You are on page 1of 5

SPE/ISRM 78216

Application of Risk Analysis to Assessment of Casing Integrity and Subsidence in


Deepwater Reservoirs
A.S. Abou-Sayed, SPE, Advantek International Corporation; J.E. Noble, SPE, Advantek International Corporation;
Q. Guo, SPE, Advantek International Corporation; F. Meng, SPE, Advantek International Corporation

Copyright 2002, Society of Petroleum Engineers Inc.


Risk Analysis and Uncertainty
This paper was prepared for presentation at the SPE/ISRM Rock Mechanics Conference held
in Irving, Texas, 20-23 October 2002. Uncertainty about numbers is prevalent in all reservoir management
decisions. How large are the reserves of a proposed new exploitation
This paper was selected for presentation by an SPE/ISRM Program Committee following project? What will be the average market price for the reservoir
review of information contained in an abstract submitted by the author(s). Contents of the
paper, as presented, have not been reviewed by the Society of Petroleum Engineers or produced? What is the optimum production rate for the project? Each
International Society of Rock Mechanics and are subject to correction by the author(s). The of these numbers is an unknown quantity. If the profit or pay out from a
material, as presented, does not necessarily reflect any position of the Society of Petroleum proposed project strategy depends on various unknown quantities,
Engineers, International Society of Rock Mechanics, its officers, or members. Papers
presented at SPE/ISRM meetings are subject to publication review by Editorial Committees of then the final answer can not be computed without making some
the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part prediction of these unknown quantities.
of this paper for commercial purposes without the written consent of the Society of Petroleum
Engineers 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
A common approach to this problem is to make a "good-faith" point
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. estimate for each of these unknown quantities and use these estimates
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. to compute the possible benefit of each proposed project strategy. But
there is a serious problem with this method because it completely
Abstract ignores the inherent uncertainty the project faces. A better approach is
to incorporate uncertainty into the analysis, assess the randomness or
Various geomechanical-modeling approaches, which cover a wide probability distributions for any variable quantities, and analyze these
range of techniques and complexity, have been developed to assess variables in a spreadsheet model that simulates the project.
the stability of a borehole and/or the integrity of well casing. The ability
to confidently use these models can be limited, however, because they When we say that a random variable in a spreadsheet represents
generally do not allow the model user to consider the "real-world" some unknown real quantity, we mean that any event for this simulated
variability of the input parameters defined in the models. Often, these random variable is just as likely as the same event for the real
geomechanical models do not adequately accommodate the innate unknown quantity. For example, if the unknown number of competitive
variability of the rock properties (mechanical and petrophysical) of the entrants has a probability 0.10 of being 1, then the random variable in
target reservoirs. Consequently, this deterministic approach too often the spreadsheet should also have probability 0.10 of being 1 after the
results in uncertainty about the "correct" value of a critical parameter to next recalculation of the spreadsheet. For any number k, the
use and insecurity in the model results. Decisions based on these probability that the random variable will be less than k after the next
results can later, not surprisingly, be found to be incorrect. Model users recalculation should be the same as the probability that the real
attempting to overcome the limitations noted above have tried various unknown quantity is less than k.
techniques. Subjective estimation, arbitrary "minimums", grading
techniques, and stepwise estimation have all been commonly used. It should be noted that there have been various geomechanical-
modeling techniques developed in the past to assess the stability of a
Recently, more powerful techniques such as Monte Carlo simulation borehole and/or the integrity of well casing. McLellan and Hawkes
and decision analysis have come into popular use. Over the past (1998) have observed that these models cover a wide range of
decade these two techniques have been extensively used in the techniques and complexity. The ability to confidently use these models
petroleum industry to evaluate and solve a wide range of analytical can be limited, however, because they typically do not allow the model
problems in reservoir engineering and the geosciences. In this paper, user to consider the "real-world" variability of the input parameters
the application of these techniques to a number of generalized defined in the models. Risk and uncertainty are not considerations in
geomechanical problems will be illustrated. A Monte Carlo simulation the models. It is too common that these geomechanical models do not
enables the user to identify, measure or estimate, and evaluate make adequate allowances for the innate variability of the rock
uncertainties in the problems being analyzed. The simulation models properties (mechanical and petrophysical) of the reservoirs being
the random behavior of the input variables much like in a game of studied. Too often, this results in uncertainty about the "correct" value
chance. That is, the variables have an uncertain value within a known of a critical parameter to use and, consequently, insecurity in the
range for any particular time or event. Numerical model and software model results. Decisions based on these results can later, not
packages are developed based on the Green's function for a nucleus surprisingly, be found to be incorrect.
of strain in the reservoir. The model is coupled with reservoir
simulations to evaluate pressure maintenance and reservoir Model users attempting to overcome the limitations noted above have
development schedule effect on casing integrity, fault stability via tried various techniques. Subjective estimation, arbitrary "minimums",
sensitivity studies. The geomechanics solutions would be coupled with grading techniques, and stepwise estimation have been commonly
both commercial or in-house developed reservoir simulators. The used techniques. Recently, more powerful techniques such as Monte
results of multiple simulations, done to determine the most likely Carlo simulation and decision analysis have come into popular use. It
outcome of various wellbore solution options, are reported in the is apparent from the literature that over the past decade these latter
current paper.
2 A.S. ABOU-SAYED, J.E. NOBLE, Q. GUO AND F. MENG SPE/ISRM 78216

