Permeability Prediction Using Artificial Neural Network (ANN): A Case Study of Uinta Basin

Abstract

Abstract The purpose of this paper is to develop a methodology to predict the permeability for wells in the same field using conventional logs (Gamma Ray, Neutron logs and Density log). This methodology involves the application of Error Back Propagation Neural Network. The advantages of this learning algorithm is that, an error in the final output gets back propagated and gradually updates the weight and hence leads to the best network structure. Present study was made using published literature on Uinta Basin, southwest Utah field (available on Utah Geological Survey's (UGS) website). It has an areal extent of 14900 km2. In this, study data from 13 wells was taken. Data from seven cored wells in the field was used for training, and subsequently prediction and verification was done on core permeability for remaining six wells. The result of this study shows that ANN generated permeability is consistent with core analysis result. This study was done using MATLAB® 6.1’s ANN Toolbox.