Ml-based porosity modeling tested on synthetic and subsurface data
Master thesis
Permanent lenke
https://hdl.handle.net/11250/3023681Utgivelsesdato
2022Metadata
Vis full innførselSamlinger
- Studentoppgaver (TN-IER) [150]
Sammendrag
This thesis investigates if synthetic porosity models are useful as a basis for comparisonbetween machine learning (ML) approaches to porosity prediction. In addition to the MLmethods, the sequential gaussian simulation (SGS) geostatistical method is used as a bench-mark. The synthetic models are porosity and impedance cubes constructed from the F3 dataset(offshore Netherlands) well-logs, to mimic specific geological geometries including a sedimentarywedge and a normal fault. Based on the performance of the different methods on the syntheticmodels, a porosity prediction is performed on the actual F3 dataset as well. The predictionmethods discussed are SGS, and ML methods such as KNN-regression, lasso-regression, randomforest-regression, and shallow neural network. The geostatistical and geophysical methods arerun using Petrel, and the ML methods using Python. ML methods are better at minimizingthe error while missing much of the detail of the SGS method. However, for the F3 dataset,random forest appears to capture more details than the other methods. The synthetic modelsprovided a better basis for comparison of the different methods, however the workflow requiresimprovement.