You are on page 1of 10

ARMA 10-162

Risks and mitigation problems in a CO2 injection project for a petroleum


onshore field in Brazil

Mendes, R.A., Costa, A.M., Sousa Jr., L.C. and Pereira, L.C.
Petrobras, Rio de Janeiro, Rio de Janeiro, Brazil
Oliveira, M.F.F.
Tecgraf/PUC-RJ, Rio de Janeiro, Rio de Janeiro, Brazil

Copyright 2010 ARMA, American Rock Mechanics Association


This paper was prepared for presentation at the 44th US Rock Mechanics Symposium and 5th U.S.-Canada Rock Mechanics Symposium, held in
Salt Lake City, UT June 2730, 2010.
This paper was selected for presentation at the symposium by an ARMA Technical Program Committee based on a technical and critical review of
the paper by a minimum of two technical reviewers. The material, as presented, does not necessarily reflect any position of ARMA, its officers, or
members. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of ARMA
is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The
abstract must contain conspicuous acknowledgement of where and by whom the paper was presented.

ABSTRACT: This work shows the methodology for the analysis of fault reactivation in terms of reservoir depletion and/or
reservoir pressurization. During reservoir depletion, the normal effective stresses at the fault plane increase and the fault tends to
close. However, in some reservoirs with different pressure levels intercalated by shales of little thickness, the fault can be
reactivated by differential pressure. On the other hand, during injection, the normal effective stresses on the fault plane may
decrease to zero and under this condition, the fault opens and fluid may migrate to another reservoir with a lower pressure. The
fault reactivation process is controlled mainly by the initial state of stresses (vertical and minimum horizontal stresses) and by the
fault cohesion and fault friction angle.

horizontal stresses in the cap rocks in order to


1. INTRODUCTION mitigate fracture propagation;
Several oil and gas fields make use of water or gas Pore pressures in the fault extension within the
injection to maintain pressures and displace oil nested in reservoir should be lower than the normal stresses
reservoirs to the producer wells, thus increasing the in the reservoir areas adjacent to the fault, i.e., the
recovery factor. In particular, CO2 injection has been reservoir fault should not present hydraulic
shown to be a way through geotechnical sequestration of fracturing.
avoiding gas disposal to the environment. For these
purposes, a CO2 continuous injection project was The stratigraphic layout of Miranga consists of
proposed for the onshore field Miranga in Recncavo successive porous layers intercalated by thin shale
Basin, northeast of Brazil. layers. Previous to injection, the field has undergone a
Miranga is comprised by a turtle-back anticline with an period of depletion that modified the original stress state
important net of axial faults compartmentalizing the and should be taken into account. Because of the
field in several structural blocks, as shown in Figure 1. differences in pore pressure along the faults through,
The main reservoir Catu1, that is receiving CO2 extensions of the faults may be reactivated, allowing
injection, is located in the structural blocks 1 and 2. intercommunication between reservoirs. During
injection, pore pressure in reservoir Catu 1 increases,
In order to properly estimate the injection system affecting strains and stresses in the whole formation and
capacity, one should determine the maximum injection herein admitting fault reactivation and/or opening.
pressures allowed, which must satisfy the following
design criteria: In order to access the stress evolution in the formation
during the pore pressure load history and determine the
Effective stresses in the reservoir should be maximum injection pressure, geomechanical analyses
compressive, i.e. there should be no tensile stresses; were carried out on three geotechnical sections, namely
Pore pressures in the fault extension within the Section 1, Section 3 and Section 6, as depicted in
reservoir should be lower than the minimal Figures 2 and 3. The sections were traced
perpendicularly to the main faults, i.e. aligned to the
minimum horizontal stresses, passing through injector
wells.
The two-dimensional stress analyses were performed by
employing the Finite Element Method (FEM) under the
assumption of plane strains in terms of the effective
stresses under drained conditions. Results indicate that
the opening of the faults within reservoir Catu 1 (third
design criterion) controls the determination of the
maxima injection pressures allowed in Miranga field.
To take into account the uncertainties of the geotechnical
parameters and their influence on fault opening,
probabilistic analyses were carried out using the
Response Surface Methodology (RSM) and Monte-
Carlo simulation. As a final result, cumulative
probability curves of the maximum injection pressure
were obtained.
This paper summarizes the methodology employed in
this study, for both deterministic and probabilistic
models. Results and discussion are presented for the
analyzed sections and faults.

