3D reservoir modelling with the aid of artificial neural networks

This paper presents a study using artificial neural networks (ANN)to perform automatic estimation of rock permeability in a reservoir scale. Three well log responses (acoustic time, gamma-ray and deep induction) were used to predict formation permeability. Permeability from core data and conventional computing have been used to test the results of the neural network. Results from the ANN are in good agreement with core analysis and better than conventional computing data. Based on artificial neural network, combining with the thought of facies controlled modeling, this study presents a reasonable and valuable method to establish the 3D reservoir model, which could provide significant geological and geophysical basis for the further research of improving recovery ratio and the potential of residual oil. Finally, it is shown that the application of artificial neural networks in permeability prediction reduces costs and predicts the reservoir properties more accurately.