The innovation will enable real-time decision support for large-scale water infrastructure systems, through the fusion of operational data and infrastructure-aware predictive models. These real-time software instruments will leverage significant investments by drinking water utilities in Supervisory Control and Data Acquisition (SCADA) systems that support operational decisions and Geographic Information Systems and infrastructure models that support infrastructure planning; these investments should and can be leveraged to support a wider scope of utility decision making. This SBIR Phase I proposal focuses on the commercial technology associated with real-time predictions of water quality evolution within water supply systems. Previous field scale evaluations of network water quality models have suffered from idealized reaction kinetics; short data collection periods; and significant parameter uncertainty and variability. By fusing operational data and infrastructure-aware predictive models, this study will, for the first time, develop a rigorous assessment of the fidelity of complex network water quality models. The data fusion software instruments inspired by this research will support the transparent integration of operational decisions with real-time data and model predictions and forecasts, leading to enhanced water quality benefits that, at present, can only be anticipated.
The commercial impact and practical benefits of fusing real-time operational data with infrastructure-aware predictive models will be enabled by the ability to simply and accurately forecast distribution system hydraulics and water quality, in real-time. This technology will allow operators to routinely engage in situational response training, and conduct operational analyses to achieve practical water quality management goals ? such as maintenance of minimum chlorine residuals or control of disinfection byproducts. Engineers will apply their infrastructure knowledge to these tasks in a collaborative fashion, while knowing their infrastructure models are continuously updated through a persistent interpretation of the operational record. Managers will review dashboards and automated reports showing trends in unaccounted for water, energy usage, and water quality, and integrate those with past and future asset management decisions. Successful completion of Phases I and II of this project will test the value proposition for real-time data fusion for water supply systems, specifically for real-time network water quality prediction and forecasts. Real-time data fusion benefits and associated workflows will be put to the test in actual operating environments, supported by a new data fusion system that allows for efficient utility workflows across the organization.
Water utilities have invested heavily in data that are essential for the design and operation of large-scale urban water supply systems. Supervisory Control and Data Acquisition (SCADA) systems support operational decisions, and Geographic Information Systems (GIS) and physics-based predictive models support infrastructure planning. Yet these information investments are not currently used to support a wider scope of utility decision making. Years of SCADA operational data sit in databases, divorced from the data that are essential for a higher-level understanding of infrastructure behavior. The real-time modeling and data processing innovations pursued here are central to knowledge discovery and visualization processes that will help water utilities throughout the world respond to pressures requiring strategic asset management and operational decision making. The merging of real-time operational data with predictive models will yield numerous practical benefits, enabled by the ability to simply and accurately forecast distribution system hydraulics and water quality, in real-time: Operators will routinely engage in situational response training, and conduct operational analyses to achieve optimization goals related to pressure, leakage, energy, and water quality management. Engineers will apply their infrastructure knowledge to these same tasks in a collaborative fashion, while knowing their infrastructure models are continuously updated through a persistent interpretation of the operational record – enabling automatic estimation of water usage, operating rules, and pump head-discharge curves. Managers will review automated periodic reports showing trends in unaccounted for water, energy usage, and water quality, and integrate those with past and future asset management decisions. Predictive models with established accuracy will allow utilities to move confidently into areas of model-supported operations and control that would be considered infeasible today. Such benefits are not unrealistic, and in fact are already supported by the existing investments in data, and by technical theories that are hundreds of years old. What has been missing is a modern software technology that connects the dots between data and infrastructure model; a well conceived delivery platform for that technology that supports end user needs; and a clear demonstration of performance and benefits from the fusion of infrastructure and operational data. During this project we developed improved methods of data processing that enhanced the accuracy of hydraulic and water quality forecasts, and measured, for the first time, the levels of error in these forecasts. Only by knowing this information about accuracy, can water utility operators, engineering, and managers learn to trust simulated values enough to base operational decisions on them -- much the same way that now individuals routinely trust short term weather forecasts enough to plan an outdoor event. We also enhanced the reliability and robustness and scalability of our data processing framework, which will translate into improved end-user tools to help water utilities better manage their resources such as energy, and help ensure the highest water quality. By working side by side with water utility personnel, we came to understand the functionality and benefits associated with preferred end user tools that utilities would be eager to use. Finally, we developed new technology strategies for advanced water quality predictions that will allow model forecasts to better adapt to the constantly changing source water conditions, due to, e.g., rainfall and temperature fluctuations.
