{"id":{"repo_id":"texas","oai_identifier":"oai:repositories.lib.utexas.edu:2152/131921"},"canonical_url":"https://search.dev.ndltd.org/etd/texas/oai:repositories.lib.utexas.edu:2152/131921","repository":{"repo_id":"texas","name":"University of Texas","base_url":"https://repositories.lib.utexas.edu/server/oai/request"},"display":{"title":"Quantification of rock fabric through multi-scale image data analysis for enhanced rock typing and formation evaluation","abstract":"Spatial variation of rock fabric (i.e., spatial distribution of minerals, pores, and saturating fluids) affects the estimation of petrophysical, mechanical, and compositional properties of sedimentary geological formations. Quantification of rock fabric and its integration with rock classification and formation evaluation efforts are critical for reliable estimation of petrophysical properties in complex formations with rapid spatial variation in rock fabric. The concept of rock fabric is often included in reservoir characterization workflows in the form of expert-derived facies obtained through qualitative interpretation of core image data. The identified facies are frequently employed in the development of geological models. However, conventional methods for facies identification can be subjective, qualitative, and time-consuming. Only the visual aspect of the core, when exposed to natural light, is used in the process of facies identification, dismissing critical visual information revealed by other types of wellbore/core image data. Furthermore, automated workflows independent of previously identified expert-derived lithofacies are not available. In this dissertation, I quantified continuous rock-fabric-related features through image analysis techniques using two-dimensional (2D) computerized tomography (CT) scan images, three-dimensional (3D) CT-scan image stacks, slabbed whole-core photos, and image logs. I computed grey-scale/color and textural features from the mentioned wellbore/core image data. The extracted features numerically quantify the visual content of the employed image data. I used the extracted image-based rock-fabric-related features to conduct image-based rock classification using unsupervised learning techniques. I integrated image-based features with (1) routine core analysis data and (2) conventional and advanced well logs for image-based rock classification and enhanced formation evaluation. I compared supervised and unsupervised learning techniques for image-based rock classification. I also used whole-core 3D CT-scan image stacks and micro-CT-scan image stacks to quantify directional rock tortuosity using pore-scale numerical simulations of fluid flow, electric current flow, and particle diffusion as well as geometrical tortuosity quantification. I evaluated the proposed workflows using wellbore/core image data, conventional well logs, specialized well logs, and routine core analysis data from siliciclastic, organic-rich mudrock, and carbonate formations, with rapid spatial variation in rock fabric. I introduced a workflow for automatic rock classification through the integration of quantitative image-based rock-fabric-related features and routine core analysis data. This was accomplished using a cost function that compares class-based estimates of permeability with core-measured permeability for increasing number rock classes, providing a novel alternative for automated rock classification compared to conventional rock classification techniques. The detected integrated rock classes honor both the visual aspect and the fluid-flow characteristics of the evaluated formations, providing a platform to expedite the identification of expert-derived lithofacies. Rock classes detected through the introduced automated rock classification workflow agreed with expert-derived lithofacies with accuracies of up to 86%. Additionally, class-based formation evaluation using the detected image-based and integrated rock classes improved estimates of petrophysical properties such as permeability and water saturation. The use of class-based rock physics models reduced the mean relative errors in permeability and water saturation estimates up to 89% and 41 % when compared to core measurements, respectively. Rock classes obtained using supervised learning techniques showed higher accuracies compared to expert-derived lithofacies than those obtained through the introduced automated rock classification workflow. However, the introduced automated rock classification workflow does not require expert-derived lithofacies. Therefore, no subjectivity is introduced in the obtained results due to the subjective nature of expert-derived lithofacies and the time required for expert-derived lithofacies identification is eliminated. Furthermore, using unsupervised learning algorithms in the proposed automated rock classification workflow eliminates the time required for the training step in supervised learning algorithms, which can be considerably high depending on the size of the dataset and the specific algorithm. The workflows introduced in this dissertation provide a framework for quantitative integration of wellbore/core image data in formation evaluation and rock classification efforts. The introduced workflows expedite the process of facies identification and remove the subjectivity introduced in conventional workflows for facies identification, impacting reservoir characterization and the development of geological/reservoir models.","abstract_html":"Spatial variation of rock fabric (i.e., spatial distribution of minerals, pores, and saturating fluids) affects the estimation of petrophysical, mechanical, and compositional properties of sedimentary geological formations. Quantification of rock fabric and its integration with rock classification and formation evaluation efforts are critical for reliable estimation of petrophysical properties in complex formations with rapid