Loading...
Diffusion potential of CO₂ into caprock and forward modeling of a CO₂ sequestration site
James, Maureen O.
James, Maureen O.
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2024
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
The increasing rates of global temperatures and climate change has necessitated the adoption of new technologies such as Carbon Capture and Utilization (CCU) and Carbon Capture and Sequestration (CCS). This thesis delves into both approaches across three comprehensive segments, highlighting their significance in the broader context of environmental sustainability and technological advancement. The first part presents an assessment of emissions reduction in CCU operations. Since CCU is presented as a more viable option to reduce emissions, the associated CO2 storage in CCU operations needs high reliability and an ability to rapidly detect and monitor leaks. The second part assesses how to monitor a compromised seal due to gas diffusion from the reservoir and the third part employs the use of machine learning for rapid investigation of gas migration and the detection of potential leaks.
The first part of this study is focused on emissions assessment. CCU and CCS technologies can play a crucial role in mitigating Green House Gas (GHG) emissions. CCU uses CO2 to enhance oil recovery (CO2-EOR) which compensates for the high cost of storing CO2 in the subsurface (CCS) via the increased production of oil (EOR-oil). CO2 used in such operations is obtained from captured or geological sources. CO2-EOR also produces additional CO2 due to usage of the increased oil production (EOR-emissions). Therefore, CO2-EOR needs an assessment of not only the economics but also the emissions generated by the additional oil as well as the complexity to capture these additional emissions. Here, I examine the CO2 budget in CO2-EOR operations by comparing the emissions generated from oil production to the amount of CO2 sequestered. I also investigate the costs associated with emissions mitigation and compare them with profits from additional oil revenue. I find that (a) the CO2 emissions from oil production operations surpass the amount of CO2 stored in the reservoir, (b) the incremental oil produced with CO2–EOR operations further contributes to the overall atmospheric emissions.
The second part of this study studies the seismic assessment of CO2 sorption in shales. CCS is limited by some risks and uncertainties; one of such risks is the issue of CO2 conformance in the reservoir. While leakage through the plume to spill points, migration along faults, fractures and abandoned wells have been studied, leakage via diffusion through the seal remains poorly determined. There are limited data that explore CO2 diffusion mechanisms in CCS studies. For example, it is acknowledged that CO2 will leak through the caprock through diffusion and dissolution, but the mechanism and the time rates of this leakage are poorly documented. Thus, in this part of the thesis, I acquired experimental data to estimate the CO2 diffusion coefficient from the reservoir formation into the shale and determined the effects of CO2 adsorption into the shale formations using ultrasonic measurements. I also analyzed mineralogical interactions in the seal with Scanning Electron Microscopy (SEM) imaging and Energy Dispersive X-ray Spectrometry (EDS).
The third part of this thesis uses machine learning to predict the velocity of CO2 storage formations. Predictive tools are required for critical decision-making for CO2 storage projects and most recently, machine learning tools have found numerous applications in geophysics. To predict long-term changes in the reservoir with CO2 injection, I applied machine learning to petrophysical properties that are critical for optimization of field scale CCS operations. In this part of the thesis, I developed models to predict the velocity of the reservoir formations. The models Random Forest Regression (RF), Multi-feed forward neural networks (MFNN) and Long Short-Term Memory (LSTM) were tested by combining all data points from the wells and by segregating only the reservoir zones. The Random Forest model performed better than the other two models, and the model development by reservoir zones reduces mismatches and allows for a more accurate prediction of the properties being trained.
All three aspects of this study; (a) the assessment of the carbon neutrality of CO2-EOR; (b) experimental studies to determine the diffusion of CO2 into the seal and the viability of the seal for CO2 storage and; (c) machine learning for the prediction of rock properties, are critical to ensure that CCU and CCS operations are safe, environmentally sustainable, and contribute towards
emissions reduction.
Associated Publications
Rights
Copyright of the original work is retained by the author.