Density log prediction using artificial intelligence tools case study: Ordovician reservoir (Oued Zine field Sbaa Basin)

Abstract

In practice, the density of rocks is measured either through data logging tools during drilling or through techniques and devices available in laboratories. However, these measurements are not always readily available and can be costly, necessitating the use of new techniques that provide accurate results in less time and at a lower cost. This work aims to apply several artificial intelligence models to predict rock density in gas reservoirs. Machine learning (ML) techniques, Gradient Boosting (GB), Random Forest (RF), and Extreme Gradient Boosting (XGB) were applied using drilling data as inputs. A vertical well containing 295 data points was used to apply the models to rocks consisting of sand and claystone. The results showed that all three models were able to predict rock density with high accuracy. The correlation coefficient (R) values were 0.807 and 0.781 in the RF1 and XGB2 models, respectively. Meanwhile, the GB3 model had an R value of 0.83, indicating higher accuracy, making GB3 the best model for predicting density values. Each model can predict rock density at a low cost and in a timely manner.