Nanometer-scale Imaging and Evaluation of Marcellus Shale 09122-04 Tim Kneafsey Lawrence Berkeley National Laboratory Dmitriy Silin Shell Summary of Results from Completed GTI Marcellus Shale R&D Project May 7, 2013 Canonsburg, PA 1
rpsea.org
Acknowledgements
o LBNL gratefully acknowledges the financial support of RPSEA and the Gas Technology Institute in this project.
2
Outline
Imaging techniques at different scales
X-ray computed tomography (CT)
FIB/SEM
3
Diamond knife vs FIB (O 10-27 * reservoir)
Observations
Core scale (O 10-16 * reservoir) Microtomography (O 10-21 * reservoir)
Heterogeneity in all scales
Modeling
Considerations o Scale - How well does a 20 micron sample represent a kilometer+ scale region? o Sampling bias - How well does the imaged portion even represent the sample? o What physical changes have occurred in the sample prior to imaging? o Where are the fractures? o What is the size of a Representative Elementary Volume? •Can image –porespace and connectivity –mineral and organic phases •Imaging useful for building conceptual models
4
Considerations o Scale - How well does a 20 micron sample represent a kilometer+ scale region? o Sampling bias - How well does the imaged portion even Volume scanned / reservoir scale represent the sample? 20μm×10μm×5μm/(100km×100km×.1km) = (O)10-27 o What physical changes have occurred in the sample prior to imaging? Volume scanned/production scale o Where are the fractures? 20μm×10μm×5μm/(30m×30m×100m) = (O)10-20 o What is the size of a Representative Elementary Volume? •Can image –porespace and connectivity –mineral and organic phases •Imaging useful for building conceptual models
5
Considerations o Scale - How well does a 20 micron sample represent a kilometer+ scale region? o Sampling bias - How well does the imaged portion even represent the sample? o What physical changes have occurred in the sample prior to imaging? o Where are the fractures? o What is the size of a Representative Elementary Volume? •Can image –porespace and connectivity –mineral and organic phases •Imaging useful for building conceptual models
6
CT Image - Core Scale
Heterogeneity and fracture present Voxel size 195 x 195 x 1000 microns
Density (g/cm3) 2.3 cm
7
Marcellus shale: core-scale tomography Direction of view
6306 ft
4099 ft 8260 ft
8300 ft
4099 ft 8
Micro CT at the ALS
X-rays from bright, monochromatic low energy source pass through sample Parallel beam Micron-scale resolution 2×2×2 - 5×5×5 mm3 sample Non-destructive 3D imaging Can perform tests under relevant controlled pressure, temperature, stress (3-D triaxial) conditions
9
ALS Beamline 8.3.2
Very High Flux
10
typically 2-3 orders of magnitude greater than tube systems (better S/N) Faster acquisition Monochromatic, tunable (no beam hardening)
Micro CT
11
Microfractures
12
Micro CT Marcellus Shale 1.3 mm cube
13
Micro CT: comparison
NAS
Marcellus 14
Haynesville
Barnett
Focused IonBeam/Scanning Electron Microscopy
Zeiss CrossBeam 1540 EsB GEMINI field emission column (FESEM) Orsay Physic Ar-ion focused ion beam Maximum resolution of 1.1 nm at 20 kV and 2.5 nm at 1kV FIB column resolution 7-5 nm at 30 kV 6-axis fully eucentric motorized stage
SEM
FIB
Sample
15
SEM/FIB guns/detectors
Collingwood Shale
FIB/SEM: Reconstructing the 3D Structure
•
Pros: – Relatively simple sample preparation – Automation of milling/imaging sequence
•
16
Ref.: Tomutsa et al SPERE 2007 Silin and Kneafsey JCPT 2012
Cons – Not repeatable – Study area: tens of microns – Image segmentation is difficult
FIB/SEM: nanometer-scale resolution and study area
17
Shale samples from different regions
Collingwood
Barnett
8.9 microns
Utica
18
Haynesville
Marcellus shale
19
Micron-scale pores
20
Kerogen Inclusion?
21
~ 21 microns
Micro (nano?) cracks
22
23
FIB/SEM: not always a success story 17.3 microns
Almost zero visible porosity at resolution 11.56 nm 24
Diamond Knife / SEM
Chalk 25
Marcellus Shale Diamond knife cutting, SEM image: J. Mancuso Gatan, Inc
3D view at 50 nm resolution
50 microns 26
Common features of shale samples
Small pores – Complex geometry – Extremely low permeability
Variable presence of organic matter – Distribution – Relative volume – Free + adsorbed gas (?) 27
Gas flow model
Assumptions Finite stimulated reservoir volume Gas storage by compressibility and adsorption
28
Gas flow model
ρ0: Gas density in standard conditions Kerogen density Gas mobility SK Kerogen relative volume p: Gas pressure φ: porosity available for flow
Cf.: Lane, H., D. E. Lancaster, and A. Watson, 1990, Estimating Gas Desorption Parameters From Devonian Shale Well Test Data: SPE-21272 29
cf=f'(p): Adsorption isotherm slope
Dimensionless equations Dimensionless distance Dimensionless
pressure2
SRV size
u
2 2
Reservoir pressure
Dimensionless time
Dimensionless initial and boundary-value problem
Initial reservoir pressure Constant well pressure No-flow SRV boundary
30
Approximate exact formula: bimodal production rate decline
“Early” (t < t*) recovery rate
“Late” (t > t*) recovery rate
Method of integral relationships 31
Pirverdian (1953) Polubarinova-Kochina (1953) Barenblatt (1954)
Production rate decline curve
Data: http://www.rrc.state.tx.us/
32
Model validation
The actual production rates follow the trend estimated from truncated data 33
Accurate prediction of the total gas recovery
Gas storage
Adsorption storage factor - ASF
… if kerogen in reservoir conditions adsorbed methane like high-quality activated carbon in laboratory at 40 Bar, 300 K ...
ASF ~ O(1) 34
Pores in activated carbon
Kerogen relative volume
Summary and Conclusions
Pore-scale complexity of gas shale requires nanometer-scale resolution
Small pores Mineral diversity Variable presence of organic matter
Pore-scale analysis + limited data → physicsbased model of fractured gas well production
Approximate analytical solution Adsorption storage factor (ASF): Relative contributions of free and adsorbed gas storage Bimodal recovery rate decline
35
Early time: Square-root of time rate decline Dividing point: the stimulated reservoir volume boundary Late time: Exponential rate decline
Pyrite framboidal structures
36
After segmentation
The pyrite cluster shape is almost exactly spherical 37
Contacts
PI: Tim Kneafsey Lawrence Berkeley National Laboratory
[email protected] (510)486-4414
38
