Prediction of Rate of Penetration (ROP) in Drilling Operations Using Empirical Correlations and Drilling Parameters Analysis

Abstract

This thesis concerns the prediction and optimisation of the Rate of Penetration (ROP) by using the real drilling data of the well IA-152 and IA-159, Ain Daoui – Ohanet field, southern Algeria. We used two approaches: Response Surface Methodology (RSM) and a machine learning model using K-Means clustering and Random Forest. RSM results showed that the most influential parameters are Weight on Bit (WOB) and Rotary Speed (RPM) with R²= 95.16%. Robust ROP predictions were obtained with the Random Forest model and validated with standard metrics with deployment through an interactive Streamlit interface. This work presents a reliable framework for intelligent drilling optimisation in Algerian field conditions.