Team looking to tap underground “thermal batteries” to cool AI Data centers and save water

Researchers at the University of Illinois Urbana-Champaign's Prairie Research Institute are investigating aquifer-based geothermal systems, known as aquifer thermal energy storage (ATES), as a solution for cooling AI data centers in the United States. Led by Illinois State Geological Survey researchers Yu-Feng Lin, Andrew Stumpf, and postdoctoral researcher Upasana Pandey, the new study published in Groundwater proposes that ATES could significantly ease the environmental strain of data centers by reducing electricity demand for cooling and conserving water.

The proposed system works by pumping cool groundwater through subsurface pipes into data centers, where it absorbs heat via a heat exchanger. The now-warm water is returned underground and stored. In seasonal climates like Illinois, excess heat from summer cooling can be stored for winter heating, and cold groundwater can be stored in winter for summer use. The researchers note that Illinois is uniquely suited for ATES due to its pronounced seasonal temperature differences, abundant aquifers, and the favorable thermal properties of its glacial deposits.

While ATES systems may have higher upfront costs compared to traditional cooling methods, they offer substantial long-term operational savings. The researchers also emphasize that the water used does not need to be potable, suggesting sources like deeper, salty aquifers or contaminated groundwater. The primary obstacles are economic evaluation horizons, which often favor short-term returns over the 20- to 40-year benefits of ATES, rather than scientific or technical feasibility.