Semester
Spring
Date of Graduation
2000
Document Type
Thesis
Degree Type
MS
College
Statler College of Engineering and Mineral Resources
Department
Petroleum and Natural Gas Engineering
Committee Chair
Khashayar Aminian
Committee Co-Chair
H. Ilkin Bilgesu
Abstract
The prediction of permeability is a critical, key step for reservoir modeling and management of oil recovery operations. Previous studies have successfully demonstrated that the new technology called Artificial Neural Network (ANN), a biologically inspired, massive parallel, distributed information processing system, is an excellent tool for permeability predictions using well log data. This technology overcomes the drawbacks caused by the inherent heterogeneity of the reservoir and lack of sufficient cores or pressure transient tests, allowing to define reservoir characterization within an acceptable accuracy while maintaining costs low. The methodology used in this study takes advantage of this technology to accomplish such a task.;An ANN was developed obtaining a correlation coefficient R2 of 0.975 when compared permeability predictions to actual measurements for seven wells using their well log data in a reservoir in West Virginia, USA. Thereafter, the ANN was used to forecast the permeability for the rest of the wells in the reservoir. Thus, based on the well permeability profile, the Flow Capacity and Average Permeability was determined and mapped throughout the field which defined the most productive areas in the reservoir and helped to improve the production history matching.
Recommended Citation
Riera, Alexis Jose, "Predicting permeability and flow capacity distribution with back-propagation artificial neural networks" (2000). Graduate Theses, Dissertations, and Problem Reports. 1184.
https://researchrepository.wvu.edu/etd/1184