PROCEEDINGS, TOUGH Symposium 2003

Lawrence Berkeley National Laboratory, Berkeley, California, May 1214, 2003

TOUGH-FLAC: A NUMERICAL SIMULATOR FOR ANALYSIS OF COUPLED

THERMAL-HYDROLOGIC-MECHANICAL PROCESSES IN FRACTURED AND POROUS
GEOLOGICAL MEDIA UNDER MULTI-PHASE FLOW CONDITIONS

Jonny Rutqvist and Chin-Fu Tsang

Lawrence Berkeley National Laboratory

Earth Sciences Division
Berkeley, CA 945 30, USA
e-mail: [email protected]

ABSTRACT
EOS PVT functions for
This paper presents development of a simulator for Fluids water, air, CO2 etc
coupled thermal-hydrological-mechanical (THM)
processes in fractured porous geological media and
recent applications of the simulator related to coupled TOUGH2 Multiphase fluid and
Reservoir heat transport analysis
THM processes under multiphase flow conditions.
The simulator is denoted TOUGH-FLAC because it Simulator
utilizes two established computer codesTOUGH2
TOUGH-FLAC
and FLAC3Dwhich are linked and jointly executed P, T, S THM Loop , k, Pc Coupling Modules
for analysis of coupled thermal-hydrologic-mechani-
cal (THM) processes. Capabilities of the TOUGH-
FLAC3D Geomechanical Stress-
FLAC simulator are demonstrated on several Soil&Rock
complex coupled problems related to injection and strain analysis with HM
Mechanics and TM coupling
storage of carbon dioxide in aquifers and to disposal
of nuclear waste in unsaturated fractured porous Constitutive model for
media. Mechanical Model
visco-elastic, poro-
=H(
,T,
/
t
t)
elasto-plastic material
THE TOUGH-FLAC THM SIMULATOR
P = Pressure k = Permeability
The TOUGH-FLAC simulator (Rutqvist et al. 2002) T = Temperature Pc = Capillary pressure
is based on a coupling of the two existing computer S = Saturation H= Material function
codes TOUGH2 (Pruess et al. 1999) and FLAC3D = Stress = Strain
(Itasca Consulting Group 1997). TOUGH2 is a well- = Porosity t = time
established code for geohydrological analysis with
Figure 1. Schematic of linking TOUGH2 and
multiphase, multicomponent fluid flow and heat
FLAC3D for a coupled THM simulation.
transport, while FLAC3D is a widely used commer-
cial code that is designed for rock and soil mechan-
ics. For analysis of coupled THM problems, the
TOUGH2 and FLAC3D are executed on compatible TOUGH2 mid element node
numerical grids and linked through external coupling
modules, which serve to pass relevant information
between the field equations that are solved in respec-
tive code (Figure 1 and 2).

TOUGH-FLAC COUPLING MODULES

A TOUGH to FLAC link takes multiphase pressures,
saturation, and temperature from the TOUGH2
simulation and provides updated temperature, and FLAC3D corner node
pore pressure to FLAC3D (Figure 1). Because a
TOUGH2 mesh uses one gridpoint within each Figure 2. Coexisting TOUGH2 and FLAC3D mesh
element, and FLAC3D nodes are located in element for a coupled THM simulation.
corners, data have to be interpolated from mid-
element (TOUGH2) to corner locations (FLAC3D). After data transfer, FLAC3D internally calculates
thermal expansion and effective stress according to:

-1-
 -2-

T = I T (1) t ( ) k +1 (6)
Rn( ) k +1 M n( ) k +1 M n( ) k Anm q nm + Vn Qn
( ) k +1

= + IP (2) Vn m

where is thermal strain T is the linear thermal

T where subscripts n and m label element and element
face, R is the residual, M is mass or energy per unit
expansion coefficient, I is the unit tensor, T is
volume, V is element volume, A is area of element
temperature, is effective stress, is total stress, face, q is flux and Q is the source term.
is a Biot effective stress parameter (Biot, 1941) and P
During this time step, the porosity and permeability
is pore fluid pressure. In a multiphase flow calcula-
are assumed to be constant, according to the values
tion, the value of P transferred to FLAC3D could
evaluated at the beginning of a time step. The mass
represent an average pore pressure calculated from
term, M n( ) k in Equation (6), represents the condi-
the pressures of the various phases (Rutqvist et al.,
2002). tions at the end of the previous time step and should
therefore include the previous porosity . On the
k

