IADC/SPE

IADC/SPE 19941

Wellbore Stability Analysis: A Review of Current Methods of

Analysis and Their Field Application
M.R. McLean and M.A. Addis,* British Petroleum
*SPE Member

Copyright 1990, IADC/SPE Drilling Conference.

This paper was prepared for presentation at the 1990 IADC/SPE Drilling Conference held in Houston, Texas, February 27-March 2, 1990.

This paper was selected for presentation by an IADC/SPE Program Committee following 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 the International Association of Drilling Contractors and are subject to correction by the author(s).
The material, as presented, does not necessarily reflect any position of the IADC or SPE, its officers, or members. Papers presented at IADC/SPE meetings are subject to publication
review by Editorial Committees of the IADC and SPE. Permission to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should
contain conspicuous acknowledgment of where and by-whom the paper is presented. Write Publications Manager, SPE, P.O. Box 833836, Richardson, TX 75083-3836. Telex, 730989 SPEDAL.

ABSTRACT effects are considered here. Time dependent effects

Hole problems during the drilling phase of operations
are often the consequence of mechanical wellbore
instability. This leads to higher than necessary drilling
costs. A number of analytical and numerical models are
available for the diagnosis and prediction of wellbore
instability; this paper reviews the merits and pitfalls of
applying these models to field situations. Attention is
focused on the peak-strength criterion and constitutive
behaviour model. Anomalies from the incorporation of
the intermediate principal stress into peak-strength
criteria are highlighted.· To illustrate the reliability of
a number of models, their predictions are compared with
laboratory results, and with a case history of a horizontal
well drilled in the Cyrus Field in the North Sea.
1.0 INTRODUCTION
The increasing demand for wellbore stability analyses
during the planning stage of a field arises from economic
considerations and the escalating use of deviated,
extended reach and horizontal wells. Wellbore insta-
bility can result in lost circulation (Figure 1a) where
tensile failure has occurred, and spalling and/or hole
closure (Figure 1b) in the case of compressive failure. In
severe cases the hole instability can lead to stuck pipe
and eventually loss of the open hole section. The causes
of instability are often classified into either chemical or
mechanical effects. Often, field instances of instability
are a result of a combination of both chemical and
mechanical effects. However, only the mechanical
References and illustrations·at end of paper
resulting from pore fluid migration (e.g. Detournay &
Cheng [1]) are not considered either.
A number of publications on the subject of mechanical
wellbore stability can be found in the literature; how-
ever, only a few actually attempt to predict the stability
of a field case. Most workers concentrate on specific
aspects of an analysis, e.g. in-situ stress determination,
stress concentrations around a borehole, rock mechan-
ical properties etc.. This paper reviews some of the
elements which go into the development of a wellbore
stability model, and applies the model to a field case.
The two main elements required in a wellbore stability
model are the failure criterion and the constitutive
behaviour model. A number of previously used criteria
and behaviour models are reviewed, and their suitability
assessed. Results are presented from a pre-spud
analysis of a horizontal well drilled in 1988 in the Cyrus
Field, North Sea. Comparisons are made between a
linear-elastic analysis and a Finite Element Method
(FEM) analysis using a constitutive model considered
more representative of the reservoir rock. The well
response during drilling indicated that the predictions
of the FEM were significantly more accurate than the
linear-elastic analysis.
2.0 BACKGROUND TO WELLBORE STABILITY
MODELLING
Before a well is drilled, compressive stresses exist
within the rock formations (Figure 2). The stresses can
be resolved into a vertical or overburden stress, Svert, and

261
2 WELLBORE STABILITY ANALYSIS: A REVIEW OF CURRENT METHODS OF .....
two horizontal stresses, ~max (the maximum horizontal
stress), and ~min (the minimum horizontal stress),
which are generally unequal. When the well is drilled,
the rock stresses in the vicinity of the wellbore are
redistributed as the support originally offered by the
drilled out rock is replaced by the hydraulic pressure of
the mud. The redistributed stresses are normally
referred to as the hoop stress, cr9 , which acts circum-
ferentially around the wellbore wall, the radial stress,
crr, and the axial stress, crz, which acts parallel to the
wellbore axis (see Figure 6 for stress state within a
hollow cylinder). In deviated wells an additional shear
component, 'tez, is generated.
