Data-Enabled Predictive Control for Robust Disinfection in Water Distribution Networks

Abstract Submission: Ensuring safe drinking water in urban water distribution networks (WDNs) requires effective disinfection strategies, typically achieved by injecting chlorine at various points in the network. However, uncertainties in water flow dynamics, varying time delays between chlorine injection and measurement, and fluctuating consumer demands present significant challenges in maintaining optimal disinfectant levels. Traditional model-based control approaches often struggle to accommodate these uncertainties and dynamic changes, leading to suboptimal performance and potential health risks. This paper proposes the application of Data-Enabled Predictive Control (DeePC) to address these challenges in WDN disinfection control. DeePC is a data-driven control method that uses historical input-output data to compute the optimal control actions without requiring an explicit model of the system's dynamics. By directly leveraging real-time data, DeePC adapts to changing water flow conditions that cause time-varying input-output time delays, and uncertainties in disinfectant decay rates, ensuring that disinfectant levels remain within safe limits throughout the network. We demonstrate the effectiveness of the DeePC approach through simulations on a WDN with significant flow variations, comparing its performance against conventional Model Predictive Control (MPC) methods that are based on system identification. The results will show that DeePC achieves superior control performance, maintaining desired disinfectant concentrations while effectively handling uncertainties and reducing the risk of harmful byproduct formation, thus contributing to improved public health and safety.

Learning Objectives/Expected Outcome (Optional) : 1. Understand the challenges of maintaining optimal disinfectant levels in water distribution networks (WDNs).
2. Explore the application of Data-Enabled Predictive Control (DeePC) for disinfection control.
3. Evaluate the performance of DeePC compared to conventional control methods used for disinfection control.