Prediction of Equivalent Circulating Density (ECD) in Drilling Operations Using Machine Learning: Case Study of Well A-1, Well A-2, and Well A-3

Abstract

This research discusses the prediction of the Equivalent Circulating Density (ECD) parameter using drilling datasets from Well A-1, Well A-2, and Well A-3 by leveraging machine learning. The objective of this study is to create a machine learning model capable of accurately predicting the ECD parameter, specifically using the Artificial Neural Network (ANN) and Random Forest (RF) algorithms. The methodology applied in this research consists of five stages: Data Collection, Exploratory Data Analysis (EDA), Data Preparation, Modeling, and Evaluation. Based on the modeling conducted, it was found that models with fewer input features exhibited better performance in blind tests. The best prediction results were obtained in the scenario using the Random Forest model with three features, with a RMSE is 0.041 and R2 value is 0.926.