Digital Solutions Advancement Roadmap for Optimization in Wastewater Treatment Operations
Wednesday, September 30, 2026 9:00 AM to 9:15 AM · 15 min. (US/Central)
New Orleans Convention Center
Interactive Technical Session
Room 238
Information
The wastewater (WW) industry is evolving with digital technologies like AI, advanced sensing, and integrated data platforms that redefine treatment monitoring and optimization. As utilities face aging infrastructure and stricter regulations, digital tools provide opportunities to leverage real-time insights, predictive decision‑making, and efficient operations. Engineers have used these tools for design and troubleshooting, but those same tools are not readily used by operations. This paper will present a proposed roadmap using case studies for migrating those tools to operations and to guide utilities planning for a digital future. Figure 1 displays the proposed progression of digital advancements. These digital solutions do not have to be implemented sequentially as presented and each solution can be implemented based on the needs of the utility and operations. Descriptions of each step are below.
1. A first step into the digital world is installing additional instrumentation and implementing advanced control algorithms utilizing the instrumentation. Boschmans, et. al, 2019, documents success with advanced real-time process controls such as ammonia-based aeration control, secondary clarifier guidance, aSRT control, and more at facilities across the nation (Figure 2).
2. Mechanistic digital twins (DTs) provide a digital tool based on a calibrated process model for simulating scenarios, training staff, and planning operational changes in a risk-free environment. The DTs use theoretical scenario inputs, such as reduced units in service, process control adjustments, or new chemical additions, to train staff. These "what-if" scenarios enhance both training effectiveness and strategic planning during plant operation transitions. Two case studies of DTs utilized for training and planning will be reviewed (Figure 3).
3. A secure one-way data from on-prem to the cloud connection enables controlled, outbound-only transfers that reduce cybersecurity risk by preventing inbound connections while still supporting cloud-based analytics.
4. Quality data is essential for informing decisions and optimizing processes, especially as utilities adopt advanced digital solutions and real-time controls. Prioritizing data quality and governance ensures the information driving automation remains accurate, reducing errors and enhancing performance. This paper discusses machine learning (ML) approaches (Figure 4) that could be deployed to perform real-time error and data anomaly detection.
5. Understanding what is coming into plants is crucial in optimization of the treatment process. Accurate prediction of flows and loads enables operators to anticipate imminent conditions and adjust treatment proactively. This foresight helps utilities improve process efficiency, reduce operational costs, and maintain regulatory compliance. The projections can be paired with mechanistic modeling for guidance on units in service and anticipated chemical dosing. Influent flow and load prediction models for a 75 mgd plant will be discussed (Figure 5).
6. DTs can be fed live influent flow and load predictions for real-time process modeling updates. The DT would predict treatment performance and provide recommendations on process setpoints and areas for optimization for operators to consider. DTs can act as pseudo-soft sensors by predicting metrics for parameters not being monitored within the plant.
7. Data-driven models combine subject matter expertise with data science tools to improve our understanding of complex environmental systems like WW treatment and bridge gaps in the current state of mechanistic models. Insights include improved decision making, enhanced efficiency, proactive issue detection, and real-time adaptability. Multiple deployed data-driven models (Figure 6) will be presented as case studies.
8. Integrating digital tools enables holistic optimization across processes and assets within a WW facility. This is accomplished by implementing comprehensive optimization algorithms (Figure 7) that utilize both data-driven and mechanistic models, incorporating information from sensors, SCADA systems, and additional data sources to rank and prioritize operational strategies in real time. Such approaches not only alleviate the workload on operational staff but also enable incremental improvements that can cumulatively result in a substantial return on investment.
9. In the future, treatment plants could be run completely based on instrumentation and digital solutions with minimal human involvement. The autonomous control of a facility would include the previous steps in the digital advancements progression. In pilot/advisory mode (Figure 8), the autonomous control program would make predictions based on input data and advise the operators on decision making. In this case, the operators are the decision makers.
