Page 157 - Machine Learning for Subsurface Characterization

132 Machine learning for subsurface characterization

travel time logs (DTC and DTS) along with an indicator of the reliability of the log
synthesis. ANN model has the best prediction accuracy among the six regression
models for log synthesis. K-means clustering can generate cluster numbers that
positively correlate with ANN prediction accuracy. By combining the shallow
ANN model and the K-means clustering, we developed a prediction workflow
that can synthesize the compressional and shear travel-time logs and
simultaneously determine the reliability of the log synthesis. This study will
enable engineers, petrophysicists, geophysicists, and geoscientists to obtain
reliable and robust geomechanical characterization when sonic logging tool is
not available due to operational or financial constraints.

2 Methodology
2.1 Data preparation

Welllogsusedinthisstudywereacquiredfromtwowells.InWell1,welllogswere
measured at 8481 depths across 4240-ft depth interval. In Well 2, well logs were
measured at 2920 depths from 1460-ft depth interval. The 13 easy-to-acquire
conventional logs used for the proposed log synthesis include gamma ray log
(GR), caliper log (DCAL), density porosity log (DPHZ), neutron porosity log

FIG. 5.1 Track 1 is depth, Track 2 contains gamma ray and caliper logs, Track 3 contains density
porosity and neutron porosity logs, Track 4 contains formation photoelectric factor and bulk density
logs, Track 5 is laterolog resistivity logs at various depths of investigation (RLA0, RLA1,
RLA2, RLA3, RLA4, and RLA5), and Track 6 contains shear and compressional travel-time
logs for a 200-ft section of the 4240-ft depth interval in Well 1.