PI: Farrah Moazeni

Co-PI(s): N/A

University: Lehigh University

Industry partner: Martz Technologies, VZRscada

This research aims to develop a framework for nonlinear economic model predictive control (NEMPC) of water distribution systems (WDSs) to facilitate online control and model adaptation. The proposed NEMPC will minimize costs and energy consumption, as well as ensure the smooth and optimal operation of modern water systems in real-time. The optimal operation of WDSs cannot be established a priori by fixing reference volumes in the tanks, making NEMPC a suitable control strategy. The proposed bi-level NEMPC strategy will optimize the scheduling of the tanks based on the electricity price fluctuation and the hourly water demand (upper-level). Then, the pump scheduling approach with a one-minute sampling time will be optimized at the lower level. We are partnering with Martz Technologies, a leader in developing artificial intelligence (AI) and optimization algorithms for smart water systems, to deliver the objectives of this research. Together, we will develop a novel NEMPC framework that improves WDS’s performance and is computationally efficient and robust to uncertainty, making it suitable for online adjustment and execution in response to sudden changes in water systems. This PITA award will contribute to modeling, analytics, and control of water systems, motivated by an emerging critical societal problem, which is the need for energy-efficient, resilient, and secure water infrastructure systems. This project will advance the computational and theoretical frontiers of optimization and control laws in smart water systems. The proposed approaches are also applicable to other uncertain cyber-physical infrastructures, such as power systems, smart buildings, or transportation systems. Together with industry involvement, this project will yield new insights into the modeling and operation of emerging automated and controllable infrastructures that are highly susceptible to changes and uncertainty. The findings will showcase the importance of innovative combinations of data analytics, optimization, physical models, and control techniques in solving complex engineering problems in critical infrastructure systems.