two techniques have been used in the petroleum industry for When each input variable has a range of possible values with an
evaluating and solving a wide range of analytical problems. Of the infinite number of values within that range, the possible values can be
two, the use of the Monte Carlo simulation technique has represented by a continuous probability distribution function. The type
predominated. In the following discussion we illustrate the application of distribution you select is based on the conditions surrounding that
of these two techniques to a number of generalized geomechanical variable. Distribution types typically are characterized as normal,
problems. Noted in the reference section of this paper are examples of triangular, uniform, and lognormal (each illustrated in the Figure 3
how various techniques have been applied by other authors to solve below). Such distributions may exist from prior analyses or can be
uncertainty problems. developed using statistical analysis techniques.

A simulation refers to any analytical method used to duplicate a real-


world system. These methods are especially useful when other
analyses are too complex or too difficult to reproduce. A Monte Carlo
simulation enables the user to identify, measure or estimate, and
analyze uncertainties in the problem(s) being analyzed. Typical
problems contain random variables whose values vary spatially and
with time. The term stochastic is often used to describe these
techniques; while deterministic is used to define the analysis of fixed-
value variables. The random behavior in games of chance is similar to
how Monte Carlo simulation selects variable values at random to
simulate a model. When you roll a die, you know that 1, 2, 3, 4, 5, or 6
will come up; however, you don't know which one for any particular roll.
It's the same with the variables that have a known range of values but
an uncertain value for any particular time or event.

Probability distributions can be used to describe the distribution of


some unknown quantity. When an unknown quantity has a definite
number of possible values, it can describe using a discrete probability
distribution (Figure 1). These values are often presented in a table that
lists the possible values of the unknown quantity and the probability of
each possible value.

0.35

0.30
Probability of Outcomes

0.25

0.20

0.15

0.10

0.05

0.00
1 2 3 4 5 6
Possible Outcomes

Figure 1. Discrete Probability Distribution

Using these data, a cumulative probability distribution curve can also


be constructed (figure 2). The cumulative-probability curve is usually
more useful for describing probabilities than point-probability bars. In
fact, point-probability bars cannot be applied to continuous probability
distributions, where there are an infinite number of possible values.
Figure 3. Continuous Probability Distribution Functions (PDFs)

1.00
Using a risk analysis approach, we can illustrate how the deterministic
0.90
and stochastic analyses of subsidence will differ. Using the following
0.80
fixed input variables we first determine a single expected value for
Cumulative Probablity

0.70
subsidence.
0.60

0.50
Pore Volume Compressibility = 38.0E-6 / psi
0.40
Porosity = 30 percent
0.30

0.20
Reservoir Thickness = 200 feet
0.10
Depth to Reservoir Center = 10000 feet
0.00
Reservoir Radius = 2000 feet
1 2 3 4 5 6 Poisson's Ratio = 0.3
Possible Outcomes Depletion = 5000 psi

The calculated amount of estimated subsidence was 3.10E-1 feet.


