Carollo Engineers To Lead Full-Scale Application Of Machine Learning To Wastewater Disinfection Control

The Water Research Foundation project will test ML-based chloramination control and develop a framework for water reclamation plants nationwide

Walnut Creek, CA - A new study from The Water Research Foundation (WRF) will evaluate how real-time machine learning (ML) can help wastewater utilities optimize disinfection chemical dosing to improve compliance and lower operating costs. Led by Carollo Engineers, with Ekster & Associates and the Los Angeles County Sanitation Districts (LACSD) as co-investigators, WRF project 5385, Machine Learning Meets Disinfection: Full-Scale Wastewater Chloramination Control for Compliance and Cost Savings, will demonstrate and evaluate a full-scale ML-based disinfection control framework at a water reclamation plant (WRP). The research will also develop practical implementation guidance that other utilities can adapt to their own facilities.

The project builds on the prior work under WRF project 5148, Transforming Aeration Energy in Water Resource Recovery Facilities through Suboxic Nitrogen Removal, where the research team successfully implemented ML-based aeration control at LACSD's Pomona Water Reclamation Plant (POWRP), demonstrating how advanced control strategies can improve performance in complex wastewater treatment processes. The work resulted in a 57% drop in blower energy consumption, significant nitrate reduction, and chemical cost savings for chloramination.

This new effort extends that approach to wastewater disinfection, where operators must continuously balance regulatory compliance, changing plant conditions, and chemical cost management.

Many chemical disinfection systems rely on conventional feedback control loops that respond to current conditions but cannot anticipate rapid changes in flow, disinfectant demand, or water quality. As a result, facilities often maintain conservative chemical dosing margins to ensure compliance, which can increase operating costs and chemical consumption.

This project will evaluate whether ML-based predictive control can improve that balance by anticipating process changes and adjusting dosing in real-time, while maintaining operational reliability and regulatory performance.

California recycled water regulations require maintaining a minimum disinfectant concentration x contact time (CT) value of 450 mg-min/L (CT450). Historical operating data at LACSD’s POWRP indicates the facility has typically maintained substantially higher CT values to preserve compliance margin under varying conditions. The project will evaluate a staged pathway toward lower, optimized CT operation, with projected annual chemical savings of $80,000 at the demonstration facility and broader applicability across LACSD’s system.

Over the 24-month study, the team will:

• Develop predictive ML models for key process drivers including flow variability and disinfectant decay.
• Integrate control logic for sodium hypochlorite and ammonia dosing.
• Deploy additional online instrumentation needed to support real-time monitoring and closed-loop control system operation.
• Integrate the control framework with plant SCADA systems using appropriate cybersecurity safeguards.
• Validate performance in shadow mode before transitioning to operator-supervised automated control.
• Conduct full-scale testing under real operating conditions.

Dr. Natalie Beach, principal investigator for the study and east region wastewater lead at Carollo, said: “When compliance is on the line, conservative dosing makes sense. This project is about evaluating whether machine learning can better anticipate changing conditions and safely adjust disinfection dosing in real-time, with a phased approach that allows operators to build confidence before transitioning to direct control. Ultimately, a primary goal of this initiative is to determine whether machine learning can deliver similar improvements in operational efficiencies and cost savings to what our research team achieved when applying machine learning to the optimization and automation of the biological nutrient removal process at the Pomona Water Reclamation Plant.”

Upon completion, the project team will publish a final report documenting the implementation considerations, key lessons learned, and practical guidance for utilities considering advanced disinfection control strategies. Findings will also be shared through WRF communication channels including technical webinars, a fact sheet, and conference presentations.

About Carollo Engineers

For over 90 years, Carollo Engineers has provided a full range of innovative planning, design, and construction management services to address the water needs of municipalities, public agencies, private developers, and industrial clients. Carollo develops robust water management strategies that leverage collaboration, provide multi-benefit solutions, and achieve sustainability and resilience. Carollo has 1,900 employees located in more than 60 offices across North America. To learn more about how Carollo is “Working Wonders With Water®” call (800) 523-5826 or visit www.carollo.com

About The Water Research Foundation

The Water Research Foundation (WRF) is the leading research organization advancing the science of all water to meet the evolving needs of its subscribers and the water sector. WRF is a nonprofit, educational organization that funds, manages, and publishes research on the technology, operation, and management of drinking water, wastewater, reuse, and stormwater systems—all in pursuit of ensuring water quality and improving water services to the public. For more information, visit www.waterrf.org