If the redistributed stress state exceeds the rock
strength, either in tension or compression, then insta-
bility may result (Figure 1). In order to evaluate the
potential for wellbore stability a realistic constitutive
model must be used to compute the stresses and/or
strains around the wellbore. The computed stresses and
strains must then be compared against a given failure
criterion.
3.0 WHICH STRENGTH CRITERION?
In a purely elastic analysis the stresses are compared
against a peak-strength criterion normally defined in
terms of the principal stresses. In an elasto-plastic
analysis, plastic strains are developed once the stress
state reaches a yield criterion, which in the case of
perfect-plasticity coincides with the peak-strength cri-
terion.
An elasto-plastic analysis of wellbore stability is more
realistic than a simple elastic analysis, since rocks rarely
behave in a purely elastic manner up to ultimate failure.
However, specifying the allowable extent of the plastic
deformation before instability occurs, is difficult and
somewhat arbitrary (e.g. Antheunis et al [2]).
In cases where well defined rock properties are obtained
from laboratory testing of core, more sophisticated
numerical analyses incorporating non-linear aniso-
tropic material behaviour may be performed to evaluate
wellbore stability (e.g. Morita & Gray [3]). However, in
the majority of cases, the poor definition of input para-
meters (in-situ stresses and strengths) only justifies a
simple elastic analysis at best. In these cases, rock
failure is determined using a peak-strength criterion.
The peak-strength criterion can be determined directly
from laboratory testing (if core is available) or from back
analysis of hole conditions recorded from caliper and
drilling logs (bearing in mind that this assumes a
knowledge of the in-situ stresses, and the likely stress
concentration around the hole).
A number of different strength criteria are commonly
262
IADC/SPE 19941
used to predict the onset of rock failure and wellbore
instability. These criteria fall into one of the four
categories (A,B,C & D) shown in Table 1.
Examples of publications which use the criteria from the
different categories are:
Category A- Woodland [7], Fuh et al[S]
Category B - MitChell et al [9], Hsaio [10], Aadn~~Jy &
Chenevert [11], Gnirk [12]
Category C - Bradley [13], Hottman et al [14],
Nakken et al [15], Marsden et al [16]
Category D - Santarelli [17], Kwakwa et al [18]
3.1 Effect of the Intermediate Stress
The question of whether the intermediate principal
stress should or should not be incorporated into a failure
criterion is an old one, and one which is still apparently
unresolved, as witnessed by the continuing investiga-
tions on the subject.
The intermediate principal stress would appear to have
some effect on rock strength as seen in true triaxial
testing (e.g. Mogi [19] and Takahashi & Koide [20]). The
variation of strength with the intermediate principal
stress found by Takahashi & Koide was investigated for
a number of rock types. In the case of the Yamaguchi
marble, cr2 had the greatest influence where cr3 =20 MPa
and cr1 > cr2= 8cr3 • For this stress state the maximum
principal stress at failure was approximately 65% higher
than the standard triaxial test strength for the same
value of cr3 • The influence of cr2 on the strength was less
marked for the other rocks tested, which included three
sandstones and a shale. The results from Mogi's tests
showed cr2 had a similar influence on strength. From
references [19] and [20] it is reasonable to expect that
the compressive strength of a rock sample tested biax-
ially (where cr1=cr2 and cr3 =0) is unlikely to be more than
twice its uniaxial strength (i.e. cr1 > 0 2 =cr3 =0).
From the above discussion, it is informative to check
various strength criteria against the two stress paths
(i.e. uniaxial and biaxial). For strength criteria which
fall into Categories B & D, Table 1, the uniaxial and
biaxial strength are the same. For criteria which fall
into Categories A & C the difference is often extreme, as
shown in Figure 3, which presents strength data for a
Gebdykes dolomite tested by Santarelli [17]. The data
has been converted from a cr1 - cr3 stress space to a 'toct-aoct
stress space (as used in references [13] & [14]), where
'toct and aoct are defined as
'toct = 1~ (crt- cr2)2+ (02- cr3)2+ (03- crt)2
3 (1)
(2)
Both the uniaxial and biaxial stress paths are plotted in
IADC/SPE 19941 M.R. McLEAN, M.A. ADDIS
Figure 3. The biaxial strength predicted by the failure
envelope exists outside the range of the experimental
data. However, from extrapolation the predicted biaxial
strength is likely to be around 6 times the uniaxial
strength, which appears excessive in the light of the
previous discussion.
