PJM and Alphabet on April 10 announced a partnership to develop a suite of new tools using artificial intelligence to speed the RTO’s generation interconnection process.
Amanda Peterson Corio, head of data center energy at Google, said grid planners face an explosion in the number of new service requests they have received, straining their ability to process applications. Google sister company X Development is leading the initiative to build on its Grid Planning Tool and Grid Aware software to create a streamlined work environment PJM can use to more quickly bring new generation onto the grid at a time when the RTO is sounding alarm bells about future resource adequacy.
The planning tool has been deployed in Chile to simulate the grid 20 years into the future with hourly granularity, while Grid Aware uses visual information from sources like Google Maps to facilitate inspections and identify where repairs may be needed.
“This initiative brings together our most advanced technologies to help solve one of the greatest challenges of the AI era: evolving our electricity systems to meet this moment,” Corio said. “We see the opportunity to help secure America’s electricity needs with the many resources seeking to provide energy to the grid and believe this work with PJM is a great catalyst for innovation across the United States.”
The sluggish pace of new entry is one of three contributors to a potential capacity deficiency that PJM has identified in the 2029/30 delivery year, alongside generation deactivations and ballooning load largely fueled by data centers. Executive Vice President of Operations, Planning & Security Aftab Khan said the RTO’s shift to a cluster-based approach to studying interconnection requests is allowing it to more expeditiously work through its backlogged queue, but it still will take about two years for projects to go through the process. Integrating more artificial intelligence into those studies can add more efficiency and quality to studies, he said. (See PJM Reaches Milestone on Clearing Interconnection Queue Backlog.)
“Innovation will be critical to meeting the demands on the future grid, and we’re leveraging some of the world’s best capabilities with these cutting-edge tools to further reduce completion times for new service requests,” Khan said. “PJM is committed to bringing new generation onto the system as quickly and reliably as possible.”
Renewable developers and consumer advocates have pointed to PJM’s interconnection queue as a central obstacle to getting clean energy onto the grid and allowing generation owners to respond to high capacity prices. In a complaint filed at FERC, Pennsylvania Gov. Josh Shapiro (D) argued for a lower maximum capacity price on the grounds the new generation cannot respond to price signals sent by upcoming Base Residual Auctions. PJM has defended the process, saying more projects are clearing the queue but are becoming mired in other issues challenging development, such as supply chain constraints and permitting requirements. (See PJM Presents Capacity Price Cap and Floor to Members Committee.)
Page Crahan, general manager of X’s electric-oriented Tapestry, said a core challenge grid operators face is information being siloed across disparate information streams and tools, an environment she said could be streamlined through using Google’s expertise in data management to create a unified model of the grid, pulling together the output of existing tools to create a “knowledge graph.” She said the name Tapestry was chosen to represent the goal of creating a platform that can stitch together the fragmented elements of the grid.
Speaking during a press conference ahead of the announcement of the partnership, Crahan said one area that could be improved by adding AI is processing PDF applications submitted by generation owners with new projects. Assessing the information in those files creates a bottleneck in the study process, where planners have to consult multiple tools, models and datasets when modeling how a new generator may impact existing equipment. She also gave the example of using AI to aid in validating information provided in interconnection applications; rather than planners having to refer to multiple documents to determine whether the land rights are associated with the correct builder, she said Tapestry software could sift through those files.
Tapestry already has partnered with system operators across the globe, including developing “near real-time grid virtualization” software to simulate AES’ distribution grids in Ohio and Indiana, as well as advanced inverter technology working with Australia’s Commonwealth Scientific and Industrial Research Organization.
Crahan said Chile’s National Electric Coordinator (CEN) has deployed the Grid Planning Tool to allow planners to simulate its grid 86% faster, allowing 30 times the number of scenarios to be run. Google’s DeepMind software also has improved CEN’s weather forecasting for wind.
Unlike those other projects, Crahan said the work with PJM will be the “first of its kind” to integrate AI into the modeling interconnection study process of a large grid coordinator. X is aiming to deliver the first tools to PJM in 2025, she said.
In response to questions on how the effort to speed interconnections may interact with President Donald Trump’s executive order April 8 seeking to ease regulations on coal generation, Khan said PJM is fuel agnostic and will welcome any resource that can improve reliability. He added there are many factors that can impact the viability of coal, including the growth of gas generation. (See related story, Trump Seeks to Keep Coal Plants Open, Attacks State Climate Policies.)
Corio said Google remains dedicated to its climate goals and will continue to seek clean energy sources that can provide firm capacity. She specified that coal is not a clean technology under that framework.