An FLAC to TOUGH link takes element stress or ( ) k +1

other hand, the mass term M n and the flux
deformation from FLAC3D and corrects element ( ) k +1
should include the new porosity and the
k+1

porosity, permeability, and capillary pressure for q mn

k+1
TOUGH2 according to the following general expres- new permeability k , which were evaluated at the
sions: beginning of the time step. In addition, the phase
= (, ) (3) saturation, S, should be corrected for the imposed
change in porosity according to:
k = k (, ) (4)
k
Pc = Pc (, ) (5) S lk = S lk +1 (7)
k +1
No interpolation in space is required for this data S gk +1 = 1 S lk +1 (8)
transfer because stress and strain are defined in
FLAC3D elements, which are identical to TOUGH2 The explicit sequential solution should be accurate if
elements. A TOUGH-FLAC coupling module for this the porosity and permeability varies slowly with time
link should calculate the hydraulic property changes, or if time step size relatively small. For example,
based on material-specific theoretical or empirical when modeling coupled THM processes around a
functions. nuclear waste repository, it is expected that slowly
evolving thermal strains cause most changes in
TOUGH-FLAC NUMERICAL PROCEDURE porosity and permeability. On the other hand, when
modeling problems with significant strain rate and
A separate batch program can control the coupling
hydraulic squeezing, time steps must be limited for
and execution of TOUGH2 and FLAC3D for the
an accurate solution. When using the implicit
linked TOUGH-FLAC simulator. In this case, it was
sequential scheme there is no time limitations for
done within the FLAC3D input file using the FLAC-
accuracy, however, large time steps may induce
FISH programming language (Itasca Consulting
numerical stability problems.
Group 1997). The calculation is stepped forward in
time with the transient TH analysis in TOUGH2, and
at each time step or at the TOUGH2 Newton iteration Tk Tk Tk+1 Tk+1
level, a quasi-static mechanical analysis is conduced Pk Pk Pk+1 Pk+1
with FLAC3D to calculate stress-induced changes in k-1 k k k+1
porosity and intrinsic permeability. The resulting kk-1 kk kk kk+1 T
T
TOUGH2 P
THM analysis may be explicit sequential, meaning P
that the porosity and permeability is evaluated only at
P P
k k
the beginning of each time step, or the analysis may T T
Pc P
be implicit sequential, with permeability and porosity c


updated on the Newton iteration level towards the FLAC3D


end of the time step using an iterative process. , k k
, k+1
k+1

tk tk+1
In the explicit sequential procedure (Figure 3), the TIME
TOUGH2 code is executed for a TH analysis
k k+1
between time t to t until mass conservation is Figure 3. Numerical procedure of a linked
assured by solving the TOUGH2 flow and heat TOUGH2 and FLAC3D simulation with
equation: explicit sequential solutions
 -3-

APPLICATION OF TOUGH-FLAC
k0 0
Two recent application examples demonstrate the Pc = Pc 0 (S l ) (11)
capability of a linked TOUGH-FLAC simulation. k
The first example is related to sequestration of CO2
The parameters for the porosity-mean stress and
green-house gas in brine aquifers (Rutqvist and
Tsang, 2002), and the second is related to a high permeability-porosity relationship0, r, b, c in
temperature nuclear waste repository in unsaturated Equations (9) and (10)were determined to repre-
rock. Within each application, material specific sent laboratory data by Davis and Davis (1999),
hydromechanical coupling relationships are which shows a one-order-of-magnitude reduction in
developed and applied. permeability from zero to 30 MPa effective stress.

