th

Proceedings of the ASME 2019 38 International

Conference on Ocean, Offshore and Arctic Engineering
OMAE2019
June 9-14, 2019, Glasgow, Scotland

OMAE2019-96319
NUMERICAL MODELING OF RADIAL FRACTURING OF CEMENT SHEATH CAUSED BY PRESSURE
TESTS

Sohrab Gheibi1, Sigbjørn Sangesland Torbjørn Vrålstad

NTNU SINTEF
Trondheim Norway Trondheim Norway

ABSTRACT cement sheath failure. The failure mechanisms of cement

To achieve an acceptable level of zonal isolation, well integrity of the cement and well system in the subsurface to
integrity should be guaranteed in hydrocarbon production and shorten the economic and environmental issues raised by
geological CO2 sequestration. Well pressure test can cause sheath in a well can be shear failure, tensile (or radial
different types of failures in the well system leading to leakages cracking/fracturing), debonding of cement-casing and cement-
through these failures. Laboratory evidences have revealed that rock interface [3].
occurrence of radial tensile fractures is likely during pressure While performing the pressure test, the cement sheath can fail
tests. due to developed excessive tensile stress in the cement sheath.
In this paper, we use a numerical code call MDEM which was As a result, radial fractures can form which can act as leakage
formulated based on discrete element method. The code can pathways compromising the cement sheath functionality. There
model discontinuum feature of fractures. A model of a lab-sized are well-known researches published in the literature reporting
pressure test was built and compared to an experiment pressure tests performed in the lab. Goodwin and Crook [4]
previously published. The model was tested under different carried out pressure tests measuring the permeability of the
confinement levels and effect of the tensile strength of rock on different cement sheath after failure [4]. They showed that the
the radial fracture was investigated at the same lab-scale. expansion of the casing due to internal pressure increase can
Fracture opening profiles are also presented showing the create radial fractures in the cement sheath. Jackson and
leakage potential of these fractures under different pressure Murphey [5] also carried out similar tests, increasing and
level. decreasing pressure inside the internal casing. They observed
Keywords: Well integrity, cement sheath, well pressure that air flows through the cement-casing system after reduction
test, fracture of the pressure followed by pressure test below a level [5].
Boukhalifa et al. [6] performed experiments simulating the
expansion and contraction of the casing leading to the
INTRODUCTION formation of radial fracture and microannuli. They showed how
Efficient zonal isolation of the subsurface formations is the radial cracks and microannuli contribute to the permeability
one of the main objectives of well cementing. Cement sheath of casing-cement sheath for different flexible, expanding and
provides a barrier to fluid migration between different foamed cement systems under several conditions.
formations. Cement acts also as a support for wellbore and There are several papers in the literature modeling the well
protects the casing in the corrosive medium of a well. However, integrity using numerical simulations. Gray et al. [7] used a
the dynamic nature of hydrocarbon production or fluid injection staged finite-element approach taking in-situ stress state,
(e.g. CO2) imposes extreme loadings on the cement sheath in nonlinear behavior of cement and formation etc. into account to
both short and long-term in terms of pressure and temperature provide a more realistic calculation of well integrity. Ravi et al.
variations. Failure of cement sheath can lead to problems in the [8] applied finite element method to model debonding,
integrity of production or injection wells as well as after well cracking, and plastic deformation of cement under different
abandonment [1,2]. Therefore, it is important to maintain the loadings such as pressure testing, well completion, hydraulic
fracturing, and hydrocarbon production.
Skorpa et al. [9] performed a lab-sized experiment of pressure
1
Contact author: [email protected] test in a cell under no confinement. They showed that radial
fracture can form due to the pressure increase. They observed

1 Copyright © 2019 by ASME

that the radial fracture initiates in the cement sheath and The fixed boundary condition in all the models is constant
propagates into the rock formation. They visualized the crack pressure on the casing wall. And the minimum mesh size was 1
path using CT-Scan. mm closer to the borehole and becomes larger as getting away
In this paper, we investigate the propagation of radial fractures from the borehole.
created in the well pressure test done by Skorpa et al. [9]. The
paper first presents the pressure test experiment performed in
the lab. Second, a numerical model is constructed to calibrate
the model against the experiment and then we investigate the
effect of rock tensile strength, stress confinement on the
behavior of the radial fracture as well as its opening profile. At Rock
the end, the conclusions are given in the last section. Cement

PRESSURE TEST EXPERIMENT

Skorpa et al. [9] developed a laboratory set-up to

investigate cement degradation during pressure cycling. In this
set-up, both debonding and cracking of the cement sheath are
characterized by X-ray computed tomography. The set-up
allows for tests with different types of rocks, casings and Crack
cement systems, as well as the inclusion of drilling fluids. They
performed a cyclic pressure test of a carbon-steel (X-52) with
outer diameter 60.3 mm and 2 mm thickness casing, Castlegate
Casing
sandstone with 200 mm height, 152 mm diameter and a
concentric borehole with 76 mm diameter. The cement had
cured at 66°C for five days, the sample was exposed to 20
pressure cycles at 30 MPa, where one cycle consisted of 1 FIGURE 1: RADIAL FRACTURE INDUCED BY CYCLIC
minute with and 1 minute without pressure [9]. A single radial PRESSURE TEST [9]
fracture was created in the cement and penetrated the rock
formation. Figure 1 shows the CT image of the casing-cement- Table 1: ELASTIC PROPERTIES OF THE MODEL
rock system and the fracture is visible. For details about the set-
up and the test the reader is referred to [9]. Domain Young’s modulus (GPa) Poisson’s ratio
Casing 210 0.3
NUMERICAL MODEL Cement 9.2 [7] 0.15 [7]
To model the initiation and propagation of the radial crack Rock 5 0.25
under pressure test, a hybrid FEM/DEM in-house code called Surrounding 5 0.25
MDEM was used [10]. The code behaves like a regular
continuum for elastic domain, but it becomes discrete as soon
as a fracture forms. This enables the code to provide a more 3.1 Calibration of the Model
realistic analysis of fracture propagation. Pre-existing fracture We performed a series of analyses to calibrate the model to
and new fractures can form in the code having the possibility to the pressure test discussed in Section 2. The pressure was
close, open and shear depending of the loading condition. The increased step by step to the final well pressure, 30 MPa.
code was described in Refs [10,11] in detail. Actually, this is different than what performed in the test.
A model of the pressure test specimen was built including the Therefore, we neglect the dynamic effects and our simulation is
casing, cement and rock domains with their actual sizes in the a quasi-static loading. Also, the model does not include the
test performed by [9]. Figure 2 presents the geometry of the cyclic pressure increase-decrease as in the lab and assumed that
model and the colors refer to the different domains. The only the failure occurs in the first loading. The reason was that the
difference between the specimen and the model is that the outer code is not yet able to consider the effect of cycling loading.
boundary is square instead of being circular and it is larger in Moreover, the cement and rock were assumed to be brittle even
the size (i.e. the region called “Surrounding” in Figure 2). The though rocks and cement show quasi-brittle behaviour. This
main reason to do this is to avoid numerical complication after assumption simplifies the model and enables a faster analysis of
the fracture reaches the boundary of the model. Table 1 pressure test. Brittle behaviour means that the fracture will
summarizes the elastic parameters of each of the domains. We propagate if the absolute value of tensile stress exceeds the
did not measure the values, but they were taken from the tensile strength. All of the assumptions in a way will affect the
literature and the corresponding references were also provided results. The simulations were performed under room
in the table. The values for Castlegate sandstone were taken temperature similar to the experiment in the lab.
after consulting with SINTEF’s formation physics laboratory.

2 Copyright © 2019 by ASME

In the calibration process, we chose a set of tensile strength for boundary (shown by arrows in the figure). This implies that the
the cement and the rock so that we create a single radial absolute value of tensile stress evolved in the rock (in the rock-
fracture at pressure 30 MPa obtained in the lab. The fracture cement interface) is greater than 4.5 MPa. It is important to
must initiate in the cement, crossing the cement-rock interface note that if the material model was assumed to be a quasi-brittle
and propagate the whole thickness of the rock (and stop before material (instead of being brittle) the set of strength values
“Surrounding” region in Figure 2). In the first round, the obtained during calibration would be different.
homogenous value for the cement and rock was chosen. It was
concluded that it was not possible to create a single fracture for
a homogenous model, but multiple fractures were created. In
the second round, we tried heterogeneous tensile strength for
the cement and homogeneous model for the rock.
Heterogeneity was imposed by a generation of random tensile
strength value from a uniform distribution U (11.5, 16.5) MPa
for the cement and a constant 7 MPa value for the rock. Figure
3 shows the numerical result and it is well in line with the
fracture in the experiment shown in Figure 1.