Figure 2. Cumulative Probability Distribution
APPLICATION OF RISK ANALYSIS TO ASSESSMENT OF CASING INTEGRITY
SPE/ISRM 78216 AND SUBSIDENCE IN DEEPWATER RESERVOIRS 3

G
Next, we apply the Monty Carlo simulation approach to calculate an
expected range of values for subsidence and the probability associated R1 = (x − x ' , y − y ' , z − z ')
with each value within that range. First, the fixed input variables from G
the deterministic analysis above are assumed to be the most likely R2 = (x − x' , y − y ' , z − z ')
value in a triangular probability distribution function of least possible G
value, most likely value, and highest possible value for each variable. R1 = R 1 = ( x − x ' ) 2 + ( y − y ' ) 2 + ( z − z ' ) 2 (3a)
These other terms are noted as follows:
G
Compressibility = {25.0, 38.0, 45.0}E-6 R2 = R2 = ( x − x ' ) 2 + ( y − y ' ) 2 + ( z − z ' ) 2 (3b)
Porosity = {25, 30, 33} percent
Reservoir Thickness = {150, 200, 250 } feet
Required data such as porosity, compressibility, initial pressure and
Depth to Reservoir = {9500, 10000, 10500} feet
pressure at a given time for each of the blocks can be obtained from
Reservoir Radius = {1500, 2000, 2500} feet
reservoir simulation. Displacement at any location (x,y,z) can be
Poisson's Ratio = 0.3
obtained by performing the numerical integration according to Equation
Depletion = {4500, 5000, 5500} psi
(1).
The cumulative probability of expected subsidence at P90, P50, and
P10 were calculated to be, respectively, 1.97E-1 feet, 2.83E-1 feet,
and 3.85E-1 feet. Instead of a single value being determined, the
Casing Strain Evaluation
risked analysis gives a more useful range of expected subsidence
values with the cumulative probability of each value occurring. Once the displacements are obtained, the induced strains in the rock
formation can be calculated from displacements. The six casing-strain
components can be computed through the continuity conditions along
Subsidence Evaluation the interface between the casing and the rock formation. It should be
While understanding the potential of subsidence occurring is import, it noted that, casing is not considered in obtaining the
is the impact of subsidence on well casing and casing integrity that is a displacements/strains in the rock formation, and strictly speaking, the
primary concern. Displacements caused by reservoir production and calculated displacements/strains only give the upper bound of casing
injection can be computed by coupling reservoir simulation with strain.
geomechanics. Suppose that the reservoir consists of many blocks,
which can be obtained from reservoir simulations. Using The continuity conditions at the interface between the casing and the
superposition, the displacement ui in the xi–direction (i = 1,2,3), based formation are that three displacement components and the three
on Mindlin and Cheng (1950) solution for source of dilation, can be surface traction components across the interface are continuous.
obtained as: Since ur, uθ and uz (see Figure 1) are continuous across the interface,
G G G G the casing strain components e’zz, e’θθ, e’θz can be computed from
u i (x , t ) = − ∫ ∫ ∫ u i* (x , x ' , t )∆ P (x , t )dx ' dy ' dz ' strains ezz, eθθ, eθz in the rock formation as:
V
(1) e’zz = ezz
N
G G G e’θθ = eθθ (4)
= −Σ ∫ ∫ ∫ ∆P ( x , t )u (x , x ', t )dx ' dy ' dz '
*
k i e’θz = eθz
k =1 Vk
where: Since the three traction components σrr, σrθ and σrz (see Figure 1) are
continuous across the interface, the casing strain components e’rr, e’rθ,
N is the total number of reservoir blocks
e’rz can be computed from strains eij (i,j = 1,2,3) in the rock formation
Vk is the volume of block K as, assuming elastic deformation both in the casing and in the rock:

∆Pk( t ) = Pk( t ) – Pk( t = 0) is the difference in formation v' G 1 − 2v '  v  α∆P


e'rr = − (eθθ + ezz ) + err + (err + eθθ + ezz ) −
pressure of block K 1 − v' G ' 1 − v'  1 − 2v  2G '
G G e’rθ = G erθ
 R1 R2 6 z ( z + z ' ) G G'
G G
i ( C
u x, x1, t = m
*


)  2 + (3 − 4 v ) 2 − 5
R2 e’rz = G erz
 R1 R2 R2 G'
(2)
G Where G’ is the shear modulus of the casing and G is the shear
2K  modulus of the rock; ν' is the Poisson's ratio of the casing. ∆P is the
− 2 [(3 − 4v)( z + z ' ) − z ' ] change of the formation pressure. It should be noted that, for the
R2  convenience of expression, we dropped the prime notation in the
above equations.
where:
α 1+V
Cm = φC p is the uni-axial compressibility
3 1 − Vp
α is the Biot’s coefficient of the reservoir rock
v is rock Poisson’s ratio
φ is porosity of the reservoir rock
Cp is the pore volume compressibility
4 A.S. ABOU-SAYED, J.E. NOBLE, Q. GUO AND F. MENG SPE/ISRM 78216

