Combining AI and EM: A Dual Approach for Dynamic Waterflood Monitoring and Geological Modelling

Abstract Waterflood monitoring is essential for managing reservoirs to maximize recovery and efficiency, yet current electromagnetic (EM) methods, such as cross-well or borehole-to-surface surveys are often too complex and costly to use at scale. In this work, we explore two techniques designed to optimize EM development by integrating sensing directly into existing well completions. One approach uses insulated well liners to track the flood front as it moves through the reservoir, while the other leverages production logging tools as temporary EM sources to capture saturation through induced polarization effects. Alongside these hardware innovations, we developed an AI-driven workflow that interprets EM data in real time. We trained a deep learning system on simulated EM responses using a controlled reservoir model, which allowed it to precisely map water saturation and flow paths between wells. Together, these developments suggest that EM monitoring will become simpler, affordable, and strongly associated with accurate data interpretation. The integration of AI-enabled analysis and completion-based acquisition techniques provide a strategic approach toward an advanced, more agile waterflood surveillance and reservoir management.