Skip to content

This repository is created to document the codes and data used in the research project titled "Generalizable Permeability Prediction of Digital Porous Media via Novel Multi-scale 3D Convolutional Neural Network".

elmorsym1/Permeability-Prediction-Via-3D-CNN

Repository files navigation

Generalizable-Permeability-Prediction-Via-Novel-Multi-Scale-3D-CNN

Data:

The data includes sub-volumes of the following rocks,

  • Bentheimer Sandstone
  • Ketton Limestone
  • Berea Sandstone
  • Doddington Sandstone
  • Estaillades Limestone
  • Carbonate (C1)
  • Carbonate (C2)

The raw CT rock cores are obtained from the Imperial Colloge London portal.

The sub-volumes are simulated for absolute permeability using OpenFOAM and their results are summerized in the provided excel sheet having the following information,

  • Number of sub-samples = 65,248
  • Labels description:
    • casename = sub-sampling index per rock type sample

    • porosity = ratio of void fraction

    • eff_porosity = the connected porosity

    • rock_type =

               {
               
               1:Bentheimer Sandstone,
               
               2:Ketton Limestone,
               
               3:Berea Sandstone,
               
               4:Doddington Sandstone,
               
               5:Estaillades Limestone,
               
               6:Carbonate (C1),
               
               7:Carbonate (C2)
               
               }
      
    • AR = anisotropy ratio

    • DOA = degree of anisotropy

    • k = absolute permeability

This work has been published under the American Geophysical Union flagship journal: Water Rseourses Research.

Paper link: Generalizable Permeability Prediction of Digital Porous Media via a Novel Multi‐Scale 3D Convolutional Neural Network.

For more information, please contact the repository owner at: elmorsym777@gmail.com

About

This repository is created to document the codes and data used in the research project titled "Generalizable Permeability Prediction of Digital Porous Media via Novel Multi-scale 3D Convolutional Neural Network".

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages