Team looking to tap underground ‘thermal batteries’ to cool AI data centers, save water

Researchers at the Prairie Research Institute at the University of Illinois Urbana-Champaign are exploring aquifer thermal energy storage (ATES) systems as a sustainable cooling solution for artificial intelligence data centers across the United States. Led by Illinois State Geological Survey researchers Yu-Feng Lin, Andrew Stumpf, and postdoctoral researcher Upasana Pandey, a new study published in the journal Groundwater details how ATES can utilize natural underground temperatures and groundwater to act as a giant "thermal battery." This innovative approach aims to alleviate the substantial environmental impact of data centers, which often allocate 10-40% of their energy consumption to cooling and use immense volumes of water, primarily through evaporative processes that deplete local supplies.

The ATES system operates by pumping cool groundwater from an aquifer through subsurface pipes into a data center to absorb heat via a heat exchanger. The now-warm water is then returned underground for storage and future use, reversing the process in different seasons to store cold water in winter for summer cooling and heat in summer for winter heating. This method significantly reduces the temperature differential that conventional cooling systems must manage, leading to considerable energy savings.

Illinois is identified as particularly well-suited for ATES implementation due to its distinct seasonal temperature variations, abundant aquifers, and favorable thermal properties of its glacial deposits. The research also suggests that water sources for ATES do not need to be potable, pointing to deeper, salty, or even contaminated aquifers and abandoned mines as potential sources. While ATES systems have higher upfront costs compared to traditional cooling methods, the researchers argue for their long-term economic and environmental benefits, emphasizing the need for project evaluations that consider 20- to 40-year lifespans.