This project will optimize and integrate smart water infrastructure monitoring, sensing, and metering data with automated hydraulic controls to establish a system-level control framework for water utility infrastructure management and operation. This smart water infrastructure monitoring and control system (SWIMCS) will be capable of: (1) quantitatively assessing the vulnerability and resiliency of a Department of Defense (DoD) Water Distribution Network (WDN); (2) Quantify the lower flow limits and overall water demand requirements for maintaining water quality (WQ); (3) Optimize the placement of smart water infrastructure, including WQ sensors, chlorine booster stations, and hydraulic controls such as directional flow control valves or automated flushing devices; (4) quantify the improvement in water resiliency for the optimized WDN with different levels of smart water infrastructure implementation; (5) quantify the impacts of low flows on WDN WQ in various water resiliency and vulnerability scenarios; and (6) enable real-time monitoring of a WDN including leak detection, threat detection, and early detection of contamination episodes. A full SWIMCS will be demonstrated on an existing DoD WDN to showcase the integration of the advanced algorithms and smart water equipment, and to establish practical guidelines for future implementation of the framework in other DoD WDNs.

Smart Water Equipment: A suit of commercially available smart WDN monitoring equipment will be systematically evaluated, including clamp-on ultrasonic flow meters, online WQ and pressure sensors, and automated flushing devices installed on fire hydrants. Equipment that requires no or minimal modifications to existing WDN piping and disruptions to water utility services will be prioritized for evaluation and implementation.

Deep-learning neural networks and numerical physics-based models will be integrated to model spatiotemporal variations in WQ throughout the WDN. This computational method will be used to baseline the existing WDN, identify the lower consumption limits needed to maintain WQ throughout the entire WDN, and enable predictive capabilities needed for facility managers to assess the impacts of water efficiency measures on WQ and waste, therefore inform decisions on water efficiency investment. Deep learning will be used to build surrogate data-driven models that can imitate the complexity of the process-based models for disinfectant decay and disinfection byproduct formation at significantly lower computational cost.

Bayesian Optimization: Novel Bayesian optimization algorithms will be developed to optimize: (1) the placement of WQ sensors to enable network-wide observability at minimal cost, (2) the location of chlorine booster systems to maintain optimal residual disinfectant throughout the entire network with minimal investment, and (3) the placement and scheduling of autonomous flushing stations (for use in irrigation) and directional flow control valves to optimize WDN operation and management while improving resiliency.

The suit of technologies will (1) enable facility managers and decision makers to anticipate future threats to water resiliency and operational security posed by deficient design of critical potable water infrastructure; (2) reduce the cost of WDN improvements such as water quality sensors, chlorine booster stations, and hydraulic controls to reduce life cycle costs and maximize return on investment; (3) increase water resiliency and reduce vulnerability by providing WDN operators with the information and insights needed to optimize daily WDN management strategies and operation; (4) enable rapid response to and mitigation of system failures, threats, and contamination events; (5) fully automate daily operations of WDN to maximize system efficiency and resiliency while reducing operation and maintenance costs; and (6) provide facility managers with the data, information, and insights needed to develop strong business cases for water resiliency upgrades.