Insights Into CO2 Storage Using the FluidFlower

Abstract

Carbon capture and storage (CCS) is essential for mitigating climate change. Geological sequestration stores CO2 in the subsurface, however, geological uncertainties and risks, especially leakage, present significant challenges. This thesis explores CO2 flow dynamics using the FluidFlower experimental rig and seismic modelling to understand the life cycle of a typical CO2 sequestration project from initial injection to long-term storage. Seismic modelling of different stages in the FluidFlower experiment provides insights into how the evolution of a CO2 sequestration project can be effectively monitored and understood in subsurface datasets.

High-resolution time-lapse images from the FluidFlower were analysed to study CO2 migration in a model based on the geology of the Horda Platform, offshore Norway, featuring tilted fault blocks bounded by normal faults and a stratigraphy comprising homogeneous and heterogeneous reservoirs separated by seals. The results of the FluidFlower experiment reveal several stages in underground storage of CO2 that can be summarised as 1) The initial injection stage, characterised by the injection of CO2 and upwards rising plume around the injector. 2) The trap-filling stage, characterised by lateral migration of CO2 along internal barriers or top seals. 3) The long-term storage stage, characterised by CO2 dissolving into solution and forming “fingers” because of density-driven convective mixing.

The time-lapse FluidFlower images were converted into 2D seismic models using a point-spread function convolution approach. Seismic modelling revealed that CO2 significantly impacts seismic responses, and even relatively minor changes in fluid saturation as low as 5% CO2 provide distinct reflections around the migrating CO2 plume. Seismic modelling also indicated which features and events during CO2 injection and storage could be resolved. Prominent features (e.g., CO2 plume and seal breaches) were resolved, while smaller, more detailed features, like fingering, could not be confidently resolved but were still detectable through changes in seismic facies.

Combining the FluidFlower experiments with seismic modelling advances our understanding of subsurface feature detection and has the potential to improve CCS evaluation and safety. Future work should involve more complex heterogeneous models, better simulation materials, and scaled scenarios. Improved scenario planning with relevant geological analogues is recommended to strengthen the coordination between physical and seismic models, advancing CO2 monitoring.