Caprock compressibility, caprock permeability and faultspermeability and the consequences for pressure development
due to CO2 storage in Vedsted structure, Northern part ofDenmark
• Ernest N. Mbia (PhD student)• Ida L. Fabricius (Professor at DTU Civil engineering) • Peter Frykman (Reservoir geologist at GEUS)• Finn Dalhoff (Senior geologist at COWI)• Christian Bernstone (senior Senior Advisor at Vattenfall R&D)• Ann T. Sørensen (Project Manager at COWI)• Gillian Pickup (Institute of Petroleum Engineering, Heriot-watt Univeristy)• Carsten N. Møller (Senior reservoir engineer at GEUS)
12-05-2014 1
• Introduction
• Objectives
• Methodology
• Main results
• Conclusions
• Acknowledgements
2
Outline
12-05-2014 3
Objectives1. To charaterize petrophysical and rock physical properties of
caprocks through laboratory study. To find out how the physical properties are related.
2. To use the measured caprock compressibility and permeability as input parameters for reservoir simulations in order to investigate and evaluate their influence on verticaland horizontal pressure developement when large volumeof CO2 is stored in Gassum Formation.
3. To investigate the effects of fault permeability in verticalpressure propagation in the Vedsted site.
Publications
1. Equivalent pore radius and velocity of elastic waves in shale. Skjold Flank-1 Well, Danish North Sea (published 2013 in Petroleum science & engineering).
2. Permeability, compressibility and porosity of Jurassic shale from the Norwegian-Danish Basin (In press, Petroleum geoscience).
3. Caprock Compressibility and Permeability and the Consequences for Pressure Development in CO2 Storage sites (Published 2013 in JGGC).
4. Modelling of the pressure propagation due to CO2 injection and the effect of fault permeability in a case study of the Vedsted structure (Accepted in JGGC).
4
5
Methodology and main results
6
Quantification of caprock properties (Permeability, compressibility and porosity of Jurassic shale from the Norwegian-Danish Basin.
12-05-2014 7
Methodology and main results
16 cuttingssamples
11 cuttingssamples
15 cuttings+ Coresamples
12-05-2014 8
Core samplesDiagonal (D)
Horizontal (H)
Vertical (V)
12-05-2014 9
Unloading Fractures
Cuttings samples - Grain density (He)
2.64 – 2.76 g/cm3
- Specific surface (BET)16 – 46 m2/g
- Pore size (MICP)̴ 10 nm range
10
- Mineralogy (XRD)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
109011151130114012221255135014451515158516751740174520052035204014751481148314861419142014211422142314281429152215271530157630513200335335203658381039594115427044204572
Mineralogical composition of shale
Dep
th [m
]
Quartz K-Feldspar Plagioclase Calcite DolomitePyrite Smectite Illite Kaolinite Chlorite
Børglum FmVedsted‐1
Haldager Fm
Fjerritslev Fm
Gassum FmVedsted‐1
Fjerritslev FmStenlille‐2
Fjerritslev FmStenlille‐5
Fjerritslev FmSkjold Flank‐1
Flyvbjerg Fm Vedsted‐1
Veds
ted-
1S
tenl
ille-
5S
kjol
d Fl
ank-
1
Stenlille-2
Porosity measurementsMercury injection capillary pressure (MICP)
Nuclear magneticresonance (NMR)
Helium porosimetry-mercury immersion (HPMI)
11
St.2
St.5
Vedsted-1
Skjold Flank-1
Porosity- HPMI porosity is comparablewith NMR while MICP measured the lowestporosity value.
12
0
1
2
3
4
0 5 10 15 20 25 30
Stra
in [%
]
Uniaxial stress [MPa]
St.2VSt.2DSt.2H
0.1
1
10
100
14 15 16 17 18 19 20M
-1x
10-4
[MPa
-1]
Uniaxial stress [MPa]
Elastic data_VElastic data_DElastic data_HSt.2VSt.2DSt.2H
12
Range of in situ stress
Elastic range
Elastoplastic range
(a) (b)
Compressibility
13
Hoek cell, uniaxial strain
Ultrasonic data
Loading
unloading
14
Velocity data
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 5 10 15 20 25 30
Vel
ocity
[km
/s]
Uniaxial stress [MPa]
Vp_St.2_V Vs_St.2_VVp_St.2_D Vs_St.2_DVp_St.2_H Vs_St.2_H
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 5 10 15 20 25 30V
eloc
ity [k
m/s
]
Uniaxial stress [MPa]
Vp_St.5_V Vs_St.5_VVp_St.5_D Vs_St.5_DVp_St.5_H Vs_St.5_H
St.2
(b)(a)
St.5HDV
HDV
HDV
HDV
Vp
Vs
Vp
Vs
Compressibility- Ultrasonic velocity data correspond with uniaxial stress strain data at the
beginning of the unloading stress path.