Fig. 2. Map with main faults and injector wells in blocks 1 and
2 of Miranga field. Sections selected for study (Sections 1, 3
and 6) are shown in red. (Petrobras)

Fig. 1. Structural map of Miranga field (top of Catu 1 layer)


showing the studied blocks (1 and 2) and their adjacent blocks
(4 and 8). Notice the structural complexity of the field.
(Petrobras)
Fig. 3. The three studied sections passing through the main
faults and wells of Miranga field. (GOCAD Paradigm)
2. FAULT REACTIVATION AND OPENING Fig.4. (a) Failure envelopes for the reservoir rock and fault,
where q stands for deviatoric stress and oct stands for
Fault reactivation and opening are controlled by the octaedric stress; (b) Stress-path of a fault material point,
stress components acting tangent and normal to the fault where: 0 = initial state of stress in the reservoir; 1 = fault
plane, respectively. These tangential and normal stresses reactivation in the reservoir due to shear failure, increasing
are related to their corresponding tangential and normal fault permeability; 2 = fault opening in the reservoir due to
stiffness properties, which are based on the filling annulment of effective normal stresses.
material and the thickness of the fault core. In the
present study, it is assumed that the fault shear 3. MECHANICAL PROPERTIES AND IN-SITU
stresses () behave according to the Mohr-Coulomb STRESS STATE
criterion with an associative plasticity law, whereas the
fault normal stresses (n) follow a linear elastic Rock properties regarding deformation and strength
constitutive model. were obtained from Mirangas well profiles [1]
calibrated with laboratory tests [2]. Dynamic Youngs
A geotechnical fault reactivates or opens whenever the modulus (Edin), Poissons ration (), cohesion (c) and
shear stress reaches the Mohr-Coulomb envelop or the
friction angle () were estimated from compressional
normal stress decreases to zero, respectively.
and shear sonic log-wells. Static Youngs modulus (Eest)
When the injection in the reservoir initiates, the fault was taken as Edin /3. Table 1 presents the mechanical
extension within the reservoir is considered to be under rock properties of the Miranga formation.
the same pore pressure of its surroundings. As the pore Table 1. Rock properties
pressure increases, strains and stresses in the formation
change, causing an increase in the effective shear Layer Edin(GPa) Eest(GPa) c(MPa) ( )
stresses and a decrease in the effective normal stresses Top 15.60 5.20 0.34 7.81 31.0
on the fault. As a consequence, the fault can reactivate
and fluid can migrate from the reservoir. In this case, the Mark B 15.60 5.20 0.34 7.81 31.0
internal fluid pressure of the reactivated fault extension Santiago 15.60 5.20 0.34 7.81 31.0
is assumed to be the same as the pore pressure acting in
the reservoir to which the fault is connected. As the Cap rock 17.15 5.72 0.35 7.46 33.0
injection continues and pore pressures increase, fault Catu 1 15.60 5.20 0.31 11.27 38.7
extensions outside the reservoir can also reactivate until
Base 15.60 5.20 0.34 11.27 38.7
the effective normal stress decreases to zero, causing the
fault to open and leading to hydraulic fracturing.
Figure 4 illustrates the stress path of a fault material The deformation properties of the faults were evaluated
point from its initial stress state to its reactivation and according to the deformation properties of their adjacent
opening, where q stands for deviatoric stress and oct for rocks and their core thickness, which were estimated
octaedric stress. being between 0.5 and 2.0 m based on the faults
through. Table 2 presents the values of Poissons ratio
and normal and tangential stiffness (Kn and Kt) adopted
for the geotechnical faults.
Table 2. Geotechnical fault properties
Layer Kndin Ktdin Knest Ktest
(GPa/m) (GPa/m) (GPa/m) (GPa/m)
Top 0.34 31.20 11.64 10.40 3.88
Mark B 0.34 31.20 11.64 10.40 3.88
(a)
Santiago 0.34 31.20 11.64 10.40 3.88
Cap rock 0.35 34.30 12.70 11.43 4.23
Base 0.34 31.20 11.64 10.40 3.88
Faults 0.33 33.27 12.44 11.09 4.15
through

