Gene Expression Programming-Based Gas Reservoir Porosity Prediction

Porosity is one of the essential properties of a reservoir rock, which relates to the volume of void space in a reservoir rock. It significantly impacts the calculation of hydrocarbon reserves and petroleum economics, making it a key petrophysical characteristic of the rock. The conventional techniques for determining porosity are time-consuming and costly. As a result, different approaches have been developed to predict porosity. The major aims of this study are i) to construct a gene expression programming (GEP)-based porosity model and ii) develop a correlation to obtain rock porosity. Through extensive analysis and experimentation, the GEP-based model is shown to be effective in accurately predicting porosity. A total number of 183 datasets, with four input variables and single output response, are randomly divided into 129 and 54 data sets for training and testing to develop the model, respectively. According to the simulated results with GEP, the prediction shows of mean absolute percentage error and a high correlation coefficient of 0.5% and 94.14 %, for the testing phase, respectively. GEP has a strong potential for predicting the reservoir porosity in the considered parameters. This research strategy can be applied to predict the reservoir rock and fluid properties for oil and gas reservoir characterization and simulation studies.