Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Snøhvit CO2 Storage Project FWP-FEW0174 Task 4
Laura Chiaramonte, Joshua A. White, Whitney Trainor-
Guitton and Yue Hao
Lawrence Livermore National Laboratory
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore
National Laboratory under Contract DE-AC52-07NA27344
U.S. Department of Energy
National Energy Technology Laboratory
Carbon Storage R&D Project Review Meeting
Developing the Technologies and
Infrastructure for CCS
August 20-22, 2013
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Benefit to Program
Project Goals & Objectives
Technical Status
Summary & Accomplishments
Appendix
Outline
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
The research project is focused on mechanical deformation in response to CO2 injection at Snøhvit
An understanding of hydromechanical interactions is essential for effective prediction and monitoring of reservoir performance
This program meets the Carbon Storage Program goal to support industry’s ability to predict CO2 storage capacity in geologic formations to within ±30 percent
Benefit to the Program
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
• The project goal is to understand hydromechanical
impacts of CO2 injection into a complex storage
reservoir:
• Study the formation/enhancement of migration pathways within
the reservoir
• Validation of results based on monitoring and characterization
data provide by Statoil
• This work can guide management and monitoring practices for
sub sea floor injections and complex geologic structures
Success is tied to ability to reproduce and predict
behavior given available monitoring and characterization
data, and provide useful guidance for the field operator
Project Overview:
Goals and Objectives
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
• Schedule was reset by sponsor to October 1st, FY2013,
due to contracting & data transfer delays
• First stage of project was completed:
• Discrete Fault Activation Analysis under Stress
Uncertainty
• Preliminary Hydromechanical Analysis – Reservoir
Pressure Response
• New data received on July 2013
Technical Status
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
• Pre-study completed
• Site characterization and geo-model completed
• Discrete fault activation & stress uncertainty
analysis complete
• Preliminary analysis of pressure response in
reservoir completed
Accomplishments to Date
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Snøhvit CO2 Project
Gas fields with a 5 – 8 % CO2
content, which needs to be
reduced before liquefaction
Separated CO2 was re-injected
into Tubåen Fm. at ~2600m
depth
Injection began in 2008, but in
2010 Statoil announced
storage capacity in Tubåen
was lower than expected.
Have since moved injection to
another formation Structural diagram of Hammerfest Basin
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Stratigraphy
Wennberg et al., 2008
Delta plain depositional
environment, with fluvial
distributary channels & some
marine-tidal influence
Highly variable sandstone facies,
interbedded with siltstones &
mudstones
Storage target: Tubåen Fm. ~2600 m depth.
45-130 m clastic wedge (over ~50 km)
Individual channels & subordinate shales
Porosity 1-16%, Permeability 130-880 mD
Caprock: Nordmela Fm.
Porosity ~13%, Permeability 1-23 mD
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Structural Configuration Top of Fuglen Fm. – depth map
Main Horst
Snøhvit segment
100m
(Wennberg et al., 2008)
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Structural complexity of the site raises
many interesting hydromechanical
questions
1. What is the role of the bounding faults at the site?
Are they reservoir seals or potential leakage pathways? Is there a risk of contaminating the producing gas?
2. Why was storage capacity lower than expected?
Is it a completely compartmentalized system? Is it a function of the depositional setting? What is the role of observed faults/fractures?
