Asset Management, Artificial Intelligence, & Connected Devices: Optimizing Collection System O&M

Asset Management, Artificial Intelligence, & Connected Devices: Optimizing Collection System O&MAbstractToday's business environment is changing faster than ever. This is especially true for water and wastewater utilities. The balancing act between affordability and level of service means that these organizations are constantly asked to do more with less. New innovations such as machine learning (ML) and connected technologies are critical tools that utilities can use to answer this call. These technologies and the data they produce can supercharge operations and maintenance programs to maximize the value of water and wastewater infrastructure for utilities and the communities they serve. Winston-Salem/Forsyth County (WSFC) Utilities is a large utility serving over 380,000 residential, business, and industrial customers in North Carolina. They recently combined both ML technology and -connected devices with the O&M components of their existing asset management program to make a huge gain in the level of service of their wastewater collection system. Using these concepts, the utility reduced collection system regulatory events by over 25% in one year. This was achieved with no additional staff, only a minor change in the volume of O&M activities completed, and no increase in the amount of capital investment within the system. WSFC Utilities' approach included three key actions to achieve these results: 1.Incorporate an ML model into O&M decision-making processes to help detect high-risk assets with a limited historical record. 2.Leverage connected devices to monitor high-risk assets and calibrate the unique maintenance strategies for these individual assets. 3.Develop a web-based O&M decision support application to direct O&M activities in accordance with these unique maintenance strategies for each individual asset. The ML model developed by WSFC Utilities combines asset attributes, environmental factors, and collection system network factors to prioritize individual pipes for maintenance. The model predicts a probability that a given pipe will be dirty. This probability is then used to identify assets with no or limited history for maintenance. As maintenance is performed on assets with a known need for intervention, nearby high-probability pipes with limited data are packaged into these same work orders to build the efficiency of crews while establishing a maintenance observation on these unknown assets. The predictive model, and the unique maintenance strategies for the individual asset, are then continuously updated to improve the performance of both the ML-model and collection system maintenance strategy. Connected devices have also been a key component of WSFC Utilities reduction in collection system regulatory events. The utility has deployed ten level sensors within the collection system to monitor high-risk assets and calibrate their unique maintenance strategies for these individual assets. For assets with the most intensive O&M programs, WSFC Utilities deploys these sensors and temporarily suspends maintenance activity for 6 months. These connected devices record system performance in real time. If an asset shows leading indicators of a regulatory event a trend of level increase within the wastewater flow an O&M event is scheduled for the asset. If there is no indicator of a need for maintenance, these activities are deferred. Upon the end of the 6-month monitoring period, the maintenance strategies for the individual asset are recalibrated. This strategy allows O&M activities on assets with the most intense programs to be deferred and lowers the overall O&M requirements for the WSFC Utilities collection system. To date, this has resulted in the deferral of over 10 miles of maintenance since program inception. WSFC Utilities updated its legacy decision-making support tool to a web-based application called FreeFlowH2O (FFH2O). The O&M decision-making is a process informed by WSFC Utilities' AI model and connected devices, along with the full history of maintenance activity performed by the utility. Leveraging utility data such as GIS-based asset information, work order histories, maintenance observations, and condition assessment findings, FFH2O creates and manages a pipe cleaning schedule to promote right-time cleaning. By actively managing the cleaning schedule within the collection system, FFH20 allows WSFC Utilities to focus cleaning resources on the highest-risk pipes and limit unnecessary maintenance on lower-risk pipes. This results in cleaning activities taking place in pipes that have grease, roots, and debris that need to be removed and avoids unnecessary cleaning on pipes that are already clean. These changes have allowed WSFC Utilities to make a significant step forward in their O&M decision-making processes. SSO events decreased from 74 in FY21 to 55 in FY22 a 25.6% decrease. Prior to the implementation of these changes, WSFC Utilities' progress in reducing SSO levels to those comparable with best-in-class utilities had slowed. These changes represent a significant improvement achieved without significant changes to investment capital rehabilitation, an increase in the volume of O&M activities performed, and no additional staff. Utilities can capitalize on new technological innovations to improve existing asset management practices. This is an essential tool to address the challenges associated with new regulatory requirements, staffing challenges, and budgetary constraints that continue to impact our industry.This paper was presented at the WEF/AWWA Utility Management Conference, February 13-16, 2024.SpeakerMueller, JacobPresentation time09:00:0009:30:00Session time08:30:0010:00:00SessionOptimization of Operation & MaintenanceSession number32Session locationOregon Convention Center, Portland, OregonTopicOptimization of Operation & MaintenanceTopicOptimization of Operation & MaintenanceAuthor(s)Mueller, JacobAuthor(s)J. Mueller1Author affiliation(s)HDR 1;SourceProceedings of the Water Environment FederationDocument typeConference PaperPublisherWater Environment FederationPrint publication date Feb 2024DOI10.2175/193864718825159248Volume / Issue Content sourceUtility Management ConferenceWord count12