The decentralized, bi-directional power of smart grids

After seeing the Texas’s winter blackouts to NYC’s late-July heat wave-driven “conserve energy” alerts, tech experts say American power grids have room for improvement, and are introducing the concept of “smart grids,” which are management systems that use a combination of sensors and AI to distribute and conserve energy.

“If we make grids smart...instead of tens and hundreds of large-scale generators, we can power the electricity supply locally through a larger number of smaller power plants,” says Yury Dvorkin, assistant professor of electrical and computer engineering at NYU Tandon.

Under this decentralized, bidirectional model, utilities can pay their customers for surplus energy at peak-demand hours to bolster the broader grid, and, ideally, use AI and sensors to more closely match energy supply with demand.  

For all of the potential benefits, building a smart grid is not foolproof. For one, with more connectivity comes more cybersecurity concerns and privacy considerations. It also requires significantly more coordination. “Instead of one nuclear power plant which is perfectly controllable, where you have engineers who are smart or intelligent, who know how this equipment works, you have to roughly coordinate 50,000 solar panels,” Dvorkin said. “That’s a big challenge because you no longer have onsite engineers. You no longer have perfect controllability. And most importantly, you have no control over the sun or wind.”