- The saturated and dry velocity (Brown-Korringa) data show only small difference
15
12-05-2014 16
PermeabilityMeasuring permeability
- Constant rate of strain experiments (Wissa et al. 1971)
where α is Biot’s coefficient from uniaxial confined experiment (Alam et al. 2011)
α = 1- (∂Ԑ/∂P)σ´/(∂Ԑ/∂ σ´) P
k = rH2μα/2Ppe
k =Permeabilityr = Strain rateH = Sample heightμ = Dynamic viscosityα = Biot’s coefficientPpe = Excess pore pressure
∂Ԑ = change in strain∂P = change in pore pressure∂ σ´ = change in net or differentialpressure
Permeability
Modelled permeability
1. Specific surface (Kozeny 1927) k = c(3/Sg2(1- )2)
2. Combined NMR and MICP data (Hossain et al., 2011)
3. Elastic data (Mbia et al., 2013) kM = c(r2PM/4)
kG = c(r2PG/4)
kK = c(r2PK/4)
4. MICP and clay fraction data (Yang and Aplin 2007)
k = 10-19.21Jv1.118 ř1.074
17
c = Kozeny’s factor= porositySg = Specific surface of grainρe = effective surface relaxivityfi = a fraction of total amplitude of each T2rpM = Equivalent pore radius from compressional modulusrpG, = Equivalent pore radius from shear modulusrpK = Equivalent pore radius from bulk
modulusJv =ř = average pore throat radiusJv = 9/8 (sin(α))2) J1
3/(1+ J1+ J12)
a = 450 – 10.240(n100 - n) e100 = 0.3024 + 1.687clay + 1.951clay2
J1 = the ratio of the largest radius of a pore to its throat radius, n = void ratio
18
Permeability
0.01
0.1
1
10
10 15 20 25 30
k [µ
D]
Porosity [%]
Specific surfaceNMRElastic dataMeasured
0.01
0.1
1
10
k_BET k_velocity k_NMR k_CRS
St.2 St.5
Modelled permeabilities are withinone order of magnitude
19
Permeability
0.00001
0.001
0.1
10
1000
100000
0.00001 0.001 0.1 10 1000 100000
Mea
sure
d pe
rmea
bilit
y (µ
D)
Modelled permeability (µD)
Yang & Aplin model - Smectite
Yang & Aplin model -Study data
Kozeny's model - Study data
Kozeny's model - Smectite
Kozeny's model - Kaolinite
Model type and dominant clay mineral
(a)
Daigle et al. 2011Yang and Aplin, 2007Dewhurst et al. 1999
Hursrud et al. 1998
Literature datata
Study data
Yang & Aplin Kozeny
12-05-2014 20
Conclusions
• Good correlation between measured and modelled permeability by Kozeny model for kaolinite rich clay and by Yang and Aplinrelation for smectite rich clay.
• Shale porosity depends on methodNMR ≥ HPMI > MICP porosity
• Compressibility from velocity data corresponds with that from stress strain data at the beginning of unloading stress path.
• Deep shale caprocks are stiffer than previously assumed for reservoir modelling studies.
