Transforming Well Log Workflows: AI and Gen AI for Digitization, QC, Auto-Splicing, and Curve Prediction

This paper introduces an AI-native workflow designed to automate and enhance the processing of well log data, with a focus on digitization, quality control (QC), auto-splicing, and missing curve prediction. By leveraging deep learning-based OCR, image segmentation, anomaly detection, and advanced machine learning models—including LSTMs and transformers—the workflow streamlines traditionally manual tasks and ensures high data accuracy. Deployed across thousands of wells in the U.S. and India, the system achieved over 90% accuracy in raster log digitization, reduced manual work by 80%, and improved log QC and splicing efficiency. Missing curve prediction models delivered R² values above 0.92 for key petrophysical logs, significantly improving data availability for interpretation. This approach demonstrates how AI and Generative AI can modernize well log management, enabling scalable, accurate, and efficient subsurface data workflows.