GeoDict Forum

Digital Rock Physics => Core Analysis => Topic started by: GeoDict User on September 29, 2021, 02:42:02 PM

Title: Compute abs. permeability of a gray value image stack of a structure
Post by: GeoDict User on September 29, 2021, 02:42:02 PM
Dear GeoDict-Team,

How can I do a flow simulation where every gray value in my scan
gets its own permeability? In my case, some of the gray values
also indicate solids,  where the permeability should be zero,
and other gray values indicate  completely empty voxels, where the permeability should be infinite.

Looking forward to your reply
Title: Re: Compute abs. permeability of a gray value image stack of a structure
Post by: linden on October 01, 2021, 11:22:31 AM
It is possible to start a Stokes-Brinkman or Naviers-Stokes-Brinkman flow simulation with SimpleFFT where each voxel can have a different permeability.
This can be done by providing the solver your own .gfr file. These *.gfr files are used to represent flow resistivities (i.e. viscosity / permeability) per voxel.
In addition, it is possible to use gray values of a CT scan and convert them with a user defined "gray value to permeability" map to get a permeability value for each voxel.

How to simulate Stokes-Brinkman flows with your own (*.gfr) flow resistivity file (manually):
How to simulate Stokes-Brinkman flows using your own "gray value to permeability" map values (automatically):
This can be done with a GeoPython script. Please find the script together with an example file in the attachment.
The script will convert the gray value image with a .csv file into a *.gfr file automatically and starts a flow simulation.
In addition, the script will consider only gray values where the corresponding material type is a porous material.
Solid materials will get zero permeability and pore materials will get an infinite permeability.

How to prepare the simulation:
The script performs the following operations:
This feature will be available using LIR in GeoDict 2022 too, but the script has to be adjusted accordingly.