Figure 4b presents the calculated injection pressure

CO2 sequestration into a brine formation during a 15-year injection period, with or without
In this example, the TOUGH-FLAC code is applied consideration of the stress-dependent rock mass
to simulate an injection operation for disposal of CO2 permeability. The difference in injection pressure is
into a permeable brine formation, which is overlain explained by permeability in the injection zone (in
by a semi-permeable caprock. The problem domain is the former case) increasing because of a general
a two-dimensional cross-section according to Figure reduction in effective stresses. However, the changes
4a and the material properties are listed in Table 1. in permeability are moderate (less than a factor 2)
The injection takes place at a depth of 1,500 meters because of a rather insensitive stress-permeability
so that the CO2 is within the temperature and pressure relationship for the porous sandstone.
range for it to be a supercritical fluid. As a
supercritical fluid, the CO2 behaves like a gas with Calculated results after 10 years of injection are
low viscosity but having a liquid-like density of 200 presented in Figures 5 to 8. These results are selected
3
900 kg/m , depending on pressure and temperature. to show how a TOUGH-FLAC simulation can be
Because the supercritical CO2 is less dense than used to evaluate the caprock integrity at a CO2
water, deep underground disposal requires that the injection site.
caprock is sufficiently impermeable to trap the
injected CO2 for a sufficiently long time. Important
CO2
rock-mechanical aspects of this simulation are to
study the integrity of the caprock and the possibilities
of rock failure.
Cap Rock 1.5 km
In this application the isotropic hydraulic properties
are corrected using empirical porosity-mean stress
and a permeability-porosity relationship (Rutqvist Injection zone
and Tsang, 2002). The porosity, , is related to the
mean effective stress, M as
Large Lateral Extension
= r + ( 0 r )exp(a M ) (9) (a) Schematics of two dimensional model geometry

where 0 is porosity at zero stress, r is residual 40

INJECTION PRESSURE, P (MPa)

H Calculation with k = k0
porosity at high stress, and the exponent a should be
experimentally determined. 35 Lithostatic
pressure
The permeability is correlated to the porosity 30
according to the following exponential function HM Calculation with k = k(')
(Rutqvist and Tsang, 2002): 25

k = k 0 exp[c( 0 1)] (10) 20

Current study
where k0 is permeability at zero stress and the expo-
15
nent c should be experimentally determined. 0 5 10 1
TIME (Years)
In addition to the two coupling relationships in (b) Simulated evolution of injection pressure
Equations (9) and (10), the capillary pressure is
modified with permeability and porosity according to Figure 4. TOUGH-FLAC simulation of CO2 injection
a function by Leverett (1941): into a brine water formation.
 -4-

Table 1. Rock properties for simulation of

CO2 disposal in a brine formation Injection Pressure 33 MPa P (MPa)
500 5

Property Cap Aquifer Upper Aquifer

1000 10

DEPTH (m)
CO2
Youngs modulus, E (GPa) 5 5 Cap 25 20
Injection Zone
1500
Poissons ratio, (-) 0.25 0.25
Base
Saturated rock density, s (kg/m3) 2260 2260 2000
20

Zero stress porosity, 0 (-) 0.01 0.1

25
2500
Residual porosity, r (-) 0.009 0.09 -10000 -5000 0 5000 10000
DISTANCE FROM INJECTION POINT (m)
Zero stress permeability, k0 (m2) 110 -17
110 -13

Irreducible gas saturation Coreys (1954) 0.05 0.05 Figure 5. Calculated fluid pressure distribution and
relative permeability function (-)
the spread of the CO2 plume after 10
Irreducible liquid saturation for Coreys 0.3 0.3 years of injection.
(1954) relative permeability function (-)
Air-entry pressure for van Genuchtens 3100 19.6
(1980) retention curve (kPa)
Exponent, m for van Genuchtenss 0.457 0.457
(1980) retention curve
Increased Horizontal Stress at XX (MPa)
500 the Injection Point Caused by
Poro-elastic Effects
Exponent for Equation (9), a (1/Pa) 510-8 510-8
1000

DEPTH (m)
Exponent for Equation (10), c 22.2 22.2
Biots parameter in Equation (2), 1.0 1.0 1500
10 5
2000 0

The integrity of the caprock could be jeopardized by 2500

a mechanical failure, which could occur as a result of -10000 -5000 0 5000 10000
hydraulic fracturing or reactivation of pre-existing DISTANCE FROM INJECTION POINT (m)
shear fractures. These failure mechanisms are driven (a) Changes in horizontal stress
by changes in effective stresses, which in turn
depends on the in situ stresses and injection fluid Increased Vertical Stress at ZZ (MPa)
500 the Injection Point Caused by
pressure. Poro-elastic Effects