spatial variation in rock fabric. The concept of rock fabric is often included in reservoir characterization workflows in the form of expert-derived facies obtained through qualitative interpretation of core image data. The identified facies are frequently employed in the development of geological models. However, conventional methods for facies identification can be subjective, qualitative, and time-consuming. Only the visual aspect of the core, when exposed to natural light, is used in the process of facies identification, dismissing critical visual information revealed by other types of wellbore/core image data. Furthermore, automated workflows independent of previously identified expert-derived lithofacies are not available. In this dissertation, I quantified continuous rock-fabric-related features through image analysis techniques using two-dimensional (2D) computerized tomography (CT) scan images, three-dimensional (3D) CT-scan image stacks, slabbed whole-core photos, and image logs. I computed grey-scale/color and textural features from the mentioned wellbore/core image data. The extracted features numerically quantify the visual content of the employed image data. I used the extracted image-based rock-fabric-related features to conduct image-based rock classification using unsupervised learning techniques. I integrated image-based features with (1) routine core analysis data and (2) conventional and advanced well logs for image-based rock classification and enhanced formation evaluation. I compared supervised and unsupervised learning techniques for image-based rock classification. I also used whole-core 3D CT-scan image stacks and micro-CT-scan image stacks to quantify directional rock tortuosity using pore-scale numerical simulations of fluid flow, electric current flow, and particle diffusion as well as geometrical tortuosity quantification. I evaluated the proposed workflows using wellbore/core image data, conventional well logs, specialized well logs, and routine core analysis data from siliciclastic, organic-rich mudrock, and carbonate formations, with rapid spatial variation in rock fabric. I introduced a workflow for automatic rock classification through the integration of quantitative image-based rock-fabric-related features and routine core analysis data. This was accomplished using a cost function that compares class-based estimates of permeability with core-measured permeability for increasing number rock classes, providing a novel alternative for automated rock classification compared to conventional rock classification techniques. The detected integrated rock classes honor both the visual aspect and the fluid-flow characteristics of the evaluated formations, providing a platform to expedite the identification of expert-derived lithofacies. Rock classes detected through the introduced automated rock classification workflow agreed with expert-derived lithofacies with accuracies of up to 86%. Additionally, class-based formation evaluation using the detected image-based and integrated rock classes improved estimates of petrophysical properties such as permeability and water saturation. The use of class-based rock physics models reduced the mean relative errors in permeability and water saturation estimates up to 89% and 41 % when compared to core measurements, respectively. Rock classes obtained using supervised learning techniques showed higher accuracies compared to expert-derived lithofacies than those obtained through the introduced automated rock classification workflow. However, the introduced automated rock classification workflow does not require expert-derived lithofacies. Therefore, no subjectivity is introduced in the obtained results due to the subjective nature of expert-derived lithofacies and the time required for expert-derived lithofacies identification is eliminated. Furthermore, using unsupervised learning algorithms in the proposed automated rock classification workflow eliminates the time required for the training step in supervised learning algorithms, which can be considerably high depending on the size of the dataset and the specific algorithm. The workflows introduced in this dissertation provide a framework for quantitative integration of wellbore/core image data in formation evaluation and rock classification efforts. The introduced workflows expedite the process of facies identification and remove the subjectivity introduced in conventional workflows for facies identification, impacting reservoir characterization and the development of geological/reservoir models.","abstract_has_math":false,"creators":["Gonzalez Barrios, Andres Ricardo"],"institution":"The University of Texas at Austin","degree_name":"Doctor of Philosophy","degree_level":"Doctoral","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Heidari, Zoya"],"committee_chairs":[],"committee_members":["Sepehrnoori, Kamy","Daigle, Hugh","Prodanovic, Masa","Lopez, Olivier"],"year":2023,"date_issued":"2023-05","date_published":"2023-05","updated_at":"2026-07-24T05:01:12Z","subjects":["Image analysis","Rock classification","Rock fabric","Petrophysics","Formation evaluation","Automation","Unsupervised learning"],"languages":["en"],"rights":[],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.26153/tsw/59265"],"render_values":[{"text":"https://doi.org/10.26153/tsw/59265","href":"https://doi.org/10.26153/tsw/59265","code":true}]}]},"links":{"outbound_url":"https://hdl.handle.net/2152/131921","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Heidari, Zoya"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Sepehrnoori, Kamy","Daigle, Hugh","Prodanovic, Masa","Lopez, Olivier"]},{"key":"dc:creator","label":"Author","values":["Gonzalez Barrios, Andres Ricardo"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-03-07T23:01:30Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-03-07T23:01:30Z"]},{"key":"dc:date.issued","label":"Date","values":["2023-05"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Doctoral"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctor of Philosophy"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["The University of Texas at Austin"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Image analysis","Rock classification","Rock fabric","Petrophysics","Formation evaluation","Automation","Unsupervised learning"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/2152/131921","https://doi.org/10.26153/tsw/59265"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Spatial variation of rock fabric (i.e., spatial distribution of minerals, pores, and saturating fluids) affects the estimation of petrophysical, mechanical, and compositional properties of sedimentary geological formations. Quantification of rock fabric and its integration with rock classification and formation evaluation efforts are critical for reliable estimation of petrophysical properties in complex formations with rapid spatial variation in rock fabric. The concept of rock fabric is often included in reservoir characterization workflows in the form of expert-derived facies obtained through qualitative interpretation of core image data. The identified facies are frequently employed in the development of geological models. However, conventional methods for facies identification can be subjective, qualitative, and time-consuming. Only the visual aspect of the core, when exposed to natural light, is used in the process of facies identification, dismissing critical visual information revealed by other types of wellbore/core image data. Furthermore, automated workflows independent of previously identified expert-derived lithofacies are not available. In this dissertation, I quantified continuous rock-fabric-related features through image analysis techniques using two-dimensional (2D) computerized tomography (CT) scan images, three-dimensional (3D) CT-scan image stacks, slabbed whole-core photos, and image logs. I computed grey-scale/color and textural features from the mentioned wellbore/core image data. The extracted features numerically quantify the visual content of the employed image data. I used the extracted image-based rock-fabric-related features to conduct image-based rock classification using unsupervised learning techniques. I integrated image-based features with (1) routine core analysis data and (2) conventional and advanced well logs for image-based rock classification and enhanced formation evaluation. I compared supervised and unsupervised learning techniques for image-based rock classification. I also used whole-core 3D CT-scan image stacks and micro-CT-scan image stacks to quantify directional rock tortuosity using pore-scale numerical simulations of fluid flow, electric current flow, and particle diffusion as well as geometrical tortuosity quantification. I evaluated the proposed workflows using wellbore/core image data, conventional well logs, specialized well logs, and routine core analysis data from siliciclastic, organic-rich mudrock, and carbonate formations, with rapid spatial variation in rock fabric. I introduced a workflow for automatic rock classification through the integration of quantitative image-based rock-fabric-related features and routine core analysis data. This was accomplished using a cost function that compares class-based estimates of permeability with core-measured permeability for increasing number rock classes, providing a novel alternative for automated rock classification compared to conventional rock classification techniques. The detected integrated rock classes honor both the visual aspect and the fluid-flow characteristics of the evaluated formations, providing a platform to expedite the identification of expert-derived lithofacies. Rock classes detected through the introduced automated rock classification workflow agreed with expert-derived lithofacies with accuracies of up to 86%. Additionally, class-based formation evaluation using the detected image-based and integrated rock classes improved estimates of petrophysical properties such as permeability and water saturation. The use of class-based rock physics models reduced the mean relative errors in permeability and water saturation estimates up to 89% and 41 % when compared to core measurements, respectively. Rock classes obtained using supervised learning techniques showed higher accuracies compared to expert-derived lithofacies than those obtained through the introduced automated rock classification workflow. However, the introduced automated rock classification workflow does not require expert-derived lithofacies. Therefore, no subjectivity is introduced in the obtained results due to the subjective nature of expert-derived lithofacies and the time required for expert-derived lithofacies identification is eliminated. Furthermore, using unsupervised learning algorithms in the proposed automated rock classification workflow eliminates the time required for the training step in supervised learning algorithms, which can be considerably high depending on the size of the dataset and the specific algorithm. The workflows introduced in this dissertation provide a framework for quantitative integration of wellbore/core image data in formation evaluation and rock classification efforts. The introduced workflows expedite the process of facies identification and remove the subjectivity introduced in conventional workflows for facies identification, impacting reservoir characterization and the development of geological/reservoir models."]