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Fault Stability Analysis: Coulomb Criteria considering thermo
poro-elasticity effects
Uncertainty Analysis using PSUADE (Problem Solving
environment for Uncertainty Analysis and Design Exploration
1.- What is the role of the bounding faults
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Stress Uncertainty
Up to 90 degrees variations in reported SHmax Azimuths
Base Case modeled as NS SHmax Azimuth Strike Slip regime
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Fault Stability Analysis indicates fairly
stable bounding faults (NS SHmax)
Fault traces color-coded by amount of extra pressure (Pcp)
necessary to initiate slip (Base Case scenario: SS environment with
NS SHmax direction)
Pcp
[MPa]
Injector
N
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Uncertainty Analysis - PSUADE
13 Parameters
1000 samples
produced with Latin
hypercube sampling
method
Variable BC Min Max Units
Sv 60.6 51.5 69.7 MPa
Shmin 43 38.6 47.2 MPa
SHmax 65 60.6 74.3 MPa
Pp 28 25.2 30.8 MPa
µ 0.6 0.35 0.85
C 0 0 5
α*dPp 0 0 10 MPa
v 0.25 0.25 0.35
T 95 85 105 °C
E 35 22 36 GPa
αT 1.5e-5 1e-6 1.5e-5 1/°C
Fault ang -85 -55 -90 °
SHmax Az 0 345 105 °
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
UQ Analysis indicates SHmax Az as main
uncertainty C
ritical P
ressure
for
Reactivation
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Faults ~ 25-35% less stable with EW SHmax
N-S SHmax
E-W SHmax
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Faults ~ 25-35% less stable with EW SHmax
N-S SHmax
Change in
scale
E-W SHmax
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
Refined Uncertainty Analysis – 12
variables (no SHmax Az)
NS SHmax EW SHmax
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
Stress tensor components, fault ang, µ, C, Pp and
ΔP indicated as the most influential parameters
Example: sensitivity indexes for Fault 10
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
2.- Why was storage capacity lower than
expected
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
• 4D seismic reveals distinct channels & vertical stratification
• Lower perforation taking ~80% of the injection
Figure: 4D difference amplitude maps, lower perforation, from (Hansen et al, 2012). Left: 2003-2009, Right: 2009-2011.
Previous Analysis (Hansen et al. 2012)
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Previous Analysis (Hansen et al. 2012)
• Previous falloff analyses suggested flow barriers at 110m,
110m, and 3000m from injector
• PVT challenges encountered using gauge ~850m above
reservoir (2009 data)
6 Author name / Energy Procedia 00 (2011) 000–000
properties near the well were modeled using log and core data while these properties were scaled away
from the well to match to the observed data.
In addition to the faults clearly visible from baseline data, some possible barriers in the vicinity of the
injector came into focus after studying post injection 4D seismic. The flow barriers depict the simulated
bottom-hole pressure versus measured data for the best match scenario that includes the modification
mentioned above. The mismatch seen in 2008, between measured bottom-hole pressures (points) and
solid line (model) is due to near well-bore salt precipitation and reduced injectivity in the well. This was
eventually solved by MEG injection. The match to seismic data is also acceptable for most layers.
The dynamic model match indicates; lower than expected permeability for all Tubåen layers, lack of
vertical communication in Tubåen, no communication across major faults, and possible extra barriers near
the well. Consequently, dynamic simulation results are in general agreement with other observations
indicating that F-2H is injecting inside a compartment with acceptable reservoir properties but with
reduced communication to the rest of the reservoir system. However, other geological models and
concepts may also match the pressure time series, Figure 3.
1E-4 1E-3 0.01 0.1 1 10 100 1000Time [hr]
1000
10000
1E+5
Ga
s p
ote
nti
al
[ba
r2/c
p]
F-2H - Model matched to PLT data (ref)
F-2H - Model matched to FO 2009
Figure 3 Best match between measured bottom-hole (crosses) and modeled pressure. Timing of the acquired seismic 4D surveys are
indicated, as well as the estimated reservoir formation fracture pressure. b) Log-log plot of (2011 PLT and 2009 FO) gas pseudo
pressure with corresponding derivative. Models shown as solid line, measured data as points.
