SAGE Record 101, Yazid and Gaci
Yazid, H., and S. Gaci, 2022, Machine learning-based prediction of rock typing from well logs: SAGE Record 102, 1 p. + supplemental video, <http://sagetech.org/sage_record_102_yazid_and_gaci/>. Oral presentation at SAGE/ESSL BIGEC 2022, 30 Aug.–01 Sept. 2022, Benghazi, Libya, and Online.
Machine Learning-Based Prediction of Rock-Typing from Well Logs
Hasna Yazid (Department of Computer Science, M’Hamed Bougara University of Boumerdès [UMBB], Boumerdès, Algeria) and Said Gaci (Sonatrach–Geology, Geophysics & Reservoir Department, Algerian Petroleum Institute [IAP], Boumerdès, Algeria)
Well cores have a significant role to play in understanding the petrophysical characteristics of geological formations because they can be used to identify reservoir depth intervals with a high potential to contain hydrocarbons and presenting a suitable permeability to allow the extraction of hydrocarbons with a satisfactory flow. Due to the spatial heterogeneity of petrophysical parameters, only one core can be used per area. As a result, more cores are needed, which necessitates a higher cost.
In this study, it has been suggested to forecast lithofacies from well log data using machine learning techniques, more specifically weak learners. Applications have been conducted on well log data recorded from ten (10) boreholes drilled in different locations. The obtained results show that the considered hybrid weak learners-based approach gives a very high accuracy even when applied on a limited amount of the available data.
