/anthills

NMR pore typing using Gaussian Mixture Models for permeability prediction

Primary LanguageJupyter Notebook

NMR Pore typing using Bayesian Gaussian mixture models

A tool for fitting Gaussian distributions to NMR T2 distributions to allow pore typing in carbonates using Bayesian Gaussian mixture models. Since scikit-learn 0.18, BayesianGaussianMixture is much faster than the PyMC3 implementation

Instructions to use:

  1. Clone the repository
  2. In the same folder location as the repository, create a folder called input_files where the LAS files will go
  3. Each LAS file should have the TCMR curve and a T2 distribution. T2_MIN & T2_MAX should be 0.3 & 6000ms
  4. Run anthills_sklearn.py
  5. The outputs will be populated in a folder called anthills_Output
  6. The calculated permeability curve can be loaded to Techlog via a CSV file