Application of multiple regression technique for predicting drilling rate in loss zone

Abstract

Prediction of drilling rate is important to optimize drilling operation, and it is typically performed based on gathering drilling data from the history of nearby wells. Previous researchers have proposed numerous techniques to improve the mathematical model's accuracy and the multiple regression techniques in predicting the drilling rate. However, 100% accuracy of the drilling rate model as compared to field data is yet achievable. This study proposes a modification of multiple regression techniques by gathering data according to the loss zone. Thus, the Bourgoyne and Young model has been used, and the drilling data from the North Kuwait field was selected for the multiple regression analysis. Multiple regression analysis was compared according to loss zone, formation type, BHA interval and single well to achieve the Bourgoyne and Young model's coefficients. The achieved coefficients were used for drilling rate estimation using the model and compared with the real field data. Based on the findings, it was observed that using multiple regression techniques according to loss zone would result in more accurate prediction than other methods with the improvement of R2 value from 0.523 to 0.646 (12.3% improvement). Therefore, it is suggested that categorizing drilling data according to the loss zone for multiple regression analysis needs to be considered for drilling rate prediction for better accuracy.

More by Sonny Irawan

• An explainable physics-informed neural network prototype for synthetic proppant transport in a two-dimensional fracture domainPhysics of Fluids · 2026
• Physics‐Guided Neural Regression Modeling of Energy‐Balance Proxies for Hydraulic Fracture Propagation in 1D–3DInternational Journal for Numerical and Analytical Methods in Geomechanics · 2026
• Modeling and Prediction of Kappaphycus alvarezii Viscosity Using Artificial Neural Network and Adaptive Neuro-Fuzzy Inference SystemJournal of Earth Energy Science Engineering and Technology · 2025 · Citations: 1
• Physics-Informed Machine Learning for Predicting Drilling Fluid Losses in Naturally Fractured Caspian FormationsSPE Annual Caspian Technical Conference and Exhibition · 2025 · Citations: 4