6. Fall-off analysis
Injection tests and fall-off (FO) analysis are good tools to investigate reservoir properties, both near the
wellbore and at larger scale. On a regular basis, the well has been shut in for only a few minutes, to
estimate the reservoir pressure and evaluate potential skin development. These tests have been made short
to neglect temperature effects and are used to establish the reservoir pressure based on the installed
gauges in the well. The estimated reservoir pressures are shown in Figure 3, and were subsequently
confirmed by pressures measured by the PLT in 2011 within a few bars. The start of the new LNG plant
at Melkøya had initial production challenges, and some caused shut-down of the full production facility,
including the CO2 injection. In particular, the 3 months shut-down in 2009 has been interesting and will
be discussed in detail. In April 2011 a PLT was run in the injector well, including a FO with for the first
time a pressure gauge at the perforations during the FO.
Figure 3b shows the log-log pressure series from the FO in 2009 (down-hole pressure gauge) and
during the PLT (sand face pressure gauge) in 2011. The shallow location of the down-hole pressure gauge
4D
Fracture pressure
4D 4D
PLT 2011 Fall-offAug 2009 Fall-off
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Is this a closed reservoir? Does rate, pressure &
temperature history imply changes in injection
behavior?
Examine entire rate, pressure, and temperature history from
the gauge at 1782 mTVDss
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Approach: Superposition Analysis
• Multi-rate injections are
difficult to analyze.
• Can often use the
principle of
superposition to simplify
the analysis (single-
phase approximation).
• Given pressure and rate
history, we solve for a
“characteristic” pressure
curve (as a linear least
squares problem).
p(t) = q × pC(t)
p(t) = (i
å qi+1 - qi ) × pC (t - ti )
Single rate:
Multi-rate:
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Thermal Correction
0.36DT
6 Author name / Energy Procedia 00 (2011) 000–000
properties near the well were modeled using log and core data while these properties were scaled away
from the well to match to the observed data.
In addition to the faults clearly visible from baseline data, some possible barriers in the vicinity of the
injector came into focus after studying post injection 4D seismic. The flow barriers depict the simulated
bottom-hole pressure versus measured data for the best match scenario that includes the modification
mentioned above. The mismatch seen in 2008, between measured bottom-hole pressures (points) and
solid line (model) is due to near well-bore salt precipitation and reduced injectivity in the well. This was
eventually solved by MEG injection. The match to seismic data is also acceptable for most layers.
The dynamic model match indicates; lower than expected permeability for all Tubåen layers, lack of
vertical communication in Tubåen, no communication across major faults, and possible extra barriers near
the well. Consequently, dynamic simulation results are in general agreement with other observations
indicating that F-2H is injecting inside a compartment with acceptable reservoir properties but with
reduced communication to the rest of the reservoir system. However, other geological models and
concepts may also match the pressure time series, Figure 3.
1E-4 1E-3 0.01 0.1 1 10 100 1000Time [hr]
1000
10000
1E+5
Ga
s p
ote
nti
al
[ba
r2/c
p]
F-2H - Model matched to PLT data (ref)
F-2H - Model matched to FO 2009
Figure 3 Best match between measured bottom-hole (crosses) and modeled pressure. Timing of the acquired seismic 4D surveys are
indicated, as well as the estimated reservoir formation fracture pressure. b) Log-log plot of (2011 PLT and 2009 FO) gas pseudo
pressure with corresponding derivative. Models shown as solid line, measured data as points.
6. Fall-off analysis
Injection tests and fall-off (FO) analysis are good tools to investigate reservoir properties, both near the
wellbore and at larger scale. On a regular basis, the well has been shut in for only a few minutes, to
estimate the reservoir pressure and evaluate potential skin development. These tests have been made short
to neglect temperature effects and are used to establish the reservoir pressure based on the installed
gauges in the well. The estimated reservoir pressures are shown in Figure 3, and were subsequently
confirmed by pressures measured by the PLT in 2011 within a few bars. The start of the new LNG plant
at Melkøya had initial production challenges, and some caused shut-down of the full production facility,
including the CO2 injection. In particular, the 3 months shut-down in 2009 has been interesting and will
be discussed in detail. In April 2011 a PLT was run in the injector well, including a FO with for the first
time a pressure gauge at the perforations during the FO.
