Support vector machines for petrophysical modelling and lithoclassification

Simple item page
dc.contributor.advisorGates, Ian D.
dc.contributor.advisorAzaiez, Jalel
dc.contributor.authorAl-Anazi, Ammal Fannoush Khalifah
dc.date.accessioned2017-12-18T22:18:32Z
dc.date.available2017-12-18T22:18:32Z
dc.date.issued2011
dc.descriptionBibliography: p. 362-370en
dc.description.abstractGiven increasing challenges of oil and gas production from partially depleted conventional or unconventional reservoirs, reservoir characterization is a key element of the reservoir development workflow. Reservoir characterization impacts well placement, injection and production strategies, and field management. Reservoir characterization projects point and line data to a large three-dimensional volume. The relationship between variables, e.g. porosity and permeability, is often established by regression yet the complexities between measured variables often lead to poor correlation coefficients between the regressed variables. Recent advances in machine learning methods have provided attractive alternatives for constructing interpretation models of rock properties in heterogeneous reservoirs. Here, Support Vector Machines (SVMs ), a class of a learning machine that is formulated to output regression models and classifiers of competitive generalization capability, has been explored to determine its capabilities for determining the relationship, both in regression and in classification, between reservoir rock properties. This thesis documents research on the capability of SVMs to model petrophysical and elastic properties in heterogeneous sandstone and carbonate reservoirs. Specifically, the capabilities of SVM regression and classification has been examined and compared to neural network-based methods, namely multilayered neural networks, radial basis function neural networks, general regression neural networks, probabilistic neural networks, and linear discriminant analysis. The petrophysical properties that have been evaluated include porosity, permeability, Poisson's ratio and Young's modulus. Statistical error analysis reveals that the SVM method yields comparable or superior predictions of petrophysical and elastic rock properties and classification of the lithology compared to neural networks. The SVM method also shows uniform prediction capability under the effect of small sample sizes.
dc.format.extentxxiv, 370 leaves : ill. ; 30 cm.en
dc.identifier.citationAl-Anazi, A. F. (2011). Support vector machines for petrophysical modelling and lithoclassification (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/4092en_US
dc.identifier.doihttp://dx.doi.org/10.11575/PRISM/4092
dc.identifier.urihttp://hdl.handle.net/1880/105093
dc.language.isoeng
dc.publisher.institutionUniversity of Calgaryen
dc.publisher.placeCalgaryen
dc.rightsUniversity of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.
dc.titleSupport vector machines for petrophysical modelling and lithoclassification
dc.typedoctoral thesis
thesis.degree.disciplineChemical and Petroleum Engineering
thesis.degree.grantorUniversity of Calgary
thesis.degree.nameDoctor of Philosophy (PhD)
ucalgary.item.requestcopytrue
ucalgary.thesis.accessionTheses Collection 58.002:Box 1991 627942841
ucalgary.thesis.notesUARCen
ucalgary.thesis.uarcreleaseyen
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
thesis_Al-Anazi_2011.pdf
Size:
148.31 MB
Format:
Adobe Portable Document Format
Description:
Thesis
Download
Collections
Legacy Theses