(169m) Using ML to Determine the Optimal Set of Operational Parameters of the RO System in a Regional Water Treatment Plant

Operating RO systems presents challenges that involve handling a large amount of data, monitoring various variables, and conducting statistical analysis before sending water for treatment. Operators must monitor feed water characteristics such as conductivity, oxidation reduction potential (ORP), total suspended solids (TSS), turbidity, and chemical oxygen demand (COD), as well as the operational parameters of RO membrane feed such as feed pressure, flow rate, and temperature, as the most impactful factors affecting RO membrane performance to optimize the operational parameters like salt passage, permeate flow rate, and pressure difference across membranes. Data from individual RO membrane units must be collected and normalized, providing criteria for choosing the best treatment methods and minimizing costs.

Currently, many facilities still rely on manual SCADA data collection and infrequent review of maintenance logs, a process that can take up to weeks. Furthermore, water varies in hybrid plants such as the one in this study, making it more difficult to select the best treatment method for an effective RO facility's operation and maintenance strategy. ML models can be employed to better monitor data and predict the optimal set of operational parameters, maximizing membrane performance in an RO system and enabling rapid response and mitigation and safety efforts.

This study aims to develop a predictive model capable of assessing the impact of feed characteristics (including conductivity, ORP, TSS, turbidity, and COD) and operational parameters of RO membrane feed (such as feed pressure, flow rate, and temperature) on the performance of RO membranes, focusing on parameters like salt passage, permeate flow rate, and pressure difference across membranes. The performances of different machine learning methods will be compared, and the model with the highest accuracy will be used to predict the best treatment method by adjusting water quality and determining the most influential factors on RO membrane performance. Subsequently, an optimization model will be developed to determine the optimal set of operational parameters for the RO system, aiming to maximize membrane performance.