Facies Prediction From Core And Log Data Using Artificial Neural Network Technology

Facies classification is widely used to divide well data to obtain, for example, meaningful porosity-permeability relationships. This classification is usually done visually on cores and then extended to wireline logs from the cored wells. The challenge of this problem is to apply this classification to uncored wells based on the relationships observed at the cored wells. This paper presents a novel method to predict facies based on artificial neural network (ANN) techniques. We use a back-propagation ANN algorithm for recognizing the patterns of different facies. The ANN is trained on each facies of cored wells based on gamma ray, density, neutron, and resistivity logs. The facies selected for training the ANN are turbidites, debris flow, shoreface, and lower shoreface. The accuracy of facies predicted from logs alone using the ANN ranges from 75% to 93%.Gamma ray and density logs are the most crucial for some types of facies while neutron porosity log are more important for others. The approach of this work can be applied to fields where quantitative classification of a large number of logs by visual observation can be time-consuming and tedious. This approach can also be used to determine which logs are the most crucial for determining different types of facies. This can provide insights into future data collection.