After a cluster of cancer cases appears near a military base, health officials find a possible link to contaminated well water around the base. But where did the chemicals in the water come from? And who else might be at risk? As he searches for answers, Dr. Gnanamanikam “Kumar” Mahinthakumar spends more time at his computer than he does examining the area around the wells.

An associate professor of civil engineering, Mahinthakumar has developed computer models to track the spread of groundwater contamination and devise the best remediation strategies. Drilling test wells and using geophysical imaging to learn the subterranean geology doesn’t provide a complete picture, he says, especially in terms of hydrology. “You can only take certain measurements in the ground, which means you’re estimating much of the time,” he says. “You need inverse computer modeling to infer environmental characteristics, such as where and when the contamination started and how it’s spreading.”

Following the movement of groundwater in three dimensions requires millions of calculations, so Mahinthakumar taps into the National Science Foundation’s (NSF) TeraGrid to run his simulations. TeraGrid combines high-performance computers and data resources from nine supercomputing sites nationwide, including Oak Ridge National Laboratory (ORNL), where Mahinthakumar worked as a computational research scientist before coming to NC State. He still holds an adjunct position at ORNL to get access to the resources there.

The native of Sri Lanka is now applying his inverse modeling algorithms to water distribution systems. Together with civil engineering professors Downey Brill and Ranji Ranjithan, he is part of a three-year NSF project using Cincinnati’s water system to study threat management in urban water distribution systems. They hope to develop a computer model that could be used as a monitoring system to limit the impact of contamination—either accidental or intentional—and allow public health officials to trace the source of the problem. Mahinthakumar says such a system might have helped Cary, North Carolina, where officials had to cut off water to most of the town for an entire weekend this summer because of E. coli contamination in one neighborhood.

Monitoring a water system requires less guesswork than tracking groundwater because everything is contained in pipes. But rapidly changing conditions require nimble algorithms for real-time characterization, Mahinthakumar says. “It’s a naturally dynamic system, so we have to be fast to adapt to new data,” he says. “The key, as always, is to narrow down the source of the problem so you can respond properly.”

For more information, please visit