks, computer systems patterned after the networks in a human brain.

Because power generated by wind and solar energy is usually cheaper than power from natural gas-fired plants — as well as having greater climate benefits — grid operators tend to rely on these forms of renewable energy as much as they can. However, because of the intermittency of these renewable resources, they must be backed up by a more reliable form of generation, typically a combined-cycle gas-fired plant.

As the levels of renewables flow up and down the operators of the combined-cycle plant have to continuously increase and decrease the electricity they produce. Because the plants are designed to operate at a particular level, this up and down cycling decreases efficiency and burns more fuel than necessary, thereby contributing to the plant’s overall carbon emissions.

“Operators have a tendency to operate in a very narrow region of the operational envelope, simply because of historical behaviors,” Kwan said. “By applying AI and neural-network technology we can have the plant operate in a much more efficient manner than before.”

Another AI application, developed by AI software company C3.ai, helps lower the carbon emissions of a power grid by reducing “the amount of fuel that you need to power the grid by 8%” according to C3.ai founder, Chairman and CEO Tom Siebel.

Siebel said in the future every successful energy company will be employing AI technology in its operations.

“Those companies who adopt AI will be delivering cleaner, more reliable, safer energy with much lower environmental impact. Those that don’t adopt will be acquired by those that do,” he said.

Follow me on Twitter