CSL assistant professor places second in inaugural Intelligent Water Systems Challenge
Artificial Intelligence (AI) has enabled advances in many industries, including healthcare, finance, transportation, and now public utilities, specifically wastewater management. At the forefront of this new area of research is CSL’s Lav Varshney.
“The water and wastewater industry is only just starting to use AI and machine learning technologies,” Varshney, an assistant professor for electrical and computer engineering and chief scientist of Ensaras, Inc., said. “Our goal for this competition was to work with a utility to define problems where AI would be useful, solve those problems using advanced algorithms, and then implement and deploy the solutions.”
The competition referenced by Varshney was the first-ever Intelligent Water Systems Challenge hosted by the Leaders Innovation Forum for Technology (LIFT), a joint effort of the Water Environment Federation and Water Research Foundation. Varshney and collaborators from Ensaras and the University of Illinois at Chicago (UIC) began working with the Metropolitan Water Reclamation District of Greater Chicago (MWRDGC) in April to develop a solution to a problem the utility was experiencing: Odor complaints near their new reservoir system, which serves the surrounding community of 500,000 people.
“The utility themselves proposed this problem and a solution was very much needed,” Varshney said. “We worked to formulate an AI-driven solution that fit in with current operational processes.”
Using AI, Varshney and his team wanted to predict when there would be an odor problem three days in advance of the actual issue. There were several factors that went into determining when a stench would occur, such as the amount of hydrogen sulfide and other compounds being produced by the water (usually the source of the odor), the weather patterns capable of dispersing the odor, and whether or not the people living nearby would be bothered enough to complain about it. The team evaluated the influence that each of these factors had on odor complaints.
“We used the data we gathered to train machine learning algorithms to predict odor events three days in advance with fairly good accuracy,” Varshney said. “One of the keys to our solution was figuring out which features were most important, so there was a direct line to taking action: I think the utility has a good solution now.”
MWRDGC is currently in the process of fully implementing the solution.
Varshney hopes the team’s results will not only prove useful for wastewater utilities globally who are dealing with any type of odor issue but also demonstrates that AI techniques can help solve problems throughout the water and wastewater industry.
The MWRDGC was not alone in their support of the research results. After reviewing entries from all over the country, the LIFT competition committee awarded Varshney’s team with second place overall.
“It’s exciting,” Varshney said. “Especially in terms of industry adoption. Since machine learning is fairly new to the wastewater space, it’s good to validate that it’s actually useable and impactful.”
A next key target for Varshney in wastewater management is to enhance remote automation technology to safely optimize operations and maintenance functions for plant operations.