DRILLING HYDRAULICS OPTIMIZATION: DESIGNING FOR FLOW AND POWER

ABSTRACT

Drilling hydraulics optimization is pivotal for minimizing non-productive time (NPT) and lowering drilling costs, thereby enhancing drilling performance, especially in less competent formations of the Niger Delta region during drilling operations. This study aim at optimizing drilling hydraulics by comparing traditional Hydraulic Horsepower–based method with flowbased design method using advanced analytical techniques on offset field data. This is addressed by developing an optimized predictive model for Rate of Penetration (ROP) using machine learning techniques incorporating estimated values of Jet Impact Force (JIF) and Hydraulic Horsepower (HHP) into the design, an unexplored area in current research. The research focused on key drilling parameters such as Flow Rate (FR), Revolutions per Minute (RPM), Yield Point (YP), and other parameters of significant impact. Machine learning models—specifically Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Support Vector Machine (SVM)—were adopted. Results from these machine learning models were compared using multiple performance metrics. It was observed that XGBoost performed well with a R-squared (R2 ) value of 98.39%. The integration of JIF and HHP significantly improved the predictive accuracy and drilling performance, leading to reduced non-productive time (NPT) and optimized bit designs. The sensitivity analysis revealed the greater impact of Jet Impact Force (JIF) on the ROP model compared to Hydraulic Horsepower (HHP), underscoring the need for parameter optimization across different depths.