Robust Prediction of Residual Chlorine Decay in a\nDrinking Water Distribution System Integrating Water Quality Sensing\nand Predictive Tools

This\nresearch paper presents a novel approach for predicting residual\nchlorine in water distribution systems, a critical component for ensuring\nsafe drinking water quality. This new method innovatively incorporates\nthe variability of source Natural Organic Matter (NOM), operational\nactions, and water demands. It combines water quality sensing with\ndata-integrated hydraulic and chemical reaction models, aiming to\noffer high-fidelity and robust water quality predictions and, thus,\nprovide operational decision support for optimized distribution system\nmanagement. Until now, the absence of real-time data and appropriate\nalgorithms to quantify and analyze source quality characteristics\nhas inhibited the development of a robust water quality predictor.\nThis research fills this gap and enables direct predictions of Total\nResidual Chlorine (TRC) instead of using water age as a surrogate.\nBy integrating kinetics modeling approaches and tracking NOM variations\nin the network as well as implementing the predictive tools, it is\nnow possible to anticipate TRC decay rates and manage water quality\nmore effectively. Three distinct models were evaluated in this study:\ntwo pure black-box models leveraging Artificial Neural Networks (ANNs)\nand the Gaussian Process Regression (GPR) algorithm and a hybrid model\ncombining a first-order kinetic model with a GPR algorithm. The results\nshowed significant improvement in MSE values, up to 90%, in prediction\naccuracy for TRC in the distribution network compared to traditional\nkinetic models.