Predicting Drilling Hazards in Carbonate Reservoirs: A Machine Learning Approach for Loss Circulation Mitigation

Summary Loss circulation in carbonate reservoirs poses a significant risk to drilling operations, resulting in substantial non-productive time and cost overruns. Despite advances in traditional geological risk assessments, these methods often fall short in capturing the complex relationships between parameters in heterogeneous environments. Recent breakthroughs in machine learning for drilling applications ( Pang et al., 2022 ) have demonstrated the potential of data-driven approaches to enhance predictability in complex subsurface settings. Building on this foundation, our study aims to develop a machine learning-based predictive system that leverages multi-domain geoscience data, including spatial coordinates, seismic attributes, geological facies, petrophysical properties, and geomechanical parameters, to forecast loss circulation zones within a carbonate reservoir. By integrating these diverse data sources, our approach enables objective, reproducible risk assessments that minimize human interpretation bias, ultimately informing proactive well placement decisions that reduce drilling risks and operational costs.