Nakken et al [15] and Marsden et al [16] express their
criteria in terms of the q-p stress space as used in critical
state soil mechanics. These stress invariants are defined
as
(3)
(4)
For standard triaxial testing (where cr2 =cr3 ) the general
form of p can be expressed as
(5)
Figures 4 and 5 show the strength criteria given in q-p
space for claystones tested by N akken et al and Marsden
et al (Note: These plots are actually in terms of the mean
effective pressure,p', since some of the tests were carried
out with non-zero pore pressures). For this stress space,
the predicted biaxial strengths in Figures 4 & 5 are
approximately 10 and 20 times the uniaxial strengths,
respectively.
The major problem with many of the criteria which fall
into Categories A & C, Table 1, are that they give far too
great a significance to the influence of cr2 on the strength
of frictional materials than is indicated by true triaxial
testing. Mogi [19] showed that if cr2 was to be incorpo-
rated into a failure criterion for competant rock, then
the 'mean' stress term, p, should be adapted to the
equation given below
(6)
where the factor n typically takes on values of around
0.1. Green and Bishop [21] arrived at similar conclusions
based on experiments performed on sands.
Despite Mogi's work and the irreconcilable differences
between the predicted uniaxial and biaxial strengths,
researchers continue using strength criteria defined in
terms of q-p and 'toct-O'oct·
It is our conclusion that a strength criterion expressed
in terms of cr1 and cr3 is adequate for the purposes of
wellbore stability. Although the intermediate principal
stress may have some influence, the effect is small
relative to the accuracy to which down-hole strength and
in-situ stresses can be determined.
263
3
3.2 Linear or Non-Linear?
Restricting our attention to failure criteria grouped in
Categories B & D, Table 1, we now consider the question
of whether a linear or non-linear strength criterion is
required. A number of researchers go to some length to
fit a non-linear criterion to triaxial test data carried out
over a wide range of confining pressures. Kwakwa et al
[18] back-analyse field conditions to produce a Hoek-
Brown failure criterion which is plotted for minimum
principal stresses up to 10,000 psi (70 MPa). This is
warranted when the minimum principal stress varies
considerably throughout the region ofinterest. In elastic
analyses ofwellbore stability we are concerned only with
the state of stress at the wellbore wall (in some specific
cases, e.g. underbalance drilling, it may also be
necessary to look at points just inside the wellbore wall).
In the case of a wellbore, the minimum effective stress
is invariably the overbalance pressure (well pressure
less formation pressure), and is generally in the region
of0-1,000 psi (0-7 MPa). In extreme cases the overbal-
ance pressure may be as high as 2,000 psi (14 MPa). As
such, there is no requirement to adapt criteria to fit
peak-strength data for confining pressures greater than
2,000 psi (14 MPa). Over this small range of confining
pressures a linear failure criterion is more than
adequate for all but the weakest formations.
In conclusion, we consider that a linear failure criterion
which incorporates only the maximum and minimum
principal stresses (i.e. Mohr-Coulomb) is the most
applicable in a wellbore stability analysis. For very weak
formations (uniaxial strength less than 1,500 psi (10
MPa)) a non-linear criterion may be justified. Any
allowance for the effect of the intermediate principal
stress can result in gross overpredictions of strength
(depending on the stress path) and should be checked
thoroughly.
4.0 WHICH CONSTITUTIVE MODEL?
There are many published constitutive models used to
determine the stress state around a wellbore. It is not
reasonable to list them all, thus, only those considered
the most indicative are discussed here. Table 2 sum-
marises these models. The models assume homogeneity
and isotropy unless stated.
Given a strength criterion expressed in terms of cr1 and
0'3 , it is known from laboratory testing that small
diameter boreholes in rock (usually around 1" dia.) are
far stronger than predicted by a linear-elastic analysis
(e.g. Santarelli [17], Guenot [27]). It is generally thought
that wellbores are also stronger than predicted by
linear-elasticity. Few publications, however, have made
the comparison between the predicted response of a well,
based on laboratory determined rock properties, and
actual response during drilling. One example, although
inconclusive, is presented by IGein and McLean [28].
