Relative permeability curve

[gxw-ll]

Hi Prof,
I meet a question in the Predicting relative permeability in using openpnm.
I generated the matrix and pores using the quartet structure generating set method, extracted the pore network model using the SNOW2 algorithm, and used gas to invade water on the pore network model, resulting in a very strange relative permeability curve, the relative permeability of water is very small, where do I need to adjust the model？
Here is my relative permeability curve:

[jgostick]

There are many things that could be wrong. You should try breaking it into manageable chunks. Is your network extraction ok? Is it's Pc curve good? Can you get good rel perm curves with a cubic network? What are the differences between the extracted network and the cubic one? Etc.

[gxw-ll]

OK,，Thank you for your prompt response! It's very helpful for my current situation. I will try the directions you provided.

[gxw-ll]

Yes, based on your advice, I decided to modify the pore network in my simulation. Initially, I used a voxel size of 30 x 30 x 30, resulting in 149 pores. After the adjustment, I switched to a voxel size of 300 x 300 x 300, yielding a total of 48,872 pores. This modification proved to be instrumental as it allowed me to obtain the desired curve in my simulation results.Why does voxel size have such a significant impact?The following is a portion of my code and fig

# Matrix,pores and hydrates are generated using the quadruple structure generation set method im_data = loadmat('D:/python_script/TestRelativePermeablity/testPF1.mat', squeeze_me=True) im_data[im_data == 2] = 0 # Hydrate is equivalent to matrix, im_data contains 0,1,2. 0 as matrix, 1 as pore, and 2 as hydrate im = im_data['E'].astype(bool) #im = 300*300*300 ps.metrics.porosity(im) length_x, length_y, length_z = im.shape shape = [length_x, length_y, length_z] net = ps.networks.snow2(phases = im, voxel_size=1e-3) pn = op.io.network_from_porespy(net.network) pn.regenerate_models()

[jgostick]

You must be mixing up terminology because voxel size is a scalar, not 30x30x30. Is 300x300x300 your image image size? If so then 30x30x30 is much too small so that would be why your results were gibberish. 300-cubed is about the minimum image size for getting a good extraction.

[gxw-ll]

Yes, you're right,I made a mistake. Thank you for your prompt response.