How do AI data center fires start?
A researcher at Texas A&M University is exploring the increased fire risks associated with the rise of AI data centers.
According to a university announcement Tuesday, chemical engineering Ph.D. students Tylee Kareck and Chi-Yang Li published a paper, with George Washington University and the University of California, Berkeley, that analyzes the common causes of such fires and provides strategies for reducing risk.
“Our work provides insights to assess fire risk so engineers can design safer and more resilient data centers,” Qingsheng Wang, a professor in the Artie McFerrin Department of Chemical Engineering, who contributed to the paper, said in the announcement.
The researchers found that data center fires can start with battery failures, electrical faults, equipment malfunction and human error. Centers with many batteries can reportedly start fires when “thermal runaway” occurs, when chemical reactions in batteries create a great amount of heat.
“While addressing individual causes can reduce risk, holistic fire prevention and mitigation efforts are necessary to properly protect infrastructure,” Kareck said.
According to the announcement, the researchers believe large-scale testing and improved data collection will enable a better understanding of how data center fires start and spread.
