Integrated multi-attribute transform and seismic driven machine learning technique for geomechanical assessment of Cenozoic reservoirs and seal integrity for carbon storage in the Central Gulf of Mexico

Abstract

Assessing the geomechanical integrity of seals and storage reservoirs is important prior to carbon dioxide (CO₂) storage because it can determine the safety of storage, containment and stability of a proposed storage, and helps minimize the possibility of CO₂ leakage. This study has integrated simultaneous seismic inversion, multi-attribute transform, and a probabilistic neural network, and uses geophysical well logs to evaluate geomechanical parameters for reservoir and seal integrity assessment of Cenozoic strata. Results indicate that candidate reservoir and seal units identified from wells in the study area possesses greater failure strength than the in-situ stresses and are geomechanically stable. However, there is possibility of tensile failure occurring when the injection get to the mature stage and the effective minimum stress crosses the zero effective normal stress line. Each candidate reservoir storage unit has higher rock strength than its overlying shale layer. The thickness of the caprock units is adequately high to provide effective seal and the thickness of the reservoirs are sufficient to support optimal CO₂ storage resources in the study area. The friction angle of Pliocene–Pleistocene strata is adequately high especially in the eastern part of the study area to minimize the risk of fault reactivation and associated deformation. Additional work can be performed to simulate the response of seals, reservoirs, and geomechanical deformation at variable rates and durations of injection.