Evolutionary genetic algorithm and expression trees-based model for enhancing reservoir porosity prediction using petrophysical logging data

Reservoir rock porosity is a crucial petrophysical property, which significantly impacts hydrocarbon reserve forecasting and petroleum economics. Traditional porosity measurement methods, such as core analysis and wireline log analysis are time-consuming and expensive. To address these limitations, the current study attempts an investigation to assess the performance of gene expression programming (GEP)-based predictive models and develops new correlations using a field petrophysical datasets for forecasting the porosity profile of clastic sedimentary rocks. The GEP-based models are trained using 183 datasets with four independent variables of gamma ray log, resistivity log, density log and shale volume to obtain the output response of rock porosity. Fitness score, statistical model performance indices, and Taylor diagrams are utilized to evaluate the expression-tree-based predictive models. It is found that the best porosity model exhibited promising performance with high precision of correlation of determination, root mean square error, and mean absolute error of 93.93%, 0.0050 and 0.0038% for testing samples. These results demonstrated that GEP has strong potential for accurately predicting reservoir effective porosity by capturing pore-space connectivity of rocks with formation radioactive concentration, bulk density, and rock resistivity environments. This systematic and innovative approach of GEP framework can be applied for crucial parameters forecasting in geoscience, geological and petroleum engineering disciplines, thereby avoiding tedious destructive testing procedures and reducing costs.