Using Artificial Intelligence to Optimize Sewer Preventative Maintenance

Using Artificial Intelligence to Optimize Sewer Preventative MaintenanceAbstractIntroduction WSSC Water has 5,700 miles of sewer pipeline. Due to foreign debris such as fats, oils and grease, root intrusion, silt/sediment, and debris, pipelines need to be cleaned periodically to remove these elements which sometimes create blockages and thus avoid Sanitary Sewer Overflows (SSOs). Cleaning a sewer pipeline requires a scouring process, which over time, can degrade wall thickness and reduce pipe life. WSSC Water currently has preventative maintenance (PM) schedules for much of our sewer pipeline system. WSSC Water would like to optimize the sewer PM schedules to save money and to extend the life of our sewer pipelines. Sewer Maintenance Prediction technology can help optimize sewer PM activities by using Artificial Intelligence (AI) to recommend when cleaning is needed even before sewer levels reach an alarm state. The Sewer Maintenance Prediction Pilot tested two technologies, ADS Echo and SmartCover. These technologies can be used for comprehensive sewer performance monitoring, early warning and notification of impending overflows, SSO monitoring, and sewer capacity studies. However, this pilot focused on the technologies' ability to accurately provide early warning and notification of sewer blockages to predict when cleaning is required. Pilot Plan The year-long pilot installed two technologies in 10 manholes each. Each technology was assigned 9 sites located upstream of pipes with 3-month cleaning schedules and 1 site with a high scouring velocity with little to no maintenance issues to be used as a control. After 6 months at the 10 initial locations, some of the equipment was swapped and data was collected at the alternate locations. The swapped locations were those that had the most activity to eliminate any site-specific advantages there may have been. During the year of testing, data was collected to confirm the technologies accurately alarm at full pipe, 1 foot above bench, and 2 feet from top of frame as well as predict when cleaning is needed. The pilot would be considered successful if the technology was able to accurately alarm at pre-defined levels and predict the need for cleaning as verified by WSSC Water crews. Avoided costs associated with reduced cleaning was also considered as part of the evaluation. Pilot Results This presentation will explore the findings of the two technologies and offer practical solutions for wastewater utilities. These findings will be presented in a manner that allows utilities to extrapolate potential cost savings from the optimization of sewer pipeline cleaning using artificial intelligence within their own systems.This paper was presented at the WEF/AWWA Utility Management Conference, February 13-16, 2024.SpeakerTitus, SaraPresentation time13:30:0015:00:00Session time13:30:0015:00:00SessionReal World Applications of Artificial IntelligenceSession number24Session locationOregon Convention Center, Portland, OregonTopicDigital Transformation including AI and ChatGPTTopicDigital Transformation including AI and ChatGPTAuthor(s)Titus, SaraAuthor(s)S. Titus1Author affiliation(s)Washington Suburban Sanitary Commission 1;SourceProceedings of the Water Environment FederationDocument typeConference PaperPublisherWater Environment FederationPrint publication date Feb 2024DOI10.2175/193864718825159278Volume / Issue Content sourceUtility Management ConferenceWord count9