A Comparison of Machine Learning Approaches for Prediction of Permeability using Well Log Data in the Hydrocarbon Reservoirs

Permeability is a vital parameter in reservoir engineering that affects production directly. Since this parameter's significance obvious, finding way for accurate determination of permeability essential as well. In paper, the two notable carbonate reservoirs (Ilam and Sarvak) southwest Iran was predicted by several different methods, level accuracy all models compared. For purpose, Multi-Layer Perceptron Neural Network (MLP), Radial Basis Function (RBF), Support Vector Regression (SVR), decision tree (DT), random forest (RF) methods were chosen. The full set real well-logging data investigated forest, five them selected potent variables. Depth, Computed gamma-ray log (CGR), Spectral (SGR), Neutron porosity (NPHI), density (RHOB) considered efficacious variables used input data, while output. It should be noted values are derived from core analysis. Statistical parameters like coefficient ( ), root mean square error (RMSE) standard deviation (SD) determined train, test, total sets. Based on statistical graphical results, SVM DT perform more accurately than others. RMSE, SD R2values 0.38, 1.63, 0.97 0.44, 2.89, 0.96 respectively. results best-proposed paper then compared with outcome empirical equation prediction. comparison indicates artificial intelligence traditional estimation, such proposed equations. Doi: 10.28991/HEF-2021-02-02-01 Full Text: PDF