Robotized petrophysics: Machine learning and thermal profiling for automated mapping of lithotypes in unconventionals
Abstract We present a method for predicting rock types. The method is based on continuous high-resolution thermal logging along full-size core samples and being applied for rocks from a major unconventional formation. The method utilizes spatial spectral decomposition and machine learning approaches allowing automatic classification of the core samples over lithological groups within an isolated stratigraphic depth interval of a wellbore. The core samples are basically classified to the particular lithotypes by means of spectral representation of profiles of thermal properties obtained by a modern contactless method.
Paper
Full text
Robotized petrophysics: Machine learning and thermal profiling for automated mapping of lithotypes in unconventionals
Semantic Scholar · Engineering · 2018
Abstract
Abstract We present a method for predicting rock types. The method is based on continuous high-resolution thermal logging along full-size core samples and being applied for rocks from a major unconventional formation. The method utilizes spatial spectral decomposition and machine learning approaches allowing automatic classification of the core samples over lithological groups within an isolated stratigraphic depth interval of a wellbore. The core samples are basically classified to the particular lithotypes by means of spectral representation of profiles of thermal properties obtained by a modern contactless method.