Methodology and main results
Caprock Compressibility and Permeability and the Consequences for Pressure Development in CO2 Storage sites
21
Methodology and main results
22
Model setupand parameters
Formation Thickness Base & standard Permeability kh kv/kh Porosity
Compressibility Measured Range
(m) x 10‐5 (bar‐1) (µD) (µD) (%)
Post Chalk 30 4.5 5 x 103 0.1 23
Chalk 420 4.5 2 x 103 0.1 25
Vedsted 390 4.5 15 x 103 0.1 21
Frederickshavn (shale) 230 0.5 & 4.5 1 1x103 – 1x10-2 0.1 13
Børglum (shale) 50 0.5 & 4.5 1 1x103 – 1x10-2 0.1 13
Flyvbjerg (shale) 20 0.5 & 4.5 1 1x103 – 1x10-2 0.1 20
Haldager sand 80 4.5 267 x 103 0.1 17
Top Fjerritslev (shale) 174 0.5 & 4.5 1 1x103 – 1x10-2 0.1 11
Middle Fjerritslev (shale) 174 0.5 & 4.5 1 1x103 –1x10-2 0.1 11
Base Fjerritslev (shale) 174 0.5 & 4.5 1 1x103 – 1x10-2 0.1 10
Top Gassum (sandstone) 64 4.5 63 x 103 0.1 19
Middle Gassum (shale) 47 0.5 & 4.5 1 1x103 – 1x10-2 0.1 9
Middle Gassum (shale) 47 0.5 & 4.5 1 1x103 – 1x10-2 0.1 9
Middle Gassum (shale) 47 0.5 & 4.5 1 1x103 – 1x10-2 0.1 9
Base Gassum (sandstone) 85 4.5 70 x 103 0.1 14
Skagerrak (sandstone) 331 4.5 20 x 103 0.1 14
1 µD = 1 x 10-18 m2 1 bar = 1 x 105 Pa
Methodology and main results
23
-The temperature at 1875 m (mid Gassum level) is estimated at 66 0C
-The PVT data including the formation volume factor, density and viscosity for the temperature of 66 0C for the Gassum reservoir are obtained from the commercial PVT software PVTsim (Calsep 2001)
-Brine is saturated (25% salinity) which leads to a brine density of 1162.2 kg/m3.
-Gassum datum pressure of 196 bar
-Eclipse 100 is a black-oil simulator
-Boundary conditions for the site model have been modified to accommodate some of the lateral pressure transmission by using pore volume multipliers (factor 200) for the outmost cells.
-The capillary pressure curve for sandstone with 0.1 bar capillary entry and 6.5 bar for shale corresponding to a permeability level of around 0.3 µD correlation by Thomas et al. (1968).
-The relative permeability function data of Bennion & Bachu (2006)
-Simulating 60 Mt CO2 for 40 yrs at rate of 1.5 Mt/yr
Methodology and main results
24
CO2 plume and migration
Top reservoir layer
After 40 yrs ofinjection
Base case 0.5 x 10-5 bar-1
3.4
km
3.6 km
25
Methodology and main resultsCompressibility and pressure development in Vedsted structure 0
500
1000
1500
2000
2500
0 10 20 30 40 50
Dept
h [k
m]
Overpressure [bar]
Series1Series2
0.5 x 10-5 bar-1
4.5 x 10-5 bar-1
Caprock
0.5 x 10-5 bar-1
4.5 x 10-5 bar-1
Dept
h [m
]
260
10
20
30
40
50
0 10 20 30 40 50
Ove
rpre
ssur
e [b
ar]
Lateral distance along x-direction [km]
Series1Series2
0.5 x 10-5 bar-1
4.5 x 10-5 bar-1
Overpressure profile along x-direction in upper reservoir layer
5 bar0.5 x 10-5 bar-1
0.5 x 10-5 bar-1
27
Methodology and main resultsPermeability and pressure development in Vedsted structure
0
500
1000
1500
2000
2500
0 20 40 60
Dep
th [m
]
Overpressure [bar]
k=100 µDk=10 µDk=1 µDk=0.1 µDk=0.01 µDk=0.001 µD
Fjerritslev Formation (Caprock)
28
Methodology and main results
0
1
2
3
4
5
6
0 10 20 30 40 50
Ove
rpre
ssur
e [b
ar]
Lateral distance along x-direction [km]
Caprock k= 0.001 µD @ 40 yrsCaprock k= 0.1 µD @ 40 yrsCaprock k= 1.0 µD @ 40 yrsCaprock k= 10 µD @ 40 yrsCaprock k= 100 µD @ 40 yrs
Profile in the top layer of thecaprock
0
0.5
1
1.5
2
0 10 20 30 40 50
Ove
rpre
ssur
e [b
ar]
Lateral distance along x-direction [km]
Caprock k =100 µD @ 40 yrs
Profile at base layer of Chalk Group Permeability and pressure development in Vedsted structure
29
Methodology and main resultsPressure relaxation after injection stop
0
10
20
30
40
50
60
0 10 20 30 40 50
Ove
rpre
sure
[bar
]
Lateral distance along x-direction [km]
40 Yrs45 Yrs50 Yrs60 Yrs140 Yrs
30
Methodology and main resultsPermeability anisotropy
0
2
4
6
8
10
12
0 10 20 30 40 50
Ove
rpre
ssur
e [b
ar]
Lateral distance along x-direction [km]
Kv/Kh = 0.1
Kv/Kh = 0.02
Top reservoir Kv/kh = 0.1
Kv/kh = 0.02
Base caprock
31
ConclusionsSensitivity of caprock compressibility to pressure buildup and propagation
- We got 5 bar overpressure with the measured caprock compressibilitythan the standard caprock compressibility normally used in reservoir simulation studies.