FIGURE 3: SINGLE RADIAL FRACTURE CREATED AFTER 30

MPA CASING INTERNAL PRESSURE

3.3. Effect of Confinement

As mentioned, no confinement case used in the test.
However, in the field, the well-casing-cement-rock system is
under confinement. Therefore, the confinement may affect the
propagation of the radial crack due to pressure test. In this
section will show to what extend the confinement level can
influence the occurrence of radial fracture. The confinement
FIGURE 2: THE MODEL OF CASING-CEMENT-FORMATION level starts from being isotropic stress 1 to 20 MPa. In contrast
to the cases in the previous sections, the pressure is increased
well above to create more fractures to see the radial fracture
3.2. Effect of Rock Tensile Strength pattern. Figure 5 shows the fracture pattern for different
Different ranges for the uniform distribution were also confinement level and the corresponding internal pressures. It
examined; however, U (11.5, 16.5) MPa for cement seems to shows that as the confinement increases the fractures become
lead to a better result. Different values (11-4.5 MPa) for the shorter. It is also important to note that the first radial fracture
tensile strength of rock were also examined. Figure 4 shows the was created under 30, 33.5, 43 and 78 MPa pressure for
single radial fracture created for different tensile strength for increasing the confinement level, respectively. In higher
the rock. It is clearly shown that if the tensile strength has a confinement, the radial fractures do not penetrate the rock
higher value the radial fracture is shorter and stops at some formation. It is surprising to see that increasing the well
point in the rock. For models with rock tensile strength lower pressure is not in the favor of propagating the fractures into the
than 7 MPa branching occurred (Fig. 4d-e). Therefore, it was rock, but rather the number of the fractures increases in the
concluded that 7 MPa as the tensile strength of the rock in the cement (Figure 5).
model leads to a more similar result compared to the lab test in In one example, only the rock strength was decreased to 1 MPa
Section 2. It also seems that the radial fracture stops in the under 20 MPa confinement stress. The results showed that for
cement-rock interface if the tensile strength of the rock is 11 very weak rocks there is a low chance that the radial fracture
MPa. It is also interesting that in a model with tensile strength propagates into the rock formation even in high confinement
lower than 4.5, the radial fracture initiates first in the rock then level (Figure 6).
propagates in two directions towards the cement and the outer

3 Copyright © 2019 by ASME

 a) b)
TR=11 TR=10 MPa

c) d)
TR=8 MPa TR=6 MPa

e) f)
TR=4.5 MPa
TR=5 MPa

FIGURE 4: EFFECT OF THE ROCK TENSILE STRENGTH ON RADIAL FRACTURE

4 Copyright © 2019 by ASME

 a) b)

c) d)

e) f)

FIGURE 5: EFFECT OF LEVEL OF CONFINEMENT ON THE EXTENSION OF RADIAL FRACTURE PROPAGATION

5 Copyright © 2019 by ASME

3.4. Fracture Opening Profile
The permeability of the induced fractures depends on their a residual opening after the pressure is reduced to zero. The
opening and length. In this section, we show the profile of the internal parameters of the model could be adjusted such a way
fracture opening in the no confinement case under different that the fracture closes completely or up to a certain range of
internal pressure after the single fracture was created. First, the values. This provides a possibility in calibrating for the
fracture profile is plotted as soon as the internal pressure permeability of fractures in pressure tests. It is important to
reaches 30 MPa so the single fracture shown in Figure 3 is note that the resolution of the CT imaging is 100-200 μm and
created. Afterwards, the pressure was reduced, and the closure width of the fracture in the experiment is much greater than the
of the fracture surface was monitored, and the opening profile value calculated in Figure 7. The main reason is the fact that the
is plotted until the pressure reaches zero. Figure 7 shows the opening of the fracture was suppressed by the surrounding
fracture profiles corresponding to the pressure level after the domain. However, the capability of calculating the fracture
fracture appears. As expected, the maximum opening is opening enables a relative comparison of the effect of different
observed in the maximum pressure and the fracture closes as parameters.
the pressure decreases.
CONCLUSION
A modified discrete element approach was adopted to model
the formation of radial fractures under pressure testing of
casing-cement-rock system. The model was calibrated against a
pressure test experiment. Effect of tensile strength of rock,
confinement on the radial fracture were investigated. The
length of the radial fractures is influenced by the value of
tensile strength of the rock; stronger the rock, shorter the
fracture. Below a specific value for the tensile strength, the
fracture can initiate first in the rock formation not in the cement
sheath. The level of confinement can also affect the fracturing
pressure level and also the extend of fractures, lower the
confinement lower the pressure required to fracture and longer
the its length. In higher confinement, the radial fracture is
confined in the cement sheath not penetrating the rock
formation. It was also observed that the number of fractures
increases rather than propagating to the formation by elevating
the internal pressure in higher confinement. We also concluded
FIGURE 6: RADIAL FRACTURE PATTERN IN HIGH that the confinement is a more influencing parameter than the
tensile strength of rock.
CONFINEMENET FOR VERY WEAK FORMATION
The fracture opening profiles was plotted for different pressure
levels. Higher the pressure level greater the fracture surface
opening, therefore, a greater leakage.

