Skip to main content

Artificial Intelligence (AI) and Machine Learning (ML) for Potable Reuse

Las Virgenes Municipal Water District
Calabasas, California
advanced treatment facility

Project Overview

From aeration to ultrafiltration – a holistic approach to evaluating machine learning for potable reuse optimization

The combination of population growth and unpredictable drought is straining reliable water supplies in the Western United States. Because of the lower relative cost compared to other (imported) water supplies, potable water reuse is gaining acceptance as the preferred solution in many instances and is projected to expand rapidly over the next decade.

However, moving from non-potable recycled water to purified recycled water more than doubles energy use and requires highly-trained operations staff, which also strains already over-taxed resources. The direct potential risk to human health requires a myriad of sensors and control systems to provide detailed process monitoring to maintain consistently safe water supply.

In light of these challenges, an international team of researchers has been evaluating the promise of artificial intelligence (AI) and machine learning (ML) to reduce energy and chemical use, supplement operations and maintenance, and provide greater confidence in water quality.

Over the last two years, the Las Virgenes–Triunfo Joint Powers Authority has been evaluating AI/ML at their Tapia Water Reclamation Facility (WRF) and their pure water demonstration facility (Pure Water Demo). Project partners include the Yokogawa Electric Corporation (Japan), the Japanese Ministry of Economy Trade and Industry (METI), the National Water Research Institute, Carollo Engineers, the Metropolitan Water District, the United States Bureau of Reclamation, and IOSight (Israel).

Extensive success has been documented, with full-scale trials initially demonstrating at least a 10% reduction in energy use at the Tapia WRF and demonstration-scale trials predicting membrane fouling at the Pure Water Demo. In addition, during blind testing, the AI models predicted the future monthly trend of trans-membrane pressure (TMP) with a very high degree of accuracy. By accurately forecasting TMP, operations staff will be able to make better decisions on when to clean the membrane systems – optimizing not only energy consumption, but also chemical usage and operations staff time.

Results and Highlights

Developed energy and chemical savings strategies while remaining focused on the fundamental goal of maintaining safety of potable reuse supply

Advanced the science of machine learning, specifically focused on potable reuse

Developed data transfer and dashboard interface tools that can be adapted on future project

Funded by numerous grants from various funding agencies, demonstrating broad support for advancing this research topic

Project Awards and Accolades

2022 Transformational Innovation Award

WateReuse Association

Have an upcoming project? Let's make it a success.