- This pressure difference can also play a significant role in the presenceof permeable fractures and faults.
- Therefore well-designed investigations of formation properties arerecommended when carrying out reservoir simulation studies in order to minimize therisk of underestimating or overestimating pressure buildup in CO2 storage sites.
Sensitivity of caprock permeability to pressure buildup and propagation
-Increasing permeability from 0.1 μD to 1.0 μD, no pressure transmission via the 530 m thick caprock
- increasing further the permeability to 10 and 100 μD, overpressure is transmitted through the caprock and up tothe Chalk Group.
- Reducing the caprock permeability by one or two orders of magnitude further reduces the vertical pressurebuildup but increases lateral pressure buildup and the extent within the storage formation.
- Permeability anisotropy
32
Methodology and main results
Modelling of the pressure propagation dueto CO2 injection and the effect of faultpermeability in a case study of the Vedstedstructure, Northern Denmark
33
Methodology and main results
Vedsted site
34
Methodology and main results
35
Methodology and main results
Fault permeability range (mD) Locality/Fault rock type Reference
9.0 – 1587 Crotone Basin, South Italy (Sandstone) Balsamo and Storti 2011
8.0 – 145 Faulted siliciclastic aquifer in Central Texas Nieto et al. 2012
1.0 x 10-2 – 1.0 x 102 Arbuckle reservoir in Kansas (Sandstone) Franseen et al. 2003
5 x 10-2 – 1.0 Restefond fault in Alpline foreland (Highly deformed sandstone lenses) Leclère et al. 2012
(0.1–200 ) x 10-3 Middle Jurassic sandstone reservoirs in North Sea Fisher and Knipe, 2001
Fault permeability derived from different methods and materials, and from different scales as described in the references.
36
Methodology and main results
37
Methodology and main results
38
Methodology and main results
39
Methodology and main results
0
1
2
3
4
5
6
0 10 20 30 40 50
Ove
rpre
ssur
e [b
ar]
Lateral distance along x-direction [km]
Fault k = 0.001 mDFault k = 0.01 mDFault k = 0.1 mDFault k = 1 mDFault k = 10 mDFault k = 100 mDFault k = 1000 mD
Profile at the base of Chalk Group
40
Methodology and main results
0
1
2
3
4
5
6
0 10 20 30 40 50
Ove
rpre
ssur
e [ba
r]
Lateral distance along x-direction [km]
Fault k = 0.001 mDFault k = 1 mDFault k = 1000 mD Fault k = 0.001 mDFault k = 1 mDFault k = 1000 mD
kv/kh= 0.1kv/kh= 0.002
Base layer of the Chalk Group
(c)
41
Methodology and main resultsGridding
42
Conclusions• Literature data on fault permeability were gathered supplying an upper and a lower range of permeabilities. The simulation results showed that by changing fault permeability from 1000 mD, which represents the worst case scenario, pressure buildup is transmitted to the base Chalk Group with about 5.0 bar overpressure.
• We used other fault permeability values (100, 10, 1, 0.1, and 0.001 mD) which span the range from the worst to the best case scenario and the results showed that between 0.5 and 5.0 bar overpressure is transmitted to the base Chalk Group.
• We also briefly investigated the effect of permeability anisotropy, relaxation after the end of injection period and grid size on the CO2 migration and the pressure propagation. We found that there is no significant difference in the results when we use kv/kh of 0.1 or the value of 0.002.
• The maximum overpressure of 5.0 bar is seen in the base Chalk Group level and falls to about 1.2 bar 100 years after the end of the injection period.
• Fine grid resolution allows free CO2 gas to migrate slightly further in the lateral direction than in the coarse model.
12-05-2014 43
Acknowledgement
• This work was conducted as part of the CO2-GS project funded by the Danish Strategic Research Council and Vattenfall.