1000 0
DEPTH (m)

Figure 5 presents the spread of the CO2 fluid and the

fluid pressure within the aquifer/caprock system after 1500 10
5
10 years of injection. The figure shows that the CO2
2000
has spread under the cap over 4 km and has pene- 0
trated upwards into the caprock by about 10 meters. 2500
At this time the injection pressure has increased to a -10000 -5000 0 5000 10000
33 MPa, which is slightly less than the lithostatic DISTANCE FROM INJECTION POINT (m)
stress at the injection point (Figure 4b). (b) Changes in vertical stress
Figure 6 shows how the vertical and horizontal in situ Figure 6. Calculated changes in total in situ stresses
stresses (total stresses) increases near the injection after 10 years of injection.
point. The in situ stresses increases as a result of
poro-elastic stresses that occurs when the porous rock
attempts to expand in a confined rock mass. Both
vertical and horizontal stresses increases in the injec-
tion aquifer. However, in the cap-rock, just above the
injection interval, the horizontal stresses increases
much more than the vertical stresses.

Figure 7 presents the changes in vertical and hori-

zontal effective stress near the CO2 plume. The
figure shows that the effective stresses are reduced
mostly at the interface between the injection aquifer
and the caprock and the vertical effective stress is
reduced more than the horizontal one.
 -5-

1100 1100
Upper Aquifer 'xx (MPa) Upper Aquifer
1200 1200
Possible Fracturing
Cap 5 Cap

DEPTH (m)
DEPTH (m)

1300 1300
CO2+Brine Water
5
1400 1400 Injection Zone Brine Water
Injection Zone C02
1500
5 1500
Base
Base 5
1600
1600 -2000 -1000 0 1000 2000
-2000 -1000 0 1000 2000
DISTANCE FROM INJECTION POINT (m)
DISTANCE FROM INJECTION POINT (m)
(a) Hydraulic fracturing
(a) Changes in horizontal effective stress
1100
1100 Upper Aquifer
'zz (MPa)
Upper Aquifer
1200
1200
Zone of Possible Slip
Cap

DEPTH (m)
Cap 10 1300
DEPTH (m)

1300 CO2+Brine Water

1 5 10
1400 Brine Water
1400 5 Injection Zone
Injection Zone
1500
1500
15 Base
10
-2000 -1000 0 1000 2000
-2000 -1000 0 1000 2000 DISTANCE FROM INJECTION POINT (m)
DISTANCE FROM INJECTION POINT (m) (b) Shear slip of pre-existing fractures
(b) Changes in vertical effective stress
Figure 8. Zones where the mechanical failure could
Figure 7. Calculated Changes in effective stress in occur calculated at 10 years of injection.
the aquifer-cap rock system after 10 years
of CO2 injection.

Coupled THM analysis of a high temperature

Figure 8 shows that if mechanical failure occurs it heater test in semi-dry fractured rocks
would most likely initiates at the interface between In this example, TOUGH-FLAC is applied for a
the caprock and injection the zone. In this zone, the coupled THM simulation of the Yucca Mountain
reduction in vertical effective stress can lead to the Drift Scale Test (DST). The DST is a large-scale,
formation of horizontal hydraulic fractures (Figure long-term high-temperature thermal test designed to
8a). Furthermore, a larger zones of possible slip on investigate coupled thermal-mechanical-hydrologi-
pre-existing fractures occurs at the upper and lower cal-chemical behavior in an unsaturated, fractured,
part of the injection zone. This implies that an and welded tuff rock mass. The DST centers around a
unfavorably oriented fault could be reactivated with heated drift, which is 5 meters in diameter and has a
accompanying micro-seismicity and possible perme- 47.5 m long heated section. Heating is provided by
ability change. However, even if fracturing or fault nine drift heaters within the heated drift, as well as 50
reactivation would take place in the lower parts of the rod heaters, referred to as wing heaters, placed into
caprock, the TOUGH-FLAC simulation indicates that horizontal boreholes emanating from the heated drift
it would be contained within the lower portion of the (Figure 9a). The DST is simulated with TOUGH-
cap and would not propagate through the upper part FLAC in a two dimensional cross section oriented
of the cap. normal to the drift axis (Figure 9a). The highly
fractured rock mass at the test site is modeled as a
This application example demonstrates how the dual-permeability medium, which consists of inter-
specially developed capabilities of the TOUGH2 acting matrix and fracture continua.
code for the complex behavior of the supercritical
CO2 could be utilized in a coupled TOUGH-FLAC
simulation. These results contribute to identifying the
most critical mechanisms for the integrity and long-
term stability of a CO2 injection site. Further studies
include effects of a permeable fault (Rutqvist and
Tsang, 2002), heterogeneous reservoir rock, and
inelastic aquifer compaction behavior.
 -6-