Figure 3b shows the log-log pressure series from the FO in 2009 (down-hole pressure gauge) and
during the PLT (sand face pressure gauge) in 2011. The shallow location of the down-hole pressure gauge
4D
Fracture pressure
4D 4D
PLT 2011 Fall-offAug 2009 Fall-off
… the gauge data
becomes consistent with
PLT observations.
• Crude estimate gives
• Adding simple
thermal correction
….
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Best-fit Results • All data used for calibration, except early salt-precipitation
period
• Fit with one pC(t) curve
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Best-fit Results
Available data constrains the shape of this curve out to 779
days (the calibration period).
• Resulting pC(t) represents an equivalent constant-
rate injection.
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Barrier indications in the 2009 falloff Log-log plot of the 2009 falloff (real pressure)
6 Author name / Energy Procedia 00 (2011) 000–000
properties near the well were modeled using log and core data while these properties were scaled away
from the well to match to the observed data.
In addition to the faults clearly visible from baseline data, some possible barriers in the vicinity of the
injector came into focus after studying post injection 4D seismic. The flow barriers depict the simulated
bottom-hole pressure versus measured data for the best match scenario that includes the modification
mentioned above. The mismatch seen in 2008, between measured bottom-hole pressures (points) and
solid line (model) is due to near well-bore salt precipitation and reduced injectivity in the well. This was
eventually solved by MEG injection. The match to seismic data is also acceptable for most layers.
The dynamic model match indicates; lower than expected permeability for all Tubåen layers, lack of
vertical communication in Tubåen, no communication across major faults, and possible extra barriers near
the well. Consequently, dynamic simulation results are in general agreement with other observations
indicating that F-2H is injecting inside a compartment with acceptable reservoir properties but with
reduced communication to the rest of the reservoir system. However, other geological models and
concepts may also match the pressure time series, Figure 3.
1E-4 1E-3 0.01 0.1 1 10 100 1000Time [hr]
1000
10000
1E+5
Ga
s p
ote
nti
al
[ba
r2/c
p]
F-2H - Model matched to PLT data (ref)
F-2H - Model matched to FO 2009
Figure 3 Best match between measured bottom-hole (crosses) and modeled pressure. Timing of the acquired seismic 4D surveys are
indicated, as well as the estimated reservoir formation fracture pressure. b) Log-log plot of (2011 PLT and 2009 FO) gas pseudo
pressure with corresponding derivative. Models shown as solid line, measured data as points.
6. Fall-off analysis
Injection tests and fall-off (FO) analysis are good tools to investigate reservoir properties, both near the
wellbore and at larger scale. On a regular basis, the well has been shut in for only a few minutes, to
estimate the reservoir pressure and evaluate potential skin development. These tests have been made short
to neglect temperature effects and are used to establish the reservoir pressure based on the installed
gauges in the well. The estimated reservoir pressures are shown in Figure 3, and were subsequently
confirmed by pressures measured by the PLT in 2011 within a few bars. The start of the new LNG plant
at Melkøya had initial production challenges, and some caused shut-down of the full production facility,
including the CO2 injection. In particular, the 3 months shut-down in 2009 has been interesting and will
be discussed in detail. In April 2011 a PLT was run in the injector well, including a FO with for the first
time a pressure gauge at the perforations during the FO.
Figure 3b shows the log-log pressure series from the FO in 2009 (down-hole pressure gauge) and
during the PLT (sand face pressure gauge) in 2011. The shallow location of the down-hole pressure gauge
4D
Fracture pressure
4D 4D
PLT 2011 Fall-offAug 2009 Fall-off
Falloff analyses from (Hansen et al, 2012)
• Superposition provides additional data beyond 2009 falloff period
(779 vs. 142 days).
• Multiple barriers appear early in the falloff history, but no strong
evidence of additional barriers appearing after ~100 hours.