10. For cloud-based ML programs to directly control operations, a secure data connection between on-premise systems and the cloud will be required.
11. When fully deployed, the autonomous control program would completely operate the plant with minimal human intervention. The digital solutions would be the decision makers. Human intervention would occur when there were process upsets or alarms triggered from the programming or digital solutions.
This structured approach empowers utilities to strategically adopt digital tools tailored to their operational needs and long-term goals.
1. A first step into the digital world is installing additional instrumentation and implementing advanced control algorithms utilizing the instrumentation. Boschmans, et. al, 2019, documents success with advanced real-time process controls such as ammonia-based aeration control, secondary clarifier guidance, aSRT control, and more at facilities across the nation (Figure 2).
2. Mechanistic digital twins (DTs) provide a digital tool based on a calibrated process model for simulating scenarios, training staff, and planning operational changes in a risk-free environment. The DTs use theoretical scenario inputs, such as reduced units in service, process control adjustments, or new chemical additions, to train staff. These "what-if" scenarios enhance both training effectiveness and strategic planning during plant operation transitions. Two case studies of DTs utilized for training and planning will be reviewed (Figure 3).
3. A secure one-way data from on-prem to the cloud connection enables controlled, outbound-only transfers that reduce cybersecurity risk by preventing inbound connections while still supporting cloud-based analytics.
4. Quality data is essential for informing decisions and optimizing processes, especially as utilities adopt advanced digital solutions and real-time controls. Prioritizing data quality and governance ensures the information driving automation remains accurate, reducing errors and enhancing performance. This paper discusses machine learning (ML) approaches (Figure 4) that could be deployed to perform real-time error and data anomaly detection.
5. Understanding what is coming into plants is crucial in optimization of the treatment process. Accurate prediction of flows and loads enables operators to anticipate imminent conditions and adjust treatment proactively. This foresight helps utilities improve process efficiency, reduce operational costs, and maintain regulatory compliance. The projections can be paired with mechanistic modeling for guidance on units in service and anticipated chemical dosing. Influent flow and load prediction models for a 75 mgd plant will be discussed (Figure 5).
6. DTs can be fed live influent flow and load predictions for real-time process modeling updates. The DT would predict treatment performance and provide recommendations on process setpoints and areas for optimization for operators to consider. DTs can act as pseudo-soft sensors by predicting metrics for parameters not being monitored within the plant.
7. Data-driven models combine subject matter expertise with data science tools to improve our understanding of complex environmental systems like WW treatment and bridge gaps in the current state of mechanistic models. Insights include improved decision making, enhanced efficiency, proactive issue detection, and real-time adaptability. Multiple deployed data-driven models (Figure 6) will be presented as case studies.
8. Integrating digital tools enables holistic optimization across processes and assets within a WW facility. This is accomplished by implementing comprehensive optimization algorithms (Figure 7) that utilize both data-driven and mechanistic models, incorporating information from sensors, SCADA systems, and additional data sources to rank and prioritize operational strategies in real time. Such approaches not only alleviate the workload on operational staff but also enable incremental improvements that can cumulatively result in a substantial return on investment.
9. In the future, treatment plants could be run completely based on instrumentation and digital solutions with minimal human involvement. The autonomous control of a facility would include the previous steps in the digital advancements progression. In pilot/advisory mode (Figure 8), the autonomous control program would make predictions based on input data and advise the operators on decision making. In this case, the operators are the decision makers.
10. For cloud-based ML programs to directly control operations, a secure data connection between on-premise systems and the cloud will be required.
11. When fully deployed, the autonomous control program would completely operate the plant with minimal human intervention. The digital solutions would be the decision makers. Human intervention would occur when there were process upsets or alarms triggered from the programming or digital solutions.
This structured approach empowers utilities to strategically adopt digital tools tailored to their operational needs and long-term goals.
