Physics-Coupled Machine Learning Toolset for Geological Carbon Storage Evaluation and Performance Analysis

A comprehensive toolset that can provide fast and accurate design, survey, planning, monitoring, and evaluation of behaviors and responses in the reservoir field is essential to achieve successful geological carbon capture and storage (CCS) development and operations, with or without enhanced hydrocarbon recovery. Based on the physics of material balance between injection and extraction, the Capacitance Resistance Model (CRM) method can perform rapid history matching (HM), forecasting, and optimizations in operational scale. Such capabilities provide key operational guidance to users with insights of an individual well regarding its injection/extraction and bottom hole pressure (BHP), as well as inter-well connectivity of multiple wells in the field along with its flexible time-window capability for operation planning and development. Moreover, advanced artificial intelligence (AI)/machine learning (ML) models developed for the virtual learning environment (VLE) are also coupled with the workflow to provide detailed three-dimensional reservoir field responses that are essential to the geological CCS monitoring and evaluation of the optimal reservoir management and risk reduction. The proposed approach with physics-informed ML demonstrates the value for emerging “SMART” field operations and reservoir management with three to four orders of magnitude speed-up in computational time in a real-time and near real-time fashion