In this study, the current fracture aperture b depends

Dual-Permeability on the current effective normal stress n, according
Heated Drift Medium: to the following exponential function (Rutqvist and
Matrix & Fracture Tsang, 2003):
z
Continua
x b = br + bm = br + bmax [exp (d n )] (16)

where br is a residual aperture, bm is mechanical

Drift Heater Wing Heaters aperture, bmax is the maximum mechanical aperture,
and d is a parameter related to the curvature of the
function (Figure 9c). This expression can be inserted
(a) into Equation (13) to derive expressions for rock
z mass permeability-correction factors in x, y and z
Fracture Aperture, b

directions
x y
z
x bmax Calculated results after 12 months of heating are
y presented in Figures 10 to 13. These results are
kx = kx(y ,z) 1/d selected to illustrate how TOUGH-FLAC can be
kx = kx(y ,z) br utilized to calculate THM-induced changes in air-
kx = kx(y ,z) Normal Stress, n permeability in fractured rocks. Changes in air-
(b) (c) permeability are particular important because they
are largely dependent on stress-induced changes in
Figure 9. TOUGH-FLAC simulation of the Yucca fracture permeability, which is a key process in
Mountain Drift Scale Test (a) Schematics coupled THM modeling of fractured rocks.
of two-dimensional model geometry. (b)
Conceptual model for stress-permeability Figure 10 shows that after 12 months of heating the
coupling. (c) Normal stress-aperture temperature has reached above the boiling tempera-
relationship for fractures. ture around the heated drift and near the wing
heaters. The high temperature induces strong
For the Yucca Mountain site, the correction of thermal-hydrological processes with evaporation of
hydraulic properties with stress are based on a liquid water near the heatsource (Figure 11). The
conceptual model of a highly fractured rock mass that evaporated water is transported as vapor away from
contains three orthogonal fracture sets as shown in the heat source toward cooler regions where it is
Figure 9b (Rutqvist and Tsang, 2003). The porosity condensed to liquid water (Figure 11). As a result, a
and permeability-correction factors are calculated dry-out zone is created near the heat source and a
from the initial and current apertures in Fracture Set condensation zone is progressively moving away
1, 2, and 3 according to: from the heat source. In the condensation zone, an
increase in fractures moisture content should result in
b1 + b2 + b3 (12)
F = a decrease air-permeability.
b1i + b2i + b3i
The high temperatures shown in Figure 10 gives rise
b 3 + b33 , b13 + b33 , b13 + b23 (13) to thermal expansion of the rock mass with associ-
Fkx = 32 Fky = F =
b13i + b33i
kz
b2i + b33i b13i + b23i ated thermal stresses (Figure 12). Near the heat
source, the horizontal compressive stresses increases
where fractures in Fracture Sets 1, 2, and 3 are strongly with a maximum at the drift wall and near
assumed to be equally spaced and oriented normal to the heat sources. Such increase in compressive
x, y, and z directions, respectively, and a parallel- stresses tends to tighten fractures to smaller aperture
plate fracture flow model (Witherspoon et al. 1980) is leading to a reduction in air-permeability. Away from
adopted. The capillary pressure is corrected with the heat source, the horizontal stresses decreases
porosity and permeability changes according to the slightly (Figure 12). A reduction in horizontal
Leverett (1941) function: stresses will tend to open pre-existing vertical
Fk
fractures to a larger aperture leading to an increase in
FPc = (14) air-permeability.
F
where
Fk = 3 Fkx Fky Fkz (15)
 -7-