The cohesion and the friction angle of the faults were


(b) adopted as being 750 kPa and 18.3, respectively, based
on residual values (1AF first after failure) obtained in
Fault Fault Fault
triaxial and direct shear tests of the intact caprock shale 1 2 3
[3], as shown in Figure 5.
Bl 2 Bl 1 Bl 1 Bl 4
Overburden

Underburden
1

Catu 1
Catu 2 4.6 km x 2.3 km
Catu 3
Fig. 5. Residual shear envelope from triaxial and direct shear
Catu 4
laboratory tests in shales [3]. (Petrobras) Catu 1
Catu 2
In-situ effective vertical stresses were evaluated by Catu 5 Catu 3

integrating the density log with depth. A unit weight Fig. 6. Finite element model of Section 1 (faults in magenta),
with a total of 47,195 nodes and 15,837 elements. In detail, the
(effective) of 12.06 kN/m was found at the top of main reservoir Catu 1, on the left and right side of Fault 1.
reservoir Catu 1. The minimum effective horizontal
stresses were obtained from fracture well tests [1],
leading to a lateral earth pressure coefficient (K0) of Fault Fault
1 2
0.58.
Reservoir Catu 1 has a porous thickness varying from 0 Bl 2 Bl 1 Bl 4
to 20 m, with porosity around 15%. A unity valued
Biots coefficient was adopted for all reservoirs.

4. FINITE ELEMENT ANALYSIS


Catu 1
Catu 2
The finite element analyses were carried out for the three Catu 3 1
selected sections by using the program AEEPECD [4], Catu 4
developed in Petrobras for two-dimensional analysis in 4.2 km x 2.3 km
terms of effective stresses under drained condition. Catu 5 Catu 1
Catu 2
Catu 6 Catu 3
Rocks were modeled by quadratic quadrilateral 8-node Catu 4
serendipity finite elements. Non-reservoir and reservoir
rocks were assumed to behave according to the Fig. 7. Finite element model of Section 3 (faults in magenta),
elastoplastic and poro-elastoplastic Mohr-Coulomb with a total of 27,246 nodes and 9,083 elements. In detail, the
main reservoir Catu 1, on the left and right side of Fault 1.
constitutive model, respectively, with associated flow
rule and zero tensile cut-off. Fault Fault
Faults were modeled by quadratic interface (or joint) 8- Fault 2 4 Fault
Fault Fault 6
node elements [5]. A poro-elastoplastic Mohr-Coulomb 1
3 5
constitutive model, with associated flow rule and zero
tensile cut-off, simulated the behavior of the effective Bl 8 Bl 2 Bl 1 Bl 1 Bl 4
normal and shear stresses (n and ) along the fault.
Zero displacements were prescribed in all directions on
the bottom boundary and in the horizontal direction on
the lateral boundaries of the model. Catu 1
Catu 2
Figures 6, 7 and 8 show the finite element models of Catu 3 1 2
sections 1, 3 and 6, respectively. Catu 4
4.0 km x 2.3 km
Catu 5