If that is not the case, or reasonably so, remembering again that this is
a screening tool, you might make some parametric variations.
r’
θ
Overburden Poisson's Ratio is a mechanical elastic property of the
overburden. By definition, it is the negative ratio of the change in strain
in one direction to the strain in an orthogonal direction. If you consider
this in terms of a core test, Poisson's ratio is the negative ratio of the
radial strain divided by the axial strain and loading of the core sample
z’ in the axial direction causes the axial strain. Strictly from elasticity, this
property varies between 0 and 0.5.

Depletion refers to overall reduction in the average reservoir pressure.


Remember, that at a producing well, you will have a drawdown
superimposed on the depletion. In general, you would assume that the
depletion governs the overall subsidence but that the compaction-
related difficulties that a wellbore might experience also need to
consider the drawdown.

Figure 4. Local polar coordinates associated with Overburden Young's Modulus is a mechanical elastic property of the
casing/wellbore overburden material. It indicates how much a material will deform when
it is stressed. A material with a high Young's modulus will not deform
Risked Casing Analysis as much as one with a low Young's modulus. Remember that steel has
6
a Young's modulus of 30 x 10 psi. For the Gulf of Mexico, you might
The calculated displacements analyzed by casing risk include the 6
assume a default overburden modulus of 1.5 x 10 psi, but this is
maximum surface subsidence and the vertical displacement at the something that needs to be evaluated very carefully.
intersection of well casing and the reservoir top. As noted above, these
displacements are caused by reservoir compaction due to pressure Reservoir Poisson's Ratio is a mechanical elastic property of the
drop inside the reservoir, and the resulting strain components are reservoir. By definition, it is the negative ratio of the change in strain in
calculated at the reservoir top (inside overburden), at both the well one direction to the strain in an orthogonal direction. If you consider
casing location and reservoir flank. While the maximum surface this in terms of a core test, Poisson's ratio is the negative ratio of the
subsidence occurs at the top the reservoir center, the maximum radial strain divided by the axial strain and loading of the core sample
displacement normally occurs at the intersection of wellbore and the in the axial direction causes the axial strain. Strictly from elasticity, this
reservoir top. The most critical strain will occur at either the intersection property varies between 0 and 0.5.
of the wellbore and the reservoir top or at the flank of the reservoir.
This is dependent on the ratio of the reservoir radius to the thickness. Angle of Well is the angle of the hole from the vertical.
Among the three normal strain components (please see list below), σz'z'
• Calculated displacements include the maximum surface
is usually the largest strain component especially at the intersecting
subsidence and the vertical displacement at the intersection
point of the wellbore and the reservoir top of a vertical well.
of a well trajectory and the reservoir top.

List of Terms used in Risked Casing Analysis • The displacements shown are estimates and are caused by
reservoir compaction alone.
Pore Volume Compressibility Cp can be determined from laboratory
testing or from logging inferences (for example, the program @Log
TM • The strain components reported are at the reservoir top (just
can be used). The basis definition is as follows: inside of the overburden), both at the reservoir center and at
a point on the reservoir's flanks.
1 ∂V p • εz'z' is the normal strain along the wellbore axis.
Cp =
V p ∂ρ • εr'r' is the normal strain in the direction vertical to the wellbore
axis.
where you are looking at the change in the pore volume as you change • εΦ'Φ' is the third normal strain which is perpendicular to
the pore pressure. NOTE: This is not the Compaction Coefficient that
εz'z'and εr'r'
is sometimes used in subsidence calculations.
• ∂r'z' is the shear strain in the plane perpendicular to the "r' "
Porosity is the porosity at the initial in-situ conditions that are being direction and the plane perpendicular to the wellbore axis.
considered.
• The maximum displacement normally occurs at the
Depth to Reservoir Center is the true vertical depth to the reservoir intersection of the wellbore and the reservoir top. For this
center. symmetric prediction, the maximum surface subsidence
occurs above the reservoir center.
Reservoir Thickness is the average reservoir dimension in the • The most critical strain that a completion will be exposed to
vertical direction. Use your judgment for lenticular scenarios. Note that in the overburden immediately above the reservoir often
this will be the net thickness, unless you have included a pore volume occurs at the intersection of the wellbore and the reservoir
compressibility that accounts for shale dewatering as well as primary top or at the flank of the reservoir depending the ratio of the
compressibility. reservoir radius and the thickness. Buckling and crushing in
the reservoir itself may be equally severe problems, or
Reservoir Radius refers to the average lateral extent of the reservoir. possibly more so.
Note that since this is a screening tool, it is premised on a uniform
reduction in reservoir pressure throughout the entire reservoir volume. • Among the three normal strain components, εz'z' is usually
the largest, especially at the wellbore for a vertical well.
APPLICATION OF RISK ANALYSIS TO ASSESSMENT OF CASING INTEGRITY
SPE/ISRM 78216 AND SUBSIDENCE IN DEEPWATER RESERVOIRS 5

