Newswise — Most of us never think about what happens to water after it swirls and flows down the drain. But for those who operate wastewater treatment plants, keeping the promise of clean water is a race against time and data.
Treatment plant operators face major challenges. Critical water testing takes several days to complete, but operators often need to make decisions within hours.
Explainable AI (XAI), a growing field of artificial intelligence, is helping fill that gap, said Fuad Nasir, a water expert with the Wisconsin Department of Natural Resources.
During his graduate studies, Nasir (’25 Civil Engineering Ph.D.) thoroughly researched research on collecting data from local treatment plants. And he realized that something important was happening in the broader technology world. “AI and machine learning are now being used in many fields, but in the U.S., AI hasn’t been used much in wastewater treatment,” he said.
trust issues
The problem wasn’t a lack of interest, it was a lack of trust. Traditional machine learning can be compared to a black box that takes in data and generates predictions.
“Carriers were hesitant to use it because they couldn’t see the process that led to the predictions,” Nasir said.
For wastewater treatment companies, which make high-stakes decisions about chemical dosing and treatment schedules, that lack of transparency was a deal-breaker.
That led him to explainable AI. XAI does more than just predict outcomes. Highlight the variables that shaped that prediction. “It reveals what’s going on in the background,” Nasir said. “When you apply XAI, you can literally visualize it.”
real example
Let me explain why this is important with a key example. Wastewater treatment plants rely on a laboratory test called biochemical oxygen demand (BOD) to measure how much organic material remains in the treated water. High BOD can have negative effects on aquatic life, so operators must adjust treatment chemicals to keep BOD low. However, a BOD inspection takes five days, which is too long for operators to judge.
XAI Guide’s models use historical data to predict BOD and show which factors, such as temperature, flow, and ammonia, are causing levels to rise or fall.
XAI has grown rapidly. While explainable AI was introduced earlier in the medical field, wastewater research is now catching up, citing Nasir’s work in the process.
“In the last few years, we’ve seen a huge increase in people using it in wastewater,” he said.
He acknowledged that real-world deployment will take time because utilities need equipment, training and funding, but he said the direction is clear.
Drawn by UWM research environment
Nasir arrived at UWM as a master’s student from Bangladesh in 2019, drawn to UWM’s strong research culture and environmental engineering faculty, including his advisor, Professor Jing Li.
“UWM is ranked R1, a top research institution, which definitely caught my attention,” Nasir said.
A year later, he transferred to a Ph.D. Currently, Mr. Nasir focuses on aspects of public water systems related to health and regulation. He believes that as AI advances rapidly, technical expertise will become increasingly valuable in the regulatory field.
His advice to undergraduates? Please be careful. “The thing about AI and machine learning is that they change very quickly.”
