A Self-Adaptive Artificial Neural Network Technique to Estimate Static Young's Modulus Based on Well Logs

Abstract

Abstract Static Young's modulus (Estatic) is an essential parameter needed to develop the earth geomechanical model, Young's modulus (E) considerably varies with the change in the lithology. Recently, artificial intelligence (AI) techniques were used to estimate Estatic for carbonate formation. In this study, artificial neural network (ANN) was used to estimate Estatic for sandstone formation. In this study, the ANN design parameters were optimized using the self-adaptive differential evolution (SaDE) optimization algorithm. The ANN model was trained to predict Estatic from conventional well log data such as bulk density, compressional time, and shear time. 409 data points from Well-A were used to train the ANN model which was then tested using 183 unseen data from the same well and validated on 11 data points from a different well (Well-B). The developed SaDE-ANN model estimated Estatic for the training data set with a very low average absolute percentage error (AAPE) of 0.46%, very high correlation coefficient (R) of 0.999 and coefficient of determination (R2) of 0.9978. And the Estatic values of testing data set were estimated with AAPE, R, and R2 of 1.46%, 0.998, and 0.9951, respectively. These results confirmed the high accuracy of the developed Estatic model.