References 7. Keilty, I.D. and Rabia, H.: “Applying Quantitative Risk


Assessment to Casing Design,” SPE paper 35038 presented
1. Baris, G. and Home, R.N. : “Uncertainty Assessment of at the 1996 IADC/SPE Drilling Conference, New Orleans,
Well Placement Optimization,” paper SPE 71625 presented Louisiana, March 12-15.
at the 2001 SPE Annual Technical Conference and
Exhibition, New Orleans, LA, September 30-October 3. 8. Manceau, E., Mezghani, M., Zabalza-Mezghani, I. and
Roggero, F.: “Combination of Experimental Design and
2. Bruno, M.S.: “Geomechanical Analysis and Decision Joint Modeling Methods for Quantifying the Risk
Analysis for Mitigating Compaction Related Casing Associated With Deterministic and Stochastic Uncertainties
Damage,” paper SPE 71695 presented at the 2001 SPE - An Integrated Test Study,” paper SPE 71620 presented at
Annual Technical Conference and Exhibition, New the 2001 SPE Annual Technical Conference and
Orleans, LA, September 30-October 3. Exhibition, New Orleans, La, September 30-October 3.
3. Charles,T., Guéméné, J.M., Corre, B., Vincent, G., and 9. McLellan, P.J. and Hawkes, C.D.: “Application of
Dubrule, O.: “Experience with the Quantification of Probabilistic Techniques for Assessing Sand Production
Subsurface Uncertainties,” paper SPE 68703 presented at and Borehole Instability Risks,” paper SPE 47334
the 2001 SPE Asia Pacific Oil and Gas Conference and presented at 1998 SPE/ISRM Eurock 98, Trondheim,
Exhibition, Jakarta, Indonesia, April 17-19. Norway, July 8-10.
4. Geertsma, J., 1973, “A Basic Theory of Subsidence Due to 10. Mindlin, R.D., and Cheng, D.H., 1950, “Thermoelastic
Reservoir Compaction: the Homogeneous Case,” stress in a semi-infinite body,” J. of Appl. Physics, Vol. 21,
Verhandelingen Kon. Ned. Geol. Mijnbouwk. Gen., Vol. pp. 931-933.
28, pp. 43-62.
11. Proehl, T.S.: “Pore Pressures, Fracture Gradients, and
5. Hegdal, T., Dixon, R.T. and Martinsen, R.: “Production Drilling Economics,” paper SPE 27493 presented at the
Forecasting of an Unstable Compacting Chalk Field Using 1994 IADC/SPE Drilling Conference, Dallas, Texas,
Uncertainty Analysis,” paper SPE 64296 was revised for February 15-18.
publication SPEREE (June 2000). Original manuscript SPE
50644 presented at the 1998 SPE European Petroleum 12. Zabalza-Mezghani, I., Mezghani, M. and Blanc, G.:
Conference, The Hague, October 20-22. “Constraining Reservoir Facies Models to Dynamic Data -
Impact of Spatial Distribution Uncertainty on Production
6. Jonkman, R.M., Bos, C.F.M., Breunese, J.N., Forecasts,” paper SPE 71335 presented at the 2001 SPE
Morgan,D.T.K., Spencer, J.A., and Søndenå, E.: “Best Annual Technical Conference and Exhibition, New
Practices and Methods in Hydrocarbon Resource Orleans, La, September 30-October 3.
Estimation, Production and Emissions Forecasting,
Uncertainty Evaluation and Decision Making,” paper SPE
65144 presented at the 2000 SPE European Petroleum
Conference, Paris, France, October 24-25.

You might also like