Reverse osmosis operation is an energy-intensive process, and the cost of reverse osmosis treatment can be a barrier to its widespread adoption

In most brackish and seawater desalination and water reuse applications, reverse osmosis membrane recovery (and production) is limited by scaling chemistry. The reverse osmosis membrane trains are typically operated in a steady state flow condition optimized for a water recovery setpoint with no ability for real-time adjustment of flow, pressure, and membrane flux to accommodate changing water quality and membrane performance conditions. Design and operating parameters are based on conservative design conditions to maximize water recovery while meeting product quality objectives and avoiding scaling or fouling. Further limiting the ability to build automation into process controls, many reverse osmosis facilities lack the ability to integrate sensor data from disparate sources and evaluate the data quality in real time. Given the conservative approach and manual data analytics and normalization there are opportunities to operate at more efficient settings. Thus, the goal of this project is to use connected sensors and real-time data acquisition coupled with machine learning and the latest in process automation to develop predictive algorithms with automated advisory process controls that can optimize pump energy, membrane flux (i.e., production), reverse osmosis recovery and reverse osmosis feed acid dosing antiscalant dosing to reduce energy, maximize production, and minimize chemical costs while maintaining membrane condition by minimizing fouling and scaling.