In Chile, Google X is taking AI-powered grid tools out of pilot purgatory

Tapestry, the Google X project focused on the electric grid, started out with a question: What would it look like to build Google Maps for electrons?

Since 2021, the moonshot project has been using artificial intelligence to answer that question, and specifically to make the grid more visible for planning and operating purposes.

In the early days of the project, Tapestry partnered with Chile’s grid operator, the Coordinador Eléctrico Nacional (CEN), for a nationwide pilot of the project’s transmission-level grid planning tool. The focus of the pilot was to help the country speed up its plans to phase out coal by 10 years, which would mean moving the target decommission date from 2040 to 2030.

According to Tapestry, over the last three years, the tool has allowed CEN’s grid planners to run simulations 86% faster, and run 30 times as many scenarios at one time. And today, CEN announced, the project is moving beyond pilot status: the grid planning tool will be deployed at full scale for use in Chile’s annual transmission planning process in 2025.

That commitment by Chile, Tapestry’s general manager Page Crahan said, is “a huge validation…that we can take AI tools out of the lab and move them into actual deployment with real world impact.” Chile’s energy transition plans — which call for being fully carbon neutral by 2050 — will ultimately cost hundreds of millions of dollars and impact the lives of the country’s 20 million people, she added.

“This feels like a moment of optimism,” Crahan told Latitude Media. “Instead of everyone hemming and hawing about how it might be and how we might demo, we’ve shown that we can do this, and we’ve done it. And that’s a really important moment.”

Inside the pilot phase

Tapestry’s Grid Planning Tool essentially enables large-scale (nationwide) and long-term (20 years out) simulations of transmission grids on an hour-by-hour basis, using AI to model and visualize the grid at scale. 

The tool itself is built on the foundations of commonly-used interfaces, like Google Maps, with all of the tools a grid planner might need layered on top, running scenarios that account for things like drought, low wind days, heat waves, or even new EV charging stations in Santiago.

In order to test out the tool though, access to data was essential — everything from weather forecasts and historical grid behavior to power generation and energy demand data.

That data might be hard to come by in many places, including the United States. But Chile eliminated that operational barrier, Crahan said, by having open access to data about their network — which is one key reason Tapestry selected Chile as the first national-scale trial run of its grid planning tools. In Chile, Tapestry could hone the tool using actual insights about the national grid, rather than the synthetic data many similar trials rely upon. That meant they could conduct a real-world pilot, and not be stuck in the lab phase.

Chile’s clear end goal — accelerate decommissioning coal in the country by ten years — also gave the pilot clear parameters, said Crahan, and came with an organizational commitment from CEN.

“We were looking for a demonstrated commitment across the organization, not just the board of directors, but also the people that are on the front lines using the tool to do something differently,” she said.

Finally, Tapestry wanted to pilot in an environment where the technology could truly solve a problem, Crahan added: make processes “not 10% better than what they’re doing today, but 10 times better,” she said. “And specifically for the planning tool, because of how challenging this is, we actually said out loud to Chile that our goal is to speed up the transmission planning by 100 times.”

Beyond Chile

As with all of Alphabet’s so-called “moonshot” projects, the goal of Tapestry was to be ahead of schedule with its solution, before its target market was quite ready for it, Crahan said.

“We’re trying to time being a little early [in] understanding the problem, but right on time with being able to deliver something people can use,” she added.

Grid congestion and the challenges that come with planning for high volumes of renewables were problems Tapestry identified early on. And so did Chile, which already faces the challenges of integrating an extremely long and narrow geography ranging from the Atacama desert in the north (with some of the world’s highest solar irradiance levels) to the windy Patagonia region in the south.

“Chile saw navigating the energy transition as something they were not going to be able to do using their current tools,” Crahan said.

They were a little ahead of what most countries were ready for at the time the pilot kicked off in 2021. But in 2024, Crahan said, there’s likely to be more demand; rising global concerns about load growth has coincided with the tool’s readiness for national-scale grid planning

“In 2024 we are able to offer a solution when it’s falling on more receptive ears,” she added.

Chile isn’t the only place where Tapestry is piloting its tools; they’re also working in five other countries on various types of projects, many targeting distribution assets. But Chile will be the project’s first national-scale endeavor focused on transmission.

Applying the grid planning tool to the U.S., where access to data is undoubtedly more challenging, is “a top goal” for Tapestry moving forward, Crahan said: “We are hoping to partner with and start rolling this technology out in the U.S. and look forward to making that a reality as soon as we can.”

Page Crahan will be speaking about "Meeting the load growth challenge" at Transition-AI 2024 in December. Registration for the conference is now open!