Realtime Lithology Interpretation Using Machine Learning Model

Summary Automatic lithology interpretation real-time supports the drilling operations and assists in identifying potential geological risks that can impact the drilling operation as soon as possible after the interval has been drilled. The project has developed a method for training machine learning models with additional layers of data. In addition to surface data model in our Tier 1, the models incorporate Logging While Drilling measurements. The geoscientists actively use the automatic interpretation to assist in decision making. There is growing attention regarding at-bit lithology interpretation from surface data that machine learning has enabled. It is a significant potential for early detection of formation changes, optimized and correct section TD setting in competent formations, input to geo-steering decisions, including timely geo-stopping. The project is continuously developing new features and methods aimed at enhancing the robustness and reliability of the interpretation. The abstract includes an example from a newly drilled well comparing the realtime at-bit lithology interpretation based on surface data, with the quality checked manual interpretation utilizing all available data.