4 WELLBORE STABILITY ANALYSIS: A REVIEW OF CURRENT METHODS OF .....
In the case oflaboratory scale wellbores, references [17]
and [27] show that a variety of hollow cylinder rock
samples, tested under conditions detailed in Figure 6,
start to fail at outer pressures of between 2 to 8 times
the failure pressure predicted by linear-elasticity and a
Category B or D criterion (Table 1). Where the failure
criterion is defined in either a q-p or 'toct-O'oct space, the
predicted strength of the hollow cylinder is often greater
than the true strength. For example the stress path
followed at the internal wall of the hollow cylinder
depicted in Figure 6 is plotted in Figure 3 for the
Gebdykes dolomite. It can be shown that the intersection
point between the failure criterion and the stress path
is predicted when the pressure applied to the hollow
cylinder is around 130 MPa (18,500 psi). Actual hollow
cylinders of this rock tested by Santarelli [17] suffered
internal wall failure at an outer pressure of 52 MPa
(7 ,500 psi). In certain cases a 't e-O'oce criterion used in
00
association with a linear-elastic constitutive model can
give accurate predictions of hollow cylinder failure, an
example ofwhich is given by Veeken et al [25]). However,
as inferred by the authors, this is likely to be pure
coincidence, rather than sound modelling.
Armed with the knowledge that the use of linear-elas-
ticity underpredicts hole stability (using Category B &
D type criteria) the main thrust in analysis is to utilise
models which are less conservative in their predictions.
To this end plasticity offers an obvious, and commonly
used, behaviour for improving predictions. Westergaard
[24] was one of the first to utilise an elasto-plastic
approach to the analysis of a wellbore; the post yield
behaviour being modelled using perfect-plasticity. More
recent models still use perfect-plasticity due to its
modelling simplicity (e.g. Mitchell & Goodman [9], and
Bratli & Risnes [29]). However, rocks are rarely able to
sustain large amounts of permanent deformation
without a change in strength, particularly at the rela-
tively low values of minimum effective stress, o 3',
encountered at the wall of a wellbore. Also, some limit
to the amount of permanent deformation must be
assigned, which is often arbitrary in nature. In effect,
a borehole can have any strength the analyst desires
depending on the chosen allowable plastic deformation.
More realistic attempts to model plastic behaviour
around boreholes by incorporating both strain-harden-
ing and strain-softening are made by a number of
researchers, notably Morita & Gray [3] and Veeken et
al [25]. Morita and Gray only presented an analysis,
while Veeken et al went one step further and compared
their predictions against hollow cylinder tests for which
they found good agreement. However, they used a
Drucker-Prager criterion (scribed within the
Mohr-Coulomb equivalent) and the plastic deformation
at which failure is deemed to occur is somewhat arbit-
rary. The use of strain-softening behaviour is also open
to criticism, since it is normally associated with localised
deformation of uniaxial and triaxial samples, whereas a
264
IADC/SPE 19941
homogeneous deformation is assumed in the develop-
ment of the constitutive model. Despite this, strength
degradation undoubtedly occurs with continued
permanent deformation and the modelling of
strain-softening should not be ruled out altogether.
Perhaps the biggest problem associated with an elas-
to-plastic model incorporating strain-softening, is the
robustness of the algorithm. Numerical instabilities can
be generated within the computations, which lead to
non-uniqueness of the results, or non-convergence in the
algorithm.
A rigid-plastic constitutive model has been incorporated
extensively in a bifurcation analysis of borehole failure
(e.g. [26], [30] & [31]). The model assumes that all shear
strain is permanent, which is acceptable providing the
material does not attempt to unload elastically. This
constitutive model coupled with bifurcation analysis has
been used in the above references to provide extremely
accurate predictions of hollow cylinder failure. The
model, although somewhat complex, has the advantage
· of only requiring uniaxial test data for defining the
constitutive behaviour (see Figure 7). However, it would
be instructive to compare its predictions of triaxial
response against test data to assess whether the model
is truely representative of rock behaviour over the range
of relevant stress states. Providing triaxial data sup-
ports the constitutive model then it would appear to be
the most powerful predictive model published to date.
4.1 Recent Developments
Santarelli [17], amongst others, noted that the elastic
modulus for rocks determined from uniaxial/triaxial
testing increased with confining pressure (e.g. Figure 8).
