The suitability of water for human consumption depends on the quality and quantity of the resource, both of which can change due to many variables such as precipitation, climate, contamination caused by natural or human activities, and changes in seasonal demands and local/regional population. The first step to ensuring high quality water is to have a variety of instruments that can measure and record the concentrations of contaminants/constituents in water (both source and treated) and the operating parameters of water treatment systems (flows, pressures, levels, temperatures, etc.). The second step involves the understanding, processing, interpreting, and using the data that was collected in the first step to improve the performance of the water treatment processes, increasing the efficiency and sustainability of the systems that provide us water. This ultimately protects both humans and the environment by proper operation, monitoring, and maintenance of the water treatment systems.

Over the 5 years of the project, we worked with wastewater treatment plants (WWTP), from small demonstration-scale to full-scale urban/centralized systems, to enable early identification of process faults that can be easily resolved instead of allowing small problems to disrupt the entire system. We developed methods to increase the efficiency of processes to avoid faults and malfunctioning of processes. First, we focused on decentralized WWTPs. They are usually not attended by operators 24/7, and their normalized maintenance cost is higher; thus, it is more important to prevent failures and alert operators to significant changes in the system. Using multivariable analysis, we have developed data-driven tools that detect abnormal behavior of systems and alert to potential failures before they occur. We have made major methodological advances in fault isolation, which allows operators to see not only that a fault is occurring but also to identify the features that are impacted by the fault, facilitating fault diagnosis. Next, we developed a hybrid statistical and machine learning model to forecast the performance of an urban WWTP, and specifically adjusted operating conditions in advanced to minimize energy demand and infrastructure degradation. Incorporating our tools into the commercial supervisory control and data acquisition system of the plant, we were able to adjust system operations to achieve most efficient performance 50 minutes ahead of the point of measurement. In our next case study, we successfully combined our tools with artificial neural networks to forecast disinfection addition to the effluent of a large WWTP under varying conditions to achieve the desired clean water before discharge to the environment.

As a result of this funding and with further support from NSF, we have developed an introduction to data science class that has now served nearly 200 undergraduate students. A five-week summer undergraduate research program that focuses on developing data science solutions to data-rich problems from our water and wastewater treatment partners has also been established. So far, nearly fifty students have participated in this program. Finally, a data science workshop has been developed specifically for professionals in the water and wastewater treatment fields that provides training in visualization, analysis, and modeling of data.

One of the greatest achievements of this project is the infrastructure that we have developed to support future research in the field. We constructed a mobile pilot laboratory to demonstrate direct potable reuse (DPR) of reclaimed water under a broad range of conditions and for communities of various sizes. We teamed up with an industrial partner, Colorado Spring Utilities, to design and build a system equipped with numerous water quality sensors and an advanced control system. We have started developing model predictive controls and early fault detection systems that will make this mobile laboratory a critical academic asset for future generation of students, scientists, and engineers.

Over the summer and fall of 2021, the DPR system has trnsformed 500,000 gallons of reclaimed water into drinkable water. Many public outreach events have been carried out around the mobile laboratory, including 49 tours (945 attendees, 85% tasted the purified water), 3 community event (760 soda tastings with soda made of the reclaimed potable water), 4 public presentations (101 attendees), 11 school/college tours, and many media events, stories, and videos created. The public will continue to play a critical role in the adoption of new, unconventional sources of drinking water. Working with utilities, communities, and regulators, we will continue to support academic research, workforce development, and protection of the environment through integration of data science and engineering.

We acknowledge the critical support from NSF, the Colorado Water Conservation Board, Colorado Springs Utilities, Denver Metro Water Recovery, the Boulder Water Resource Recovery Facility, Aqua Aerobic Systems/METAWATER, and the Zoma Foundation for this research. Moving forward, the research infrastructure and training developed under this research will be supported by grants from the US Department of Energy (2022-2024), the NSF Data Science Corps program (Grant #1924146), and potentially other local, state, and federal agencies.

Last Modified: 01/30/2022
Modified by: Tzahi Y Cath