Micro-CT analysis of porous rocks and transport prediction

molassesitalianΤεχνίτη Νοημοσύνη και Ρομποτική

6 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

70 εμφανίσεις

Micro
-
CT analysis of porous
rocks and transport prediction

Hu Dong and Martin Blunt

Department of Earth Science and Engineering

Imperial College London


To generate the Pore Network, we
need …


The rock samples (11 samples)


Image
acquisition

and image processing


Image analysis based on maximal ball
algorithm


Results’ verification (compare with
experiments’ results, LBM simulation)

Samples we have



Image
acquisition



MicroCT scanner


Image
acquisition


Specimen preparation


Image
acquisition

Digital Detector

16 Bit Resolution

512 x 512 (or Virtual 1024 x 512)

Pixel Size 0.4 mm x 0.4 mm

X
-
Ray Tube

770 mm

Image processing


Segmenting



Due to the memory size
limitation and some side
effects of the image
itself, we currently use a
128
3

cubic image for
analysis.


Image processing


Thresholding


The original image from
scanning is gray scale.


The thresholded image is
only represented by 0
(void) and 1(grain).

Image processing


Eliminate the small
holes and small grains
in the image.

µ
-
CT
images …


Fontainebleau Sandstone


resolution: 7.5
µm

µ
-
CT
images …


Berea Sandstone


resolution: 10
µm

µ
-
CT
images …


Saudi Aramco SAMPLE1


resolution: 8.683
µm

µ
-
CT
images …


Shell CARBONATE1


resolution: 5.345
µm

µ
-
CT
images …


Saudi Aramco SAMPLE4


resolution: 8.96
µm

µ
-
CT
images …


Saudi Aramco SAMPLE3


resolution: 9.1
µm

µ
-
CT
images …


Saudi Aramco SAMPLE2


resolution: 11.497
µm

Network Extraction


Maximal ball algorithm

(SPE84296
-

Silin and Patzek)


Maximal ball

In the 3D image, from a voxel
(voxel [i, j, k]=0) in the void
space, the radius is increase by
one step until the ball hits a
solid phase voxel(1). We call
the ball a maximal ball at voxel
[i, j,k].

Network Extraction


Building the hierarchy

After finding all the maximal balls, we
compare them to build the
hierarchy. If two balls are
overlapped, the bigger one is the
smaller’s master, and recognize
the smaller a slave.

If a ball has no master, it is a
supermaster and defined as the
pore body; if a ball has no slaves,
it is a superslave and gives
information for minimum radius
of the throat.

Network Extraction

Maximal balls superimposed on MicroCT images. These
represent the pores.

Network Extraction


To build the skeleton of pore network:

Finished work:

a.
Find all the effective Maximal Balls to fill in the void
space in the image and build the hierarchy.

b.
Calculate the distribution of pore size and the co
-
ordination number.

Ongoing work:


Configure the throats to connect the pores and get
the throat size distribution

Sample case

We did a test on a sandstone SAMPLE1.
The
core
-
plug we used is 38mm in diameter and
26.5 mm in length.
A cylinder drilled from the
sandstone that is 8 mm in diameter has been
scanned to get the 3D image and a set of
processing and analysis has been done to the
image.

The image we used for simulation is 128
3

voxels
which represents a piece of rock of 1.1
3
mm
3
.

Coordination number distribution

0
2
4
6
8
10
12
14
16
18
20
1
3
5
7
9
11
13
15
17
19
21
Coordination Number
Percentage
Pore size distribution


0
20
40
60
80
100
120
140
160
180
0
1
2
3
4
5
6
7
Pore radius (voxels)
Frequency
Combination to Berea network


Our network code at present does not output a
full network. To predict relative permeability we
took a network based on Berea sandstone and
adjusted the pore size distribution to match that
measured on the image. We preserved the spatial
locations and rank order of pore size. The
coordination number 4.2 of the network is close
to that estimated from the image (5.2).

Result comparison

Experiment

LBM

PoreNetwork

Porosity


15.1%

(Gas)

14.3%

(NMR)

13.2%

(effective)

13.2%

(effective)
15.1%

(total)

Permeability


906mD

Kx 1466mD

Ky 1489mD

Kz 1605mD

1423mD

Predicted oil flood relative
permeability (primary drainage)

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Predicted water flood relative
permeability

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Future Work

1.
Experiments


Traditional

experiments
;


MicroCT

scanning

(optimize

the

parameters

during

the

scanning

for

correction

and

calibration

to

get

high

quality

images)
;


Sample

preparation

(drill

the

sample

into

proper

size

and

shape

to

meet

the

requirement

of

scanning)


3
D

image

library

2.
Network

generation


Identify the throats correctly;


Use the skeleton from a thinning algorithm [W.B.
Lindquist] as a quality control.



Acknowledgement


Supervisor
:

Martin

J
.

Blunt



Members

of

Pore
-
Scale

Modelling

Group,

Mariela

Araujo
-
Fresky,

Carlos

A
.

Grattoni,

Stefano

Favretto,

Hiroshi

Okabe



Members

of

Imperial

College

Consortium

on

Pore
-
Scale

Modelling



(BHP,

ENI,

JOGMEC,

Saudi

Aramco,

Schlumberger,

Shell,

Statoil,

Total

)