The rate of increase is particularly marked at low
confining pressure. He incorporated the variation of
elastic modulus with confining pressure into a consti-
tutive behaviour model assuming a power law rela-
tionship between the secant modulus, E and the 8
,
minimum principal stress, o 3 , given by
(7)
(8)
where E 0 is the uniaxial modulus and A and b are
constants.
The minimum principal effective stress, o3', is assumed
when non-zero pore pressures are present, where
effective stress is defined by total stress minus pore fluid
pressure.
In addition to the elastic modulus variation, Santarelli
incorporated pre-peak yielding into the constitutive
model to predict the stress state developed within hollow
cylinders and obtained an improved prediction of the
IADC/SPE 19941 M.R. McLEAN, M.A. ADDIS
failure pressure for three rock types (see Table 3). The
analysis assumed that failure occurred when a stress
point at the wall of the hollow cylinder reached its
uniaxial peak-strength.
Santarelli's model provides better predictions of failure
compared to linear-elasticity and simple elasto-plastic
analyses, although it still underestimates the pressure
at which failure is initiated within the hollow cylinders.
It is easy to assign the difference between the actual
failure and true failure to the influence of the inter-
mediate principal stress. However, another factor which
has received attention recently, and may account for the
'abnormal' strength of hollow cylinders is the effect of
scale.
Laboratory testing showing the effects of scale on the
failure around a circular opening are presented by
Antheunis et al [2] and Haimson and Herrick [32].
Antheunis showed some scale effect, but the limitation
of the testing equipment made the results inconclusive.
However, Haimson and Herrick's tests on a Alabama
limestone showed a very consistent trend over a range
ofhole sizes (Figure 9b). These tests indicate that beyond
a certain hole size (in this case around 6 em) a
linear-elastic analysis coupled with a simple failure
criterion is perfectly adequate for predicting the onset
failure of the borehole wall. Below this hole size, scale
effects may enhance the strength of the opening con-
siderably. Therefore, it should be considered dangerous
to make quantitative predictions of the stability of full
scale wellbores based on laboratory simulations, which
are often carried out on holes of 1" diameter or less,
unless the effect of scale can be quantified.
Despite the evidence presented by Haimson and Herrick
suggesting that, in the case of full scale wellbores,
linear-elasticity gives an adequate description of the
stress state, their test results do not contradict the use
of Santarelli's model which utilises a confining pressure
dependent modulus. This is due to the stress path
adopted within their test samples. The circular holes
were formed within blocks which were loaded in one
direction only (see Figure 9a). Under these conditions
the minimum principal stress is either zero or tensile
throughout the sample. In this case, assuming a
modulus variation given by equations (7) and (8), E 8 is
constant throughout the sample and equivalent to
linear-elasticity. Thus, Santarelli's model and
linear-elasticity are equivalent for this loading path. (It
may be argued that Es is less than E 0 for tensile values
of 0'3 • However, the rate of change of modulus must
quickly reduce in the tensile region, otherwise Es would
become zero or negative.)
In conclusion, it is difficult to assess the quality of a
constitutive model in its application to wellbore stability
analysis due to a number of uncertainties, in particular
265
5
the effect of scale. To correctly judge the quantitative
predictions of consititutive models and failure criteria
we need to have a better understanding of any scale
effects inherent in laboratory testing. Two constitutive
models which appear to improve predictions of hollow
cylinder failure relative to a linear-elastic analysis are
those presented by Sulem and Vardoulakis [26] and
Santarelli [17]. In the field application given in the
following section, the model proposed by Santarelli is
used. A brief description of how the model is incorpo-
rated into an FEM analysis is given by Duncan-Fama
and Brown [33]. A full description is found in [34].
5.0 FIELD APPLICATION: CASE HISTORY
5.1 Background
The Cyrus Field in the UK Continental Shelf (UKCS),
Block 16/28, will be developed using the Single Well Oil
Production System (SWOPS). Development will be from
two horizontal wells drilled from a single location. The
first well was drilled in 1988, and the second in 1989.
Prior to drilling the first horizontal well a study was
carried out to assess the stability of the 8 1/2", 600m long
horizontal section through the reservoir. Figure 10
shows the lithology, planned well profile and casing
programme.