Table 2. Rock properties for simulation of a heater

25
test in fractured welded tuff.
20
Property Value
Matrix Permeability 1.24E-17 m 2
15 Above Boiling
Hydraulic and Temperature
Thermal Porosity 0.11
10
Properties 40
van Genuchten, m 2.25E-6 Pa-1

Z (m)
5
van Genuchten mm (or ) 0.247 100
0
Residual saturation 0.18 160

Rock grain density, 2530 kg/m3 -5 100

Rock grain specific heat 953 J/kg K) Wing Heaters

-10
40
Dry thermal conductivity 1.67 W/m K Drift Heater
-15
Wet thermal conductivity 2.0 W/m K -25 -20 -15 -10 -5 0 5 10 15 20 25
X (m)
Fracture Permeability, k 1.00E-13 m 2

Hydraulic and
Hydro- Porosity 0.263E-3
Figure 10. Calculated temperature distribution after
mechanical van Genuchten, f (1/Pa) 9.73E-5 Pa-1
Properties
12 months of heating.
van Genuchten, mf (-) (or ) 0.492
Residual saturation 0.01 25
Initial fracture S l (-)
Fracture frequency 4.32 m-1
20 saturation 0.1
0.6
Initial fracture aperture, bI 51.8 m 0.4
bmax for Equation (16) 150 m
15 Condensation Zone with 0.2
Wetting of Fractures 0
-1
Exponent d for Equation (16) 0.6 MPa 10

Rock Mass
Z (m)

Youngs Modulus 14.77 GPa 5

Mechanical
Properties Poissions ratio 0.21
0 Dry-out
Thermal Expansion Coefficient, T 5+0.0583T zone
106/C -5

-10

Figure 13 presents the calculated THM-induced -15

-25 -20 -15 -10 -5 0 5 10 15 20 25
changes in air-permeability. These changes in air- X (m)
permeability are caused by the combined effect of
thermally induced changes in fracture moisture Figure 11. Calculated liquid saturation in fractures
content (Figure 11) and stress-induced fracture aper- after 12 months of heating.
ture changes (Figure 12). Near the heat source, the
permeability decreases mainly because of fracture
closure, but is also affected by thermally induced 25
xx (MPa)
Slight Reduction
wetting and drying. Away from the heat source, a 20 in Horizontal Stress 1
zone of increased permeability has developed as and Opening of
-1
Pre-existing Fractures
results of opening of vertical fractures. 15
-5
-10
10
-15
-20
Z (m)

5
-25
-30
0 Strong Thermal
Compressive Stress
-5 and Fracture Closure

-10

-15
-25 -20 -15 -10 -5 0 5 10 15 20 25
X (m)

Figure 12. Calculated thermally induced horizontal

stresses after 12 months of heating.
 -8-

performance of a potential repository at Yucca

25
Increased k k/ki (THM) Mountain are discussed in Rutqvist and Tsang
20 74:4 1.5 (2003).
1.1
15 0.9 CONCLUDING REMARKS
0.5
10 0.1 We have linked two codesTOUGH2 and
76:3
FLAC3Dfor analysis of coupled THM processes in
Z (m)

