Modeling Equivalent Circulating Density During Drilling Operations in the Gulf of Thailand

Abstract

Equivalent Circulating Density (ECD) represents the total hydrostatic pressure generated by drilling fluid while in motion. This prevents the internal pressure within the well from exceeding the fracture resistance of the rock, which could lead to lost circulation and an inability to effectively control the wellbore pressure. This research aims to predict ECD in 6.125-inch production section in Gulf of Thailand field by using five machine learning algorithms were utilized, namely Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Networks (ANN), Gradient Boosting (GB), and Extreme Gradient Boosting (XGBoost). Various sensors from the Measure While Drilling (MWD), Logging While Drilling (LWD), and Pressure While Drilling (PWD) tools were used to collect raw data, totaling 38,863 records and 24 variables to predict the ECD value. The dataset was randomly split into 80% for training and validation and 20% for testing. The results indicate that the RF technique outperformed the other models in predicting ECD values, producing the lowest RMSE of 0.031. Therefore, the RF model is most suitable for further development and real-time application in predicting ECD values.