Previous exploration and appraisal wells were success-
fully drilled through the reservoir with a 1.15 S.G.
density mud. In order to limit the potential for formation
damage it was considered necessary to restrict the mud
weights while drilling the reservoir section of the hori-
zontal wells to a similar mud density. The rock mech-
anics study was performed to determine whether mud
weights of around 1.15 S.G. would be sufficient to
prevent spalling of the wellbore wall.
5.2 Approach
The use of numerical/analytical models to predict the
mechanical behaviour of a wellbore requires a number
of input parameters to be defined or assessed. The input
parameters required to fully utilise the models used in
this study are the in-situ stress state (magnitudes and
directions), the orientation of the wellbore with respect
to the in-situ stresses, the formation pressure and the
mechanical properties of the formation (elastic, plastic
and strength).
5.3 In-Situ Stress State
Knowledge of the in-situ stress state in the Cyrus
Field is limited. It was therefore necessary to per-
form a stability analysis using a range of assumed but
6 WELLBORE STABILITY ANALYSIS: A REVIEW OF CURRENT METHODS OF .....
reasonable values for the in-situ stresses. The following
assumptions were made:
1. Principal stresses are vertical and horizontally
oriented. This is a reasonable assumption in relatively
simple sedimentary basins away from salt domes, faults
and folds.
2. The overburden or vertical stress gradient can be
approximated to the weight of overlying rock. A value
of 1 psilft (0.023 MPalm) is reasonable in sedimentary
basins below depths of around 2,000m.
3. No Instantaneous Shut in Pressures (!SIP's) from
hydraulic fracturing were available from nearby wells.
However, !SIP's carried out in other reservoirs
throughout the UKCS, show that the minimum hori.;.
zontal stress gradient in normally pressured reservoirs
invariably resides between 0.6-0.9 psilft (0.014-0.021
MPalm).
4. Little information is available on the magnitude of
the maximum horizontal stress. However, ifit is greater
than the vertical stress, then a horizontal well will be
more stable than a vertical well (assuming isotropic rock
properties). Thus, the maximum horizontal stress is
taken to be no greater than 1 psilft (0.023 MPalm), since
values in excess of this do not constitute a problem.
5. The direction of the maximum and minimum hori-
zontalprincipal stresses are determined routinely using
breakout analysis (e.g. IOein and Barr [35]). In the
Cyrus Field no consistent elongation trends were found
in the exploration and appraisal wells, indicating that
the two horizontal stresses are similar in magnitude.
6. From Drill String Tests (DST's) the formation had
been established to be normally pressured; equivalent
to 0.45 psilft (0.10 MPalm).
From the above considerations, the wellbore stability
analysis was carried out assuming a vertical in-situ
stress gradient of 1.0 psilft (0.023 MPalm), and a range
ofisotropic horizontal stress gradients equal to 0.6, 0.7
and 1.0 psilft (0.014, 0.016 and 0.023 MPa/m); a
horizontal stress gradient of 0. 7 psilft (0.016 MPalm)
being considered the most probable.
5.4 Rock Material Properties
A series ofinstrumented triaxial and uniaxial tests were
carried out on plugs taken from reservoir core from an
offset well to determine the mechanical properties of the
rock anticipated in the horizontal section of the Cyrus
production wells. The rock testing was carried out in
accordance with ISRM recommendations [36]. Prior to
266
IADC/SPE 19941
testing the samples were saturated under vacuum with
water. However, no pore fluid pressures were applied
during testing.
The strengths of the plugs tested as a function of
confining pressures are plotted in Figure 11. To these
results a Mohr-Coulomb criterion is fitted, giving a
failure envelope defined by
CJ/ = 4000 + 5.5 x CJ3' (psi)
CJ/ = 28 + 5.5 X CJ3' (MPa)
The values of the tangent modulus at 50% strength, E 50 ,
are plotted against confining pressure in Figure 12. A
linear relationship between the elastic modulus and
minimum principal effective stress was assumed for the
FEM analysis, given by
E = 1.3 X 106 + 490 X CJ3' (psi)
E= 9.0 X 103 + 490 X CJ3' (MPa)
There was no apparent relationship between Poisson's
ratio and confining pressure; a constant value of0.2 was
assumed for the analysis.
