The continuous rise of CO₂ concentration in the atmosphere has a series of negative impacts on the Earth's climate and environment, such as rising temperatures, ocean acidification, forest fires, and increased frequency of extreme weather events. Therefore, as a bottom-line method to reduce CO₂ content, carbon sequestration technology will play an increasingly important role in mitigating climate change. During the process of CO₂ injection in carbon storage, the migration behavior and specific state of CO₂ are closely related to formation pressure. Obtaining formation pressure facilitates better reservoir prediction and flow monitoring, while also providing information for formation stability (e.g., calculating geomechanical parameters such as in-situ stress). Considering the characteristics of shallow offshore reservoir (typically loose in texture and mostly unconsolidated), a rock physics model is proposed to calculate the formation pressure more accurately. This rock physics model considered multiple influencing factors, including various mineral compositions, fluid distribution condition and formation pressure. Through the use of a field data in China, the forward modeling results generated by the rock physics model are used to form the training dataset for Deep Belief Network (DBN). Then the formation pressure is calculated by the trained DBN. The proposed approach yields a notably small error. This method enables the rapid calculation method of formation pressure, providing indispensable information and a reliable basis for late-stage site selection, fluid flow monitoring, and safety assessment for CO₂ storage.

Natural gas hydrates; CO2-CH4 replacement; CO2 sequestration; continuous CO2 injection; dual-well injection-production