GEOLOG has developed a proprietary workflow named AI-PetroMech that takes data from different depth-based data sets. The tests performed in different geological contexts have shown that the minimum number of wells for creating and training the Model ranges from 4 to 6 depending on the quality of the data.
Model Building and Validation – Offset Wells
AI-PetroMech was trained with 10 wells in the same basin across carbonate formations. These wells have a full range of petrophysical well logs provided by the client (Density, Sonic and Gamma logs), and the depth-based drilling parameters and the XRF data provided by GEOLOG.
Prediction of GeoMechanical Parameters – New Wells
AI-PetroMech requires depth-based drilling parameters provided by GEOLOG and gamma ray (Wireline or LWD or MWD) provided by the client. XRF can replace GR in case of data unavailability or tool failure.
Results: Poisson’s Ratio and Young’s Modulus
In the graphs the results of AI-PetroMech Poisson’s ratio and Young’s Modulus. In blue, the “Conventional approach”, in red, the parameter value predicted by GEOLOG. The derived values were closely aligned with the well-log derived data for the same well with errors consistently within 5% of log values derived by conventional logs.