Supplementary document for EPAPS
Title: Pore network extraction from micro-CT images
Authors: Hu Dong and Martin J Blunt
Micro-CT images and pore network properties of investigated samples
The Maxima Ball algorithm was applied to all the samples studied in the paper to
extract pore networks. The cross-sections of the original datasets, the binarized micro-CT
images that were used to extract networks, the visualized network pictures, and the
distributions of coordination numbers, pore sizes and throat sizes of the networks are
presented in Figs. 21-44. The petrophysical properties calculated on both micro-CT
images and pore networks can be found in Table III in the main body of the paper.
For the coordination number distribution, it is found that some samples have large
fractions of pores with a coordination number of 0 (isolated pores) or 1 (dead ends),
which indicates that the image resolution can be insufficient to identify the throats
smaller than a voxel for the samples. Similarly, if a large fraction of pores and throats are
found with a size around the voxel size of the image, a better image resolution is required
to capture more detailed pore-space features. It is also found from the validations that a
resolution of a few microns is sufficient to image most of the sandstone rocks, but it
cannot resolve the features of carbonate rocks (sample C1 and C2) and some low
permeability sandstone (sample S4); this is evident by comparing the calculated absolute
permeability in Table III in the main body of the paper.
We find that most of the sandstone samples have an average coordination
numbers in the range 3.15-4.77 which is similar to that obtained from our two benchmark
sandstones, while sandstone sample S4 has a low average coordination number (2.72) due
to the poor connectivity caused by the insufficient resolution as discussed earlier. The
synthetic silica sample A1 with high porosity (42.9%) and high permeability (8076 mD)
has a high average coordination number (6.65) indicating its good connectivity. More
heterogeneous media – such as the carbonates C1 and C2 are characterized by relatively
low coordination numbers (3.0 and 2.37 respectively) because of a large number of
poorly connected pores. On the other hand, some heterogeneous samples, such as S7 and
S8, have high coordination numbers (5.23 and 5.94 respectively) because we have
relatively large pores connected to many throats. There is clearly a large disparity
between pore and throat size and the relatively open structure leads to lots of connections
in the pore space. Overall though, there is no simple way to relate qualitative, visual
features of the images to the computed coordination number distribution. Similarly, while
the typical pore radius can easily be inferred from the images, the segmentation into
pores and throats is sufficiently subtle to elude quantitative assessment from visual
inspection alone.
FIG. 1 (a) Cross-sections of the original micro-CT data (800×775 pixels); (b) image
used for network extraction (300×300 pixels) ; and (c) pore network for the
synthetic silica sample A1. The image resolution is 3.9 µm. The side length of the 3D
image and the pore network is 1170 µm.
FIG. 2 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample A1. The network porosity is 42.9% and the average coordination
number is 6.65.
FIG. 3 (a) Cross-sections of the original micro-CT data (850×850 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network for sandstone
S1. The image resolution is 8.7 µm. The side length of the 3D image and the pore
network is 2610 µm.
FIG. 4 Distributions of coordination number (top) and pore size and throat size
(bottom) for sample S1. The network porosity is 14.1% and the average
coordination number is 3.15.
FIG. 5 (a) Cross-sections of the original micro-CT data (950×950 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network for sandstone
S2. The image resolution is 5.0 µm. The side length of the 3D image and the pore
network is 1500 µm.
FIG. 6 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S2. The network porosity is 24.6% and the average coordination
number is 4.77.
FIG. 7 (a) Cross-sections of the original micro-CT data (875×900 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network of sandstone S3.
The image resolution is 9.1 µm. The side length of the 3D image and the pore
network is 2730 µm.
FIG. 8 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S3. The network porosity is 16.9% and the average coordination
number is 3.32.
FIG. 9 (a) Cross-sections of the original micro-CT data (550×550 pixels); (b) image
used for network extraction (300×300 pixels) ; and (c) pore network of sandstone S4.
The image resolution is 9.0 µm. The side length of the 3D image and the pore
network is 2700 µm.
FIG. 10 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S4. The network porosity is 17.1% and the average coordination
number is 2.72.
FIG. 11 (a) Cross-sections of the original micro-CT data (775×750 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network of sandstone S5.
The image resolution is 4.0 µm. The side length of the 3D image and the pore
network is 1200 µm.
FIG. 12 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S5. The network porosity is 21.1% and the average coordination
number is 3.32.
FIG. 13 (a) Cross-sections of the original micro-CT data (875×875 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network of sandstone S6.
The image resolution is 5.1 µm. The side length of the 3D image and the pore
network is 1530 µm.
FIG. 14 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S6. The network porosity is 24.0% and the average coordination
number is 4.0.
FIG. 15 (a) Cross-sections of the original micro-CT data (950×950 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network of sandstone S7.
The image resolution is 4.8 µm. The side length of the 3D image and the pore
network is 1440 µm.
FIG. 16 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S7. The network porosity is 25.1% and the average coordination
number is 5.23.
FIG. 17 (a) Cross-sections of the original micro-CT data (950×950 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network of sandstone S8.
The image resolution is 4.9 µm. The side length of the 3D image and the pore
network is 1470 µm.
FIG. 18 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S8. The network porosity is 34.0% and the average coordination
number is 5.94.
FIG. 19 (a) Cross-sections of the original micro-CT data (512×512 pixels); (b) image
used for network extraction (300×300 pixels); and (c) pore network of sandstone S9.
The image resolution is 3.4 µm. The side length of the 3D image and the pore
network is 1020 µm.
FIG. 20 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample S9. The network porosity is 22.2% and the average coordination
number is 3.32.
FIG. 21 (a) Cross-sections of the original micro-CT data (512×512 pixels); (b) image
used for network extraction (400×400 pixels); and (c) pore network of carbonate C1.
The image resolution is 2.9 µm. The side length of the 3D image and the pore
network is 1160 µm.
FIG. 22 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample C1. The network porosity is 23.3% and the average coordination
number is 3.0.
FIG. 23 (a) Cross-sections of the original micro-CT data (600×600 pixels); (b) image
used for network extraction (400×400 pixels); and (c) pore network of carbonate C2.
The image resolution is 5.3 µm. The side length of the 3D image and the pore
network is 2120 µm.
FIG. 24 Distributions of coordination number (top) and pore size and throat size
(bottom) of sample C2. The network porosity is 16.8% and the average coordination
number is 2.37.