Catu 6 Catu 1

Fig. 8 Finite element model of Section 6 (faults in magenta),


with a total of 39,512 nodes and 13,325 elements. In detail, the
main reservoir Catu 1, on the left and right side of Catu 1.
In-situ vertical and horizontal stresses were modeled as redistributed to the surrounding rock material in the
initial geostatic stresses varying with depth. For each iteration process itself.
section, the following three load cases were analyzed:
Load case 1: Depletion of all reservoirs.
5. PROBABILISTIC ANALYSIS
Static values of Youngs modulus and their
A probabilistic analysis was performed for each fault of
corresponding fault stiffness were adopted as
sections 1, 3 and 6 to take into account the uncertainties
rock and fault properties. Pore pressures values
of the geotechnical parameters and their influence on
were measured from Miranga wells and
fault opening.
increments (P) were applied to each reservoir
of the model. The main objective of this load The Response Surface Methodology (RSM) was
case is to identify fault extensions adjacent to employed together with the Design of Experiments
Catu 1 that reactivate due to pore pressure (DoE) in the Cougar software [6] to obtain, with
differences between both sides of the fault reasonable accuracy, a response surface that
through. approximates the pressurization value of reservoir
Catu 1 for which the analyzed fault opens, for a given set
Load case 2: Injection of reservoir Catu 1 from of uncertain geotechnical parameters.
the initial in-situ stress state.
An average of 80 numerical simulations was necessary
Dynamic values of Youngs modulus and their to build each response surface with quadratic
corresponding fault stiffness were adopted as polynomials using the Small Face Centered Composite
rock and fault properties. Pore pressures were Design technique as the DoE to optimize the choice of
obtained from flux simulation and increments experiments in the region of interest. One may cite the
(P) were applied to reservoir Catu 1. Fault unit weight (effective), the lateral earth pressure coefficient
extensions that lay within reservoir Catu 1 or (K0), the fault cohesion (c) and the fault friction angle
had been reactivated during depletion (load (), among others, as the parameters varied in the
case 1) were loaded from the beginning of the domain of interest.
analysis with the same pore pressure increments
of Catu 1. A sensitivity analysis of the response surface made
possible to build a Pareto chart and identify the most
Load case 3: Pressurization of reservoir Catu 1 influential geotechnical parameters in the fault opening
from an initial stress state equivalent to the occurrences.
stress state of Catu 1 after depletion.
Finally, a Monte-Carlo simulation of 10,000 experiments
Similar to load case 2, being the only difference was performed on the surface response, by assigning the
the initial stress state, for which triangular probability law and suitable interval values to
effective = 15.56 kN/m3 and K0 = 0.56 were the 8 most relevant geotechnical parameters. As a final
adopted to approximate the stress state of Catu 1 result of this probabilistic analysis, one obtains a
after depletion. cumulative probabilistic curve (S-Curve) which provides
On the interface elements, pore pressure simulates the the probability of a given pore pressure increment (P)
effect of internal fluid pressure acting on the fault in reservoir Catu 1 to cause fault opening.
extensions that are hydraulically connected to the
pressurized reservoir. Only the effective normal stress
components are directly affected by the pore pressure 6. RESULTS AND DISCUSSION
increment. In the following, results for both deterministic and
In load cases 2 and 3, pore pressure is not applied on probabilistic analysis are presented in detail for
interface elements external to the reservoir from the Section 3. Only final results will be shown for Sections 1
beginning of the analysis, except for the extensions that and 6.
are adjacent to Catu 1 and those that have already been
reactivated due to depletion. As the reservoir pressure 6.1. Deterministic Analyses
increases, interface elements that fail in shear are
automatically loaded with the same pore pressure acting
In load case 1, the main reservoir Catu 1 of Section 3
in the reservoir.
was depleted by P=-65 kgf/cm2, as shown
When the effective normal stress of an interface element schematically in Figure 9, which also presents the
decreases to zero, the fault opens and does not transmit depletion increments for the adjacent porous layers.
shear stresses. In this case, these shear stresses are
Results indicate that fault reactivation due to depletion Fault Fault
1 2
occurred only at isolated points of shale layers.
Figure 10 shows in detail the reactivation results for Bl 2 Bl 1 Bl 4
Fault 1, where the red circles indicate the reactivated -45
points. The graphic on the right plots the Shear Stress
Ratio (R = /Mohr-Coulomb), in which values equal to 1 -55
-35
configure failure based on the Mohr-Coulomb criterion -65
-85-75
and, accordingly, reactivated fault points. Figure 11 -55
-65 -15
-15 -75 -95
shows friction angles that would be required to avoid -55 -35
-85
-115

fault reactivation due to depletion (req) in the cases of -55


-95 Depletion
c=0 and c=750 kPa. Notice this value exceeds the faults Injection
friction angle =18.3 in the case of null cohesion for
Fig. 9. Depletion pore pressure increments P (kgf/cm2) of
points 2, 3, 4 and 5 of the fault. load case 1 for Section 3.

