Improving De-Oiling Efficiency in Offshore Produced Water Treatment Using Deep Learning and NMPC-Based Control

Offshore oil production demands stricter produced water treatment (PWT) systems, with Danish regulations requiring real-time monitoring of oil-in-water (OiW) concentrations. While this offers opportunities to optimize de-oiling, OiW data are rarely used in PWT control due to the absence of models capturing hydrocyclone separation dynamics. To bridge this gap, this study combines predictive deep learning with Nonlinear Model Predictive Control (NMPC) to develop an OiW-based closed-loop framework. Using OiW concentration as feedback, the system enhances de-oiling efficiency, stabilizes separation, and reduces the environmental impact of produced water (PW) discharge. Simulation results demonstrate improved performance, showing the framework’s potential to optimize operations and support compliance with evolving regulations.