Methodologies to detect leakages from geological carbon storage sites

Abstract

Geological carbon storage (GCS) has been proposed as a favorable technology to reduce carbon dioxide (CO2) emissions to the atmosphere. Candidate storage formations include abandoned oil and natural gas reservoirs, un-mineable coal seams, and deep saline aquifers. The large global storage capacity and widespread occurrence of deep saline formations make them ideal repositories of large volumes of CO2, however they generally lack of data for geological characterization in comparison to oil and gas reservoirs. Thus, properties of the injected formation or the sealing formation are unknown, which implies that the evolution and movement of the CO2 plume are uncertain in these geological formations. The first part of this research aims to provide an understanding of the main sources of uncertainty during the injection of CO2 that cause leakage variability and fluid pressure change near the injection well, which could be responsible for fracturing the sealing formation. With this purpose the effect of uncertain parameters such as permeability and porosity of injected aquifer, permeability of CO2 leakage pathways through the sealing layers, system compressibility, and brine residual saturation are investigated using stochastic and global sensitivity analyses. These analyses are applied to a potential candidate site for GCS located at the Michigan Basin. Results show aquifer permeability and system compressibility are the most influential parameters on fluid overpressure and CO2 mass leakage. Other parameters, such as rock porosity, permeability of passive wells, and brine residual saturation do not influence fluid overpressure nearby the injection well. CO2 mass leakage is found to be sensitive to passive well permeability as well as the type of statistical distribution applied to describe well permeability. Scarce data of the Michigan Basin exist that can be used directly to describe the spatial distribution at the basin scale of the caprock overlying the candidate site. The continuity of this formation is uncertain. The second part of this investigation explores the application of binary permeability fields for the study of CO2 leakage from GCS at the candidate site. A sequential indicator simulation algorithm is used to populate binary permeability fields representing a caprock formation with potential leaky areas (or inclusions). Results of the caprock continuity uncertainty conclude that increasing the probability of inclusions occurrence increases the CO2 leakage. In addition, the correlation length used by the sequential indicator simulator affects the occurrence of inclusions. The third part investigates the detection and location of the presence of possible brine or carbon leakage pathways at the caprock during the injection operations of a GCS system. A computational framework for the assimilation of changes in head pressure data into a subsurface flow model is created to study the evolution of the CO2 plume and brine movement. The capabilities of two data assimilation algorithms, the ensemble smoother (ES) and the ensemble Kalman smoother (EnKS), to identify and locate the leaky pathways are examined. Results suggest that the EnKS is more effective than the ES in characterizing caprock discontinuities.