Quantitative seismic interpretation and machine learning applications for subsurface characterization and modeling

Simple item page
dc.contributor.advisorPranter, Matthew J.
dc.contributor.authorCaf, Abidin
dc.contributor.committeeMemberMarfurt, Kurt J.
dc.contributor.committeeMemberBedle, Heather
dc.contributor.committeeMemberCarpenter, Brett M.
dc.contributor.committeeMemberRai, Chandra
dc.date.accessioned2022-12-06T16:45:59Z
dc.date.available2022-12-06T16:45:59Z
dc.date.issued2022-12
dc.date.manuscript2022-12-05
dc.description.abstractQuantitative seismic interpretation and geostatistical modeling methods have been widely used for subsurface reservoir characterization. However, the task becomes challenging due to the reservoir complexity and limited well control. To address these challenges, this research explores workflows that combine supervised machine learning, quantitative seismic interpretation, and seismic-constraining reservoir modeling methods to effectively reduce uncertainty in predicting multiscale subsurface heterogeneity. These workflows help mitigate the risks and uncertainties of exploring and developing potential reservoirs for hydrocarbon exploration and production or subsurface carbon sequestration. Techniques applied in this study integrate multiple sources of data to characterize complex reservoirs across different fields in north America. This dissertation presents three case studies combining new and traditional subsurface characterization techniques at different scales. The research starts with supervised machine learning, 3D seismic data, and well-log information to map the seismic scale diagenetic imprint and its corresponding reservoir quality on a Permian Basin reservoir. Then, I present a workflow that integrates core-derived petrophysical measurements, well logs, and pre-stack seismic data through supervised machine learning to map the seismic-scale spatial variability of petrophysically significant facies of a carbonate reservoir targeted for carbon geosequestration. Lastly, I present a seismic-constrained reservoir modeling and simulation workflow that combines the seismic-scale petrophysically defined facies information with well log and core data to map small-scale stratigraphic variability of petrophysical properties, CO2 storage capacity, and subsurface fluid flow behavior for long-term carbon sequestration. The illustrated workflows showed that the subsurface properties, such as lithology and petrofacies information, could be extracted on a seismic scale with the help of supervised machine learning. Additionally, this information can be used to better constrain reservoir models and reduce uncertainty where the well control is sparse.en_US
dc.identifier.urihttps://shareok.org/handle/11244/336875
dc.languageen_USen_US
dc.subjectGeophysicsen_US
dc.subjectReservoir modelingen_US
dc.subjectReservoir characterizationen_US
dc.subjectSupervised machine learningen_US
dc.subjectQuantitative seismic interpretationen_US
dc.subjectSeismic interpretationen_US
dc.subjectCarbon sequestrationen_US
dc.subjectReservoir simulationen_US
dc.thesis.degreePh.D.en_US
dc.titleQuantitative seismic interpretation and machine learning applications for subsurface characterization and modelingen_US
ou.groupMewbourne College of Earth and Energy::School of Geosciencesen_US

Files

Original bundle
Now showing 1 - 2 of 2
Thumbnail Image
Name:
2022_Caf_Abidin_Dissertation.pdf
Size:
13.77 MB
Format:
Adobe Portable Document Format
Description:
Download
No Thumbnail Available
Name:
2022_Caf_Abidin_Dissertation.docx
Size:
54.56 MB
Format:
Microsoft Word XML
Description:
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

OU - Dissertations