5
complex geological media. The codes were linked
0
with modules representing the coupled thermo-
Reduced k mechanical and hydrologic-mechanical behavior of
-5 rocks. The coupling modules contain nonlinear stress
-10 Reduced k versus permeability functions, which were calibrated
against site-specific data. These coupling modules
-15 could be exchanged with modules containing any
-25 -20 -15 -10 -5 0 5 10 15 20 25
X (m) other type of empirical or theoretical hydrologic-
mechanical coupling relationship. We have demon-
Figure 13. Calculated changes in air-permeability as strated the usefulness of linked, sequentially coupled
a result of TH induced changes liquid THM analyses for complex problems related to
fracture saturation and TM induced injection and storage of CO2 in brine aquifer forma-
changes in fracture aperture. 74:4 and tions and to the expected conditions at a high
76:3 are borehole sections at which temperature nuclear waste disposal in unsaturated
calculated and measured air-permeability rock. In both these examples, the hydromechanical
is compared in Figure 14. changes are relatively slow, which is most suitable
for the sequentially explicit solution. Problems with
Figure 14 presents a comparison of calculated and higher strain rates relative to fluid mobility may
measured air-permeability at two borehole sections. require the sequential implicit approach or ultimately
The calculated changes in air-permeability is in good a fully implicit coupled approach.
agreement with the measured ones both in trends and
magnitude. The good agreement between the calcu- ACKNOWLEDGMENT
lated and measured results shows that the adopted Technical review and comments by Dr. Kazumasa Ito
conceptual model for stress induced changes in and Dr. Guomin Li, Lawrence Berkeley National
permeability depicted in Figure 9 is sound and Laboratory are much appreciated. This work was
validates the numerical values of bmax and d that jointly supported by the Director, Office of Science,
defines the normal stress-aperture function in Office of Basic Energy Sciences, Division of
Figure 9c. Chemical Sciences, Geosciences and Biological
Sciences, of the U.S. Department of Energy, under
contract No. DE-AC03-76-SF00098, and by the
PERMEABILITY CORR. FACTOR

2
Measurements
1.8 Calculations Director, Office of Civilian Radioactive Waste
1.6
1.4 74:4 Management, U.S. Department of Energy, through
1.2 Inecreased k Memorandum Purchase Order EA9013MC5X
1 between Bechtel SAIC Company, LLC and the
0.8
0.6
Ernest Orlando Lawrence Berkeley National
0.4 76:3 Laboratory (Berkeley Lab) through the U.S.
Reduced k
0.2 Department of Energy Contract No. DE-AC03-
0 76SF00098, and by the Swedish Nuclear Power
0 10 20 30 40 50
TIME (Months) Inspectorate.
Figure 14. Calculated and measured permeability
REFERENCES
correction factor Fk (see Equation 15 for
definition of Fk). Biot, M. A., General theory of three dimensional
consolidation. J Applied Physics, 12, 155164, 1941.
In this example, we could utilize the capabilities of
the TOUGH2 code to simulate complex above Corey AT. The interrelation between oil and gas
boiling two-phase flow behaviorfor efficiency, this relative permeabilities. Producers Monthly Novem-
requires special treatment with exchange of primary ber, 38-41, 1954.
variables in the numerical solution (Pruess et al.,
1999)and a full THM simulation was achieved Davis, J.P., and D.K. Davis, Stress-dependent perme-
through coupling with the FLAC3D code. More long- ability: characterization and modeling. Society of
term results and the impact of these processes on the Petroleum Engineers, SPE Paper no 56813, 1999.
 -9-

Itasca Consulting Group Inc. FLAC-3D Manual: Fast Rutqvist, J., and C.-F. Tsang, A study of caprock
Lagrangian Analysis of Continua in 3 Dimensions hydromechanical changes associated with CO2
Version 2.0. Itasca Consulting Group Inc., Minne- injection into a brine aquifer. Environmental
sota, USA, 1997. Geology, 42, 296-305, 2002.

Leverett, M.C., Capillary behavior in porous media. Rutqvist, J., and C.-F. Tsang, Analysis of thermal-
Trans, AIME, 142, 341-358, 1941. hydrologic-mechanical behavior near an emplace-
ment drift at Yucca Mountain. J. Contaminant
Pruess, K., C. Oldenburg, and G. Moridis, TOUGH2 Hydrology (In press), 2003.
Users Guide, Version 2.0, Report LBNL-43134,
Lawrence Berkeley National Laboratory, Berkeley, van Genuchten MT. A closed-form equation for
Calif., 1999. predicting the hydraulic conductivity of unsaturated
soils. Soil Sci Soc Am J 1980;44:892-898.
Rutqvist, J, Y-S. Wu, C-F Tsang and G. Bodvarsson,
A modeling approach for analysis of coupled multi- Witherspoon, P.A., J.S.Y. Wang, K. Iwai and J.E.
phase fluid flow, heat transfer, and deformation in Gale. Validity of the cubic law for fluid flow in a
fractured porous rock, Int. J. Rock Mech. Min. Sci. deformable fracture. Water Resour. Res.
39, 429-442, 2002. 1980;16:1016-1024.