},{"key":"dc:format.mimetype","label":"Dc Format Mimetype","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Quantification of rock fabric through multi-scale image data analysis for enhanced rock typing and formation evaluation"]}]}],"canonical_facts":{"dc:contributor.advisor":["Heidari, Zoya"],"dc:contributor.committeemember":["Sepehrnoori, Kamy","Daigle, Hugh","Prodanovic, Masa","Lopez, Olivier"],"dc:creator":["Gonzalez Barrios, Andres Ricardo"],"dc:date.accessioned":["2025-03-07T23:01:30Z"],"dc:date.available":["2025-03-07T23:01:30Z"],"dc:date.issued":["2023-05"],"dc:description.abstract":["Spatial variation of rock fabric (i.e., spatial distribution of minerals, pores, and saturating fluids) affects the estimation of petrophysical, mechanical, and compositional properties of sedimentary geological formations. Quantification of rock fabric and its integration with rock classification and formation evaluation efforts are critical for reliable estimation of petrophysical properties in complex formations with rapid spatial variation in rock fabric. The concept of rock fabric is often included in reservoir characterization workflows in the form of expert-derived facies obtained through qualitative interpretation of core image data. The identified facies are frequently employed in the development of geological models. However, conventional methods for facies identification can be subjective, qualitative, and time-consuming. Only the visual aspect of the core, when exposed to natural light, is used in the process of facies identification, dismissing critical visual information revealed by other types of wellbore/core image data. Furthermore, automated workflows independent of previously identified expert-derived lithofacies are not available. In this dissertation, I quantified continuous rock-fabric-related features through image analysis techniques using two-dimensional (2D) computerized tomography (CT) scan images, three-dimensional (3D) CT-scan image stacks, slabbed whole-core photos, and image logs. I computed grey-scale/color and textural features from the mentioned wellbore/core image data. The extracted features numerically quantify the visual content of the employed image data. I used the extracted image-based rock-fabric-related features to conduct image-based rock classification using unsupervised learning techniques. I integrated image-based features with (1) routine core analysis data and (2) conventional and advanced well logs for image-based rock classification and enhanced formation evaluation. I compared supervised and unsupervised learning techniques for image-based rock classification. I also used whole-core 3D CT-scan image stacks and micro-CT-scan image stacks to quantify directional rock tortuosity using pore-scale numerical simulations of fluid flow, electric current flow, and particle diffusion as well as geometrical tortuosity quantification. I evaluated the proposed workflows using wellbore/core image data, conventional well logs, specialized well logs, and routine core analysis data from siliciclastic, organic-rich mudrock, and carbonate formations, with rapid spatial variation in rock fabric. I introduced a workflow for automatic rock classification through the integration of quantitative image-based rock-fabric-related features and routine core analysis data. This was accomplished using a cost function that compares class-based estimates of permeability with core-measured permeability for increasing number rock classes, providing a novel alternative for automated rock classification compared to conventional rock classification techniques. The detected integrated rock classes honor both the visual aspect and the fluid-flow characteristics of the evaluated formations, providing a platform to expedite the identification of expert-derived lithofacies. Rock classes detected through the introduced automated rock classification workflow agreed with expert-derived lithofacies with accuracies of up to 86%. Additionally, class-based formation evaluation using the detected image-based and integrated rock classes improved estimates of petrophysical properties such as permeability and water saturation. The use of class-based rock physics models reduced the mean relative errors in permeability and water saturation estimates up to 89% and 41 % when compared to core measurements, respectively. Rock classes obtained using supervised learning techniques showed higher accuracies compared to expert-derived lithofacies than those obtained through the introduced automated rock classification workflow. However, the introduced automated rock classification workflow does not require expert-derived lithofacies. Therefore, no subjectivity is introduced in the obtained results due to the subjective nature of expert-derived lithofacies and the time required for expert-derived lithofacies identification is eliminated. Furthermore, using unsupervised learning algorithms in the proposed automated rock classification workflow eliminates the time required for the training step in supervised learning algorithms, which can be considerably high depending on the size of the dataset and the specific algorithm. The workflows introduced in this dissertation provide a framework for quantitative integration of wellbore/core image data in formation evaluation and rock classification efforts. The introduced workflows expedite the process of facies identification and remove the subjectivity introduced in conventional workflows for facies identification, impacting reservoir characterization and the development of geological/reservoir models."],"dc:format.mimetype":["application/pdf"],"dc:identifier.uri":["https://hdl.handle.net/2152/131921","https://doi.org/10.26153/tsw/59265"],"dc:language.iso":["en"],"dc:subject":["Image analysis","Rock classification","Rock fabric","Petrophysics","Formation evaluation","Automation","Unsupervised learning"],"dc:title":["Quantification of rock fabric through multi-scale image data analysis for enhanced rock typing and formation evaluation"],"dc:type":["Thesis"],"thesis:degree_level":["Doctoral"],"thesis:degree_name":["Doctor of Philosophy"],"thesis:institution_name":["The University of Texas at Austin"]},"updated_at":"2026-07-24T05:01:12Z"}