5.5 Results
Figure 13 shows the computed maximum effective hoop
stress, CJ8', at the wellbore wall as a function of mud
density using both linear-elasticity and the FEM
analysis. Also shown is the maximum allowable hoop
stress as determined by the Mohr-Coulomb failure
criterion.
Using linear-elasticity, a minimum mud density of 1.24
S.G. is predicted for prevention of compressive failure
assuming the most favourable horizontal stress gradient
(1.0 psilft). This compares with a minimum mud density
of 1.07 S.G. predicted by the FEM assuming the same
horizontal in-situ stress state. If we assume the least
favourable horizontal stress gradient (0.6 psilft), the
minimum mud density requirement predicted by the
FEM is 1.18 S.G. (Note: The mud weights predicted
neglect the effects of swab and surge pressures).
On the basis of the FEM, it was considered that a 1.18
S.G. mud would be sufficient to support the wellbore,
provided swab and surge pressures were minimised and
good fluid loss characteristics were maintained to pro-
vide an effective mud cake. Although, swab pressures
will reduce the effective mud weight below 1.18 S.G. this
was not considered a problem as the recommendations
were based on the least favourable stress state and
plasticity effects had not been allowed for in the analysis,
which would mean the results are of a conservative
nature.
 IADC/SPE 19941 M.R. McLEAN, M.A. ADDIS 7

5.6 Actual Well Response in order to improve the predictions of wellbore stability
models.
The reservoir section of the first Cyrus horizontal well
was actually drilled with a 1.17 S.G. mud. It is not
possible to state whether the hole suffered any com-
pressive failure, since no calipers were run in the 8 1/2" NOMENCLATURE
hole. However, during drilling of this section the hole
appeared to be in perfect condition. No excessive (Compression assumed positive throughout)
overpulls while tripping out nor drag while tripping in
were encountered and drilling proceeded without inci- E Elastic modulus (secant modulus assumed)
dent. In addition, good agreement between the predicted
E0 Elastic modulus for uniaxial loading
and actual torque while drilling (Child and Cocking [37])
suggests an in-gauge hole. This response suggests that Ea Secant modulus
the linear-elastic analysis carried out was extremely
conservative. However, it is not possible to determine q Maximum deviator stress
the accuracy of the FEM analysis, since even lower mud
p Mean pressure
weights may have proved successful.
p' Mean effective pressure (mean pressure
minus pore pressure)
6.0 CONCLUSIONS
Svert Vertical/Overburden stress
1. Careful consideration should be given to the type of ~max Maximum horizontal in-situ stress
strength criterion used. Those criteria which incorpor-
ate the intermediate principal stress may lead to ~min Minimum horizontal in~situ stress
unacceptable overpredictions of formation strength
resulting in optimistic predictions of hole stability. ar,O'a,O'z Radial, circumferential and axial stresses
Criteria which do not consider the influence of the a11a2,a3 Maximum, intermediate and minimum
intermediate principal stress are likely to be conserva- principal stresses
tive in nature, particularly where used in association
with linear-elasticity. a' Effective stress (total stress minus pore
pressure)
2. Fitting of peak-strength criteria should be directed 'toct Octahedral shear stress
towards the appropriate confining pressure range,
which in most downhole instances is unlikely to include O'oct Octahedral normal stress
confining pressures greater than 2000 psi (14 MPa).

3. Incorporation of a stress dependent elastic modulus

determined from triaxial and uniaxial testing appears
to improve the predictions of elastic and elasto-plastic
models. However, the main basis for this assumption
are hollow cylinder test results, which may be subject to
scale effects relative to full scale wellbores. The
importance (or otherwise) of the scale effect should be
quantified before any degree of confidence can be
assigned to models validated against hollow cylinder test
results.
4. Application of a FEM incorporating a stress
dependent modulus has been applied to a horizontal
well drilled in the North Sea. The actual well response
implies that the FEM analysis gives much improved
predictions of hole stability relative to a linear-elastic
approach.
ACKNOWLEDGEMENTS
The authors wish to thank the British Petroleum
Company for permission and encouragement to publish
this paper. Further we wish to thank Bob Klein and
Marc Greenway for assistance with the analysis and
testing.
REFERENCES
1. Detournay, E. and Cheng, A.H. (1988). Poroelastic
response of a borehole in a non-hydrostatic stress
field. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr.. ,
No.25, pp.171-182.