Fig. 10. On the left, detail of finite element mesh of Fault 1 of Section 3. On the right, graphic of the Ratio= /Mohr-Coulomb along
the fault.

Fig. 11. Required friction angle (req) of the Fault 1 of Section 3 to prevent reactivation. Points 4 and 5 correspond to shale layers
underlying the reservoir Catu 1 that have been reactivated during depletion (req>18.3).
During depletion, the effective normal stresses increase pressure increments (P) in Catu 1 for Section 3 to
and faults tend to close. For this reason, for all sections, prevent faults 1 and 2 to open.
fault reactivation was observed only at isolated points,
mainly where substantial differences in pore pressure
exist between the two sides of the fault.
On the other hand, when pressurization is applied, fault
reactivation is more likely to occur. The increase in pore
pressure leads to an effective stress decrease; hence the
fault tends to open and fluid can migrate. In load cases 2
and 3, the main reservoir Catu 1 of Section 3 was
pressurized, as shown schematically in Figure 9, which
Fig. 13. Maximum pore-pressure increments in Catu 1 to
also indicates the pore pressure increments (P) applied
prevent fault opening of Fault 1 and Fault 2 in Section 3.
to the faults. Tables 3 and 4 summarize the results for
Section 3 of load cases 2 and 3, respectively, where P is Tables 5 and 6 show the results obtained for Sections 1
the pore pressure increment; a positive P means pore and 6, respectively. In Table 5, the symbol > in some
pressure increase (injection); and a negative P means values of P indicates that convergence could not be
pore pressure decrease (depletion). achieved for higher values, due to substantial differences
in the pore pressure distribution along Catu 1. Therefore,
Fault Fault
1 2 for these cases in Fault 1, reactivation and opening will
Bl 2 Bl 1 Bl 4 occur for higher values than reported.
-15
+11 Table 5. Results for load case 2, Section 1.

+66 Fault Reservoir P (kgf/cm2)


+55 Depletion
Fault Fault
Injection reactivates opens
1 Left Catu 1 >6.8 >6.8
Fig. 12. Injection pore pressure increments P (kgf/cm2) in Right Catu 1 24.1 >41.6
Catu 1 for load cases 2 and 3, Section 3. 2 Left Catu 1 32.9 60.5
Right Catu 1 39.0 63.5
3 Left Catu 1 36.0 60.5
Table 3. Results for load case 2, Section 3.
Fault Reservoir P (kgf/cm2)
Table 6. Results for load case 2, Section 6.
Fault Fault
reactivates opens Fault Reservoir P (kgf/cm2)
1 Left Catu 1 22.9 65.8 Fault Fault
Right Catu 1 19.9 69.5 reactivates opens
2 Left Catu 1 33.1 60.3 1 Right Catu 1 21.4 64.3
2 Left Catu 1 24.5 64.3
Table 4. Results for load case 3, Section 3. Right Catu 1 21.4 64.3
3 Left Catu 1 24.5 61.2
Fault Reservoir P (kgf/cm2)
Right Catu 1 27.6 67.3
Fault Fault
4 Left Catu 1 30.6 55.1
reactivates opens
Right Catu 1 30.6 58.2
1 Left Catu 1 -45.1 19.2 5 Catu 1 61.2 67.3
Right Catu 1 -45.1 22.9 6 Left Catu 1 27.6 64.3
2 Left Catu 1 -31.9 10.5