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Observations from Pressure Analysis
• Reservoir does not exhibit significant changes in
injection behavior over time. No evidence of large
geomechanical or permeability changes.
• Reservoir does not appear completely closed, and had
not reached pseudo-steady state.
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
4D seismic analysis suggests stratigraphic
compartmentalization, can it also have a
structural component? 4D difference amplitude maps
Hansen et al, 2012
Hypothetical sub-seismic faults (Az = 335-355º) expected
“permeable” under NS SHmax
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Reservoir does not appear completely closed,
is it possible a local vertical migration at F10?
F10 expected “sealing” under NS SHmax, but “permeable”
with EW SHmax
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Strong stress uncertainties difficult predictions
Faults fairly stable under “most likely” stress state:
SS & NS SHmax. Caprock failure would happen
before fault reactivation. Under those conditions, it
is unlikely that a theoretical sub-seismic fault could
act as flow barrier
Faults are ~ 30% less stable with EW SHmax, where
several segments are close to critically stressed.
Fault reactivation could happen before caprock
failure if injection continues with risk of gas
contamination.
Summary
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Summary, cont.
• Superposition analysis provides a complement to standard falloff testing, allowing one to analyze multi-rate pressure data over long periods
• Reservoir does not exhibit significant changes in injection behavior over time. No evidence of large geomechanical or permeability changes over time
• Reservoir does not appear completely closed, and had not reached pseudo-steady state. New storage volume was still being accessed at end of injection
• Potential structural component in compartmentalization/fluid migration difficult to assess due to stress orientation uncertainty
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Acknowledgments
Dataset and funding provided by Statoil and the Snøhvit Production License partners
Phil Ringrose, Olav Hansen, Bamshad Nazarian for useful discussions and contributions
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. We acknowledge funding from the U.S. Department of Energy, Fossil Energy.
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT 34
Appendix
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Carbon Fuel Cycles (Roger Aines)
Carbon Management
(Susan Carroll)
LLNL Carbon Sequestration
Program
Task 1. Active Reservoir
Management
Task 2. In Salah
Task 3. China
Task 4. Snøhvit
Task 5. Carbonates
Technical Staff
Wolery
Buscheck Aines
McNab, Chiaramonte,
Ezzedine, Hao, Foxall Ramirez, White
Friedmann
Chiaramonte, White,
Hao, Trainor-Guitton
Carroll, Hao, Smith
Expertise
Experimental and Theoretical Geochemistry
Subsurface Hydrology
Computational Geomechanics
Seismology
Structural Geology
Organization Chart
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Gantt Chart
Task FY2012 FY2013 FY2014
4.0 Pre-study (complete)
4.1 Site characterization & geomodel
4.2 Coupled hydromechanical analysis
4.3 Geomechanical modeling
Forecasting fault failure
Caprock deformation & fracture
Complete
on schedule
milestone
Lawrence Livermore National Laboratory LLNL-PRES-642912-DRAFT
Journal Papers in Preparation:
Chiaramonte, L., White, J.A. and Trainor-Guitton, W, Effect of Stress Field
Uncertainty on Modeling Geomechanics and Seal Integrity for CO2 Storage
Sites, (in preparation)
White, J.A. and Chiaramonte, L., Pressure Analysis, (in preparaation)
Peer Reviewed Papers:
Chiaramonte, L., White J.A., Hao, Y., and Ringrose, P., 2013, Probabilistic Risk
Assessment of Mechanical Deformation due to CO2 Injection in a
Compartmentalized Reservoir, Proceedings of the 47th U.S. Rock Mechanics /
Geomechanics Symposium, San Francisco, CA, 23-26 June
Chiaramonte, L., White J.A., and Johnson, S., 2011, Preliminary geomechanical
analysis of CO2 injection at Snøhvit, Norway. Proceedings of the 45th U.S.
Rock Mechanics / Geomechanics Symposium, San Francisco, CA, 26-29 June
Bibliography