
Xudong Fan's research suggests that AI can help utilities better predict water main breaks.
Release Date: February 9, 2026
BUFFALO, N.Y. – In colder climates during the winter, water main breaks are not uncommon.
The soil freezes and thaws, causing underground pipes in municipal water systems to crack or burst, which leads drinking water to spill out into yards and streets. It’s a scene that has played out in Buffalo and surrounding communities in recent weeks.
Xudong “Andrew” Fan, PhD, assistant professor in the Department of Civil, Structural and Environmental Engineering at the University at Buffalo, studies municipal water systems.
“Fundamentally, water main breaks occur when the internal and external stress on a pipe exceeds the pipe’s structural strength capacity,” he says. “During the winter, when you have freeze and thaw cycles, this leads to soil deformation and adds additional stress to the pipes, which potentially is the main reason of a higher number of pipe breaks.”
Other factors at play include older pipes made of materials that do not expand or contract easily, corrosive soils and excavation work, says Fan.
While studying at Case Western Reserve University, Fan co-authored a paper that explored how machine learning (a subfield of artificial intelligence) can help predict when underground water pipes are likely to break. The researchers found that AI can outperform traditional methods, potentially giving utilities a tool to improve the reliability of aging infrastructure and save money over time.
“Overall, we found that pipe failures are the result of physical factors, such as the age, material and size of the pipe; climate factors, including temperature and precipitation; and operational factors, such as the amount of water pressure and how it’s delivered,” says Fan.
Cory Nealon
Director of Media Relations
Engineering, Computer Science
Tel: 716-645-4614
cmnealon@buffalo.edu