Full text loading...
-
Permeability Prediction Using Artificial Neural Network and Least Square Support Vector Machine Methods
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 80th EAGE Conference and Exhibition 2018, Jun 2018, Volume 2018, p.1 - 5
Abstract
Permeability of hydrocarbon reservoirs plays a significant part in all stages of oil and gas recovery including well completion, production, and reservoir management. A wide number of statistical methods have already been implemented to estimate reservoir permeability based on the available well logs and core data. Nevertheless, determining representative and accurate permeability values still remains an active challenge. This paper introduces the application of two statistical methods known as “artificial neural network” (ANN) and “least square support vector machine” (LSSVM) to estimate permeability in an Iranian oil field. Prior to building the models, a pre-processing was performed and from all the available well logs, three with better correlations were selected for regression. Gamma ray, sonic, and thermal neutron porosity logs together with core porosity were employed to estimate permeability. Obtained results indicate that predictions by both models accord ideally with core data. LSSVM was also proven to outperform the ANN model yielding an overall correlation coefficient of 0.9848.