Load case 2 is based on the initial in-situ stress state 6.2. Probabilistic Analyses
while load case 3 on an equivalent initial stress state,
with higher stress values, and as a consequence, Table 3 As mentioned before, load case 2 has been chosen for the
presents P values greater than Table 4. Since probabilistic analyses to obtain the maximum injection
reactivation due to depletion has shown to be difficult to pressure of Catu 1.
happen, in this study load case 2 was chosen for the
Figure 14 shows, in a Pareto chart, the results of the
probabilistic analyses to determine the maximum
sensibility analysis of the geotechnical parameters
injection pressure of the reservoir Catu 1 for each
chosen as input for the response surface of Fault 2 of
analyzed section. Figure 13 shows the maximum pore
Section 3. The Monte-Carlo simulation over the response just presented. The other two criteria were also
surface leads to the cumulative probability curve (S- addressed.
curve) depicted in Fig. 15.
Regarding the first criterion, to maintain reservoir Catu 1
The maximum pore pressure increment allowed in in compressive regime, a maximum pore pressure
Mirangas CO2 injection project was adopted as being increment of P=+115 kgf/cm2 was found, preventing
the P5 value in the S-curve. This choice was based on the null effective stresses.
curves inflexion point, after which the cumulative
For the second criterion, to avoid pore pressures greater
probability increases rapidly. The maximum pore
than the minimum horizontal stresses in reservoir Catu 1,
pressure increment should not be exceeded in the
a maximum pore pressure increment of P=+93 kgf/cm2
reservoir areas adjacent to the fault in order to avoid fault
was obtained, preventing hydraulic fractures in the cap
opening and fluid leakage to upper layers.
rock.
Both the above values are higher than the maximum pore
6.3. Other Design Criteria pressure increments obtained for preventing the fault to
open and therefore, the third criterion, for prevention of
As mentioned in the introductory section of this paper, fault opening, was proved to be critical.
the maximum injection pressure must be in accordance
with three design criteria. The third criterion has been
determined in the deterministic and probabilistic analyses

Fig. 14. Pareto chart for Fault 2 of Section 3.

Fig..15. Cumulative probability curve (S-curve) from Monte-Carlo method for Fault 2 of Section 3.
Fig.16. Cumulative probability curve (S-curve) from Monte-Carlo method for Fault 2 of Section 1.

Fig.17. Cumulative probability curve (S-curve) from Monte-Carlo method for Fault 1 of Section 6.

results; however these field data are difficult to obtain.


The in-situ stress state also presented relevance. In this
7. CONCLUSIONS
sense, it is important to plan mini-frac tests in wells to
The establishment of the maximum injection pressure determine the magnitude of the stress state during the
is essential for proper evaluation of the injection injection.
system capacity in CO2 injection projects. The
The analyses simulating the pressurization of reservoir
proposed approach simulates the geomechanical
Catu 1 has shown that fault reactivation may occur.
behavior of the formation by applying the finite
Well pressures in all reservoir layers should be
element method, while the uncertainties of the
monitored during CO2 injection, to detect any possible
involved geotechnical parameters are taken into
intercommunication through existing faults.
account in probabilistic analyses. This methodology is
suitable for the determination of the maximum increase In order to manage pore pressure, stress state and fault
in pore pressure that does not cause the opening of pre- reactivation in the reservoir and neighboring layers it is
existing faults and their subsequent hydraulic necessary to monitor the rock mass in terms of
fracturing above reservoirs. measurements of injector well pressure and permanent
microseismic.
From the sensibility analyses, the cohesion and friction
angle of the filling material of the fault core proved to
be the most influent geotechnical parameters in the
REFERENCES
1. Arajo, E. 2007. Geomechanical Model and Analysis
of the Seal Faults in Miranga Field. Houston: Report
from GeoMechanics International for Petrobras.
2. Bloch, M. 1993. An Experimental Study of
Mechanical Properties of a Sandstone from an Oil
Reservoir. M.Sc. Thesis. Rio de Janeiro: Catholic
Universitiy of Rio de Janeiro (PUC-Rio).
3. Santos, E.S.R. 2008. Residual Mechanical Properties
of Shales. Petrobras internal report no.RT/TEP 020/08
project PD10320.
4. Costa, A. M. 1984. An Application of Computational
Methods and Principles of Rock Mechanics in the
Design and Analysis of Underground Excavations for
the Underground Mining. Ph.D. Thesis. Rio de
Janeiro: Graduate School of Engineering (COPPE),
Federal University of Rio de Janeiro (UFRJ).
5. Goodeman R.E., Taylor, R.L., and Breeke, T.L.
1968. A model for the mechanics of jointed rocks.
ASCE j. Soil Mech. Found. Div., 94: 637-659.
6. Reference Manual. COUGARTM Software. Version
2007.2.

You might also like