Investigation on Estimation and Prediction of Resistivity of Limestone Rocks based on Physico-Mechanical Properties of Rocks

Abstract

Prediction of rock resistivity indirectly is of paramount importance in several geophysical and civil engineering applications. Physico-mechanical properties such as p-wave velocity, porosity and dry density tend to have a good correlation with electrical resistivity of rocks. Conventional approaches for measuring resistivity produce results which may consume more time and efforts and are not accessible every location. To overcome this, an Artificial Neural Network (ANN) model was evolved in this study, using Python and TensorFlow. The model was trained using known values to predict electrical resistivity of unknown and similar materials. Actual results of resistivity were compared with resistivity values obtained from ANN model. The obtained values were evaluated for reliability using non-linear regression models. It was observed that predicted resistivity values generated using p-wave velocity were more reliable. Also, validations made based on the ANN model, using mean absolute error (MAE) and average residuals indicate that P-wave velocity is the most reliable predictor, achieving the lowest MAE (4.638) and near-zero residuals (-0.005), while porosity and dry density showed higher errors and weaker correlations. This study revealed that the ANN model developed results in reliable predictions of rock resistivity based on p-wave values. © 2025, World Researchers Associations. All rights reserved.