Artificial Intelligence and Machine Learning (AI/ML) has significant potential to optimize wastewater treatment processes, improve forecasting, detect leaks, expedite alerts about pollutants, and address other long-standing challenges in the industry. However, adoption of  AI/ML by wastewater systems in the United States is lower than other industries with the use of AI currently standing at 10-15%, mainly by larger utilities. 

How can wastewater utilities start taking advantage of smart technology to optimize their operations, improve demand forecasting, and extend the life of their assets? Here are some tips to consider.  

Quality Data In, Smart Decisions Out 

Models are only as good as the data on which they’re trained. High quality, continuous data, such as real-time sensor data, historical SCADA data, weather patterns, and population growth trends, is necessary for effective model training since trustworthy results can only come from a model’s exposure to a high volume of data.  

Not only are large volumes of data required for model training but the data must be carefully labeled and organized before it is used. Utilities should invest in data management best practices to ensure that they have the systems in place to collect, store, and process the extensive data that is needed.  

Build Trust Through Transparency 

Complex AI solutions can be hard to understand with results that can’t be easily interpreted or verified. This lack of transparency impacts trust and may lead to second guessing. Investing in Explainable AI (XAI) helps bridge that gap by providing clear explanations for predictions and decisions. DARROW, an ambitious, multi-year European Union project to drive transformation in the wastewater industry, recognized something important early on: AI needs to be trustworthy and understandable in order to be successful. DARROW’s AI, designed to optimize how resources are recovered from wastewater, provides users with direct visibility to insights, including visualization tools for decision trees as well as clarifications for recommendations. 

It's About The People 

Resistance to change is part of human nature. Operators and technicians may be worried that their jobs are at risk or that their hands-on experience will be devalued with AI/ML. It’s important to emphasize that AI’s value lies in being used as a tool to assist people in their everyday operations, not as a fully independent system with no human oversight.  

Start Small 

Get the tools in place for a specific use case before investing in a full roll-out. This approach will make it easier to test feasibility, identify data needs/gaps, and understand operational impacts. Small wins add up, and success will build confidence and buy-in for bigger changes.  

To help you on your journey, BRYX offers easy access to pretrained machine learning models that are specifically designed for the wastewater industry. Our Sewer Pipeline Defect Detection model detects and classifies structural and operational defects in CCTV inspection video, using NASSCO standards. And our Smart Sewer Repair model generates ranked recommendations and cost estimates for repairs such as pipe replacement, manhole to manhole, light cleaning, root treatment, sectional liners, test and seal, and more. Both models come with free trials with no commitment required. Start yours today at BRYX!