ACKNOWLEDGEMENTS
The authors acknowledge the Research Council of
Norway, AkerBP, ConocoPhilips, Equinor and Wintershall for
financing the work through the DrillWell research centre-
Drilling and Well Centre for Improved Recovery, a research
cooperation between NORCE, NTNU, SINTEF and UiS. We
appreciate Dr. A.-P. Bois for discussion of the results.

REFERENCES
[1] Vignes, B., and Aadnoy, B. S., 2008, “Well-Integrity
FIGURE 7: FRACTURE OPENING PROFILE IN DIFFERENT Issues Offshore Norway,” IADC/SPE Drilling Conference,
PRESSURE LEVEL FOR THE SINGLE FRACTURE Society of Petroleum Engineers.
https://doi.org/10.2118/112535-PA
It was also observed that the fracture opening is not uniform [2] Vrålstad, T., Saasen, A., Fjær, E., Øia, T., Ytrehus, J.
through the cement sheath and rock formation and the peak D., and Khalifeh, M., 2019, “Plug & Abandonment of
opening was in the rock formation at a distance from the Offshore Wells: Ensuring Long-Term Well Integrity and Cost-
cement-rock interface in this case. In this example, we observe

6 Copyright © 2019 by ASME

Efficiency,” J. Pet. Sci. Eng., 173, pp. 478–491. Operations: Part I,” SPE Drill. Complet., 24(01), pp. 127–136.
https://doi.org/10.1016/j.petrol.2018.10.049 https://doi.org/10.2118/106998-PA
[3] Bois, A.-P., Garnier, A., Rodot, F., Sain-Marc, J., and [8] Ravi, K., Bosma, M., and Gastebled, O., 2002,
Aimard, N., 2011, “How To Prevent Loss of Zonal Isolation “Improve the Economics of Oil and Gas Wells by Reducing the
Through a Comprehensive Analysis of Microannulus Risk of Cement Failure,” IADC/SPE Drilling Conference,
Formation,” SPE Drill. Complet., 26(01), pp. 13–31. Society of Petroleum Engineers. https://doi.org/10.2118/74497-
https://doi.org/10.2118/124719-PA MS
[4] Goodwin, K. J., and Crook, R. J., 1992, “Cement [9] Skorpa, R., Øia, T., Taghipour, A., and Vrålstad, T.,
Sheath Stress Failure,” SPE Drill. Eng., 7(04), pp. 291–296. 2018, “Laboratory Set-Up for Determination of Cement Sheath
https://doi.org/10.2118/20453-PA Integrity During Pressure Cycling,” Volume 8: Polar and Arctic
[5] Jackson, P. B., and Murphey, C. E., 1993, “Effect of Sciences and Technology; Petroleum Technology, ASME, p.
Casing Pressure on Gas Flow Through a Sheath of Set V008T11A039. doi:10.1115/OMAE2018-78696
Cement,” SPE/IADC Drilling Conference, Society of Petroleum [10] Alassi, H. T., 2008, “Modeling Reservoir
Engineers. https://doi.org/10.2118/25698-MS Geomechanicsusing Discrete Element Method:Application to
[6] Boukhelifa, L., Moroni, N., James, S. G., Le Roy- Reservoirmonitoring,” NTNU.
Delage, S., Thiercelin, M. J., and Lemaire, G., 2004, [11] Gheibi, S., Vilarrasa, V., and Holt, R. M., 2018,
“Evaluation of Cement Systems for Oil and Gas Well Zonal “Numerical Analysis of Mixed-Mode Rupture Propagation of
Isolation in a Full-Scale Annular Geometry,” IADC/SPE Faults in Reservoir-Caprock System in
Drilling Conference, Society of Petroleum Engineers. CO<inf>2</Inf>storage,” Int. J. Greenh. Gas Control, 71.
https://doi.org/10.2118/87195-MS https://doi.org/10.1016/j.ijggc.2018.01.004
[7] Gray, K. E., Podnos, E., and Becker, E., 2009, “Finite-
Element Studies of Near-Wellbore Region During Cementing

7 Copyright © 2019 by ASME