Author ORCID Identifier
Semester
Summer
Date of Graduation
2024
Document Type
Thesis
Degree Type
MS
College
Statler College of Engineering and Mineral Resources
Department
Petroleum and Natural Gas Engineering
Committee Chair
Ebrahim Fathi
Committee Co-Chair
Samuel Ameri
Committee Member
Samuel Ameri
Committee Member
Kashy Aminian
Abstract
This thesis delves into the utilization of machine learning methodologies to quantify the spatial extent of CO₂ plumes by leveraging microseismic data obtained from the Illinois Basin Decatur Project (IBDP) site spanning November 2011 to June 2018. This initiative, focused on the geological sequestration of carbon dioxide, furnishes a unique and comprehensive dataset comprising well logs, microseismic activity records, and CO₂ injection metrics, all crucial for quantifying the subsurface CO₂ saturation plume dynamics. The primary objective is to forecast the temporal evolution of CO₂ saturation plumes in the subsurface, a critical undertaking for ensuring both the environmental integrity and operational efficacy of CO₂ sequestration activities.
The findings reveal that the application of machine learning for interpreting microseismic data can forecast plume behavior exhibiting vertical clustering within a confined range of distances from the injection well, indicative of periodic migration and following an invasion percolation model. The buoyant CO₂ plume is partly trapped within the sandstone intervals periodically breaching discrete barriers or baffles. This observation aligns with earlier investigations that uncovered the presence of cemented or shale-rich intra-formational baffles. These intervals act as leaky seals impeding the vertical migration of injected CO₂ into the Mt. Simon sandstone, confining it within thin, highly saturated layers until buoyancy overcomes gravity and capillary forces, leading to periodic breakthroughs along vertical zones of weakness. By employing clustering algorithms such as K-Means and DBSCAN, we were able to identify patterns and trends in the seismic data that would be challenging to detect through traditional methods. These machine learning models allowed for a more precise quantification of CO₂ plume expansion, both vertically and horizontally. The results indicated that the CO₂ plume primarily expands vertically within the Mt. Simon B and C formations, with significant vertical migration observed during the injection phase in the order of several hundred feet. Horizontal migration, while less pronounced, was still notable and provided valuable insights into the lateral spread of the CO₂ plume.
This capability of application of machine learning for quantifying the extension of CO₂ saturation plume holds immense significance for real-time monitoring and management of CO₂ sequestration sites. The models demonstrate commendable accuracy in analyzing the spatial dispersion of CO₂ , validated against physical models. This research not only reinforces the viability of CO₂ geological sequestration as a climate change mitigation strategy but also adds valuable advanced tools using machine learning for analyzing and safely managing these operations.
Recommended Citation
Iyegbekedo, Ikponmwosa Bright, "Machine Learning-Driven Quantification of CO₂ Plume Dynamics at IBDP Sites using Microseismic Data" (2024). Graduate Theses, Dissertations, and Problem Reports. 12525.
https://researchrepository.wvu.edu/etd/12525