5. The problems and costs of wellbore instability con-
tinue to be a major cost factor in drilling wells. To reduce
these costs continued R&D effort needs to be directed
towards a better understanding of rock behaviour
around circular openings in both the laboratory and field
2. Antheunis, D., Vriezen, P.B., Schipper, B.A. and van
der Vlis, A. C. (1976). Perforation collapse: Failure
of perforated friable sandstones. Proc. SPE-Euro-
pean Meeting, Amsterdam, April. SPE 5750.

267
8 WELLBORE STABILITY ANALYSIS: A REVIEW OF CURRENT METHODS OF .....
3. Morita, N. and Gray, K.E. (1980). A constitutive
equation for non linear stress-strain curves in rocks
and its application to stress analysis around a
borehole during drilling. 55th Annual Fall Tech.
Con{. and Exhib. of SPE, Dallas, Sept. SPE 9328.
4. Drucker, D.C. and Prager, W.J. (1952). Soil mech-
anics and plastic analysis or limit design. Quat. of
Appl. Math., Vol.10, pp.157-165.
5. Pariseau, W.G. (1968). Plasticity theory for aniso-
tropic rocks and soils. Proc.lOth U.S. Symp. onRock
Mech., Austin, Tx, pp.267 -295.
6. Hoek, E. and Brown, E.T. (1980). Underground
Excavations in Rock. The Institute of Mining and
Metallurgy, London, pp.131-182.
7. Woodland, D.C. (1988). Borehole instability in the
Western Canadian overthrust belt. SPE Rocky
Mountain Regional Meeting, Casper, Wy, May,
pp.319-331. SPE 17508.
8. Fuh, G.F .. Whitfill,D.L. and Schuh, P.R. (1988). Use
of borehole stability for successful drilling of high-
angle hole. IADC I SPE Drilling Con{., Dallas, Tx,
Feb. pp.483-491. IADC/SPE 17235.
9. Mitchell, R.F., Goodman, M.A. and Wood, E.T.
(1987). Borehole Stresses: Plasticity and the drilled
hole effect. SPE I IADC Drilling Conf., New Orleans,
La, Mar., pp.43-49. SPE/IADC 16053.
10. Hsiao, C. (1988). Growth of plastic zone in porous
medium around a wellbore. 20th Annual OTC Conf.,
Houston, Tx, May, pp.439-448. OTC 5858.
11. Aadnc;Jy, B.S. and Chenevert, M.E. (1987). Stability
of highly inclined boreholes. SPE I IADC Drilling
Con{., New Orleans, La, Mar., pp.25-41, SPE/IADC
16052.
12. Gnirk, P.F. (1972). The mechanical behaviour of
uncased wellbores situated in elastic/plastic media
under hydrostatic stress. SPE J., Feb., pp.49-59.
SPE 3224.
13. Bradley, W.B. (1979). Mathematical stress cloud -
stress cloud can predict borehole failure. Oil & Gas
J., Vol.77, No.8, Feb., pp.92-102.
14. Hottman, C.E., Smith,J.H. and Purcell, W.R. (1978).
Relationship among earth stresses, pore pressure,
and drilling problems offshore Gulf of Alaska. 53rd
Annual Tech. Con{. & Exhib. of SPE, Houston, Oct.
SPE 7051.
15. Nakken, S.J., Christensen, T.L., Marsden, J.R. and
Holt, R.M. (1989). Mechanical behaviour of clays at
268
IADC/SPE 19941
high stress levels for wellbore stability applications.
Proc. Int. Symp. ISRM-SPE, Pau, France, Aug.,
pp.141-148.
16. Marsden,J.R., Wu, B., Hudson,J.A. andArcher,J.S.
(1989). Investigation of peak rock strength behav-
iour for wellbore stability application. Proc. Int.
Symp. ISRM-SPE, Pau, France, Aug., pp.753-760.
17. Santarelli, F.J. (1987). Theoretical and Experi-
mental Investigation of the Stability of the Axisym-
metric Wellbore. PhD Thesis, University ofLondon.
18. Kwakwa, K.A., Batchelor, A.S. and Clark, R. (1989).
An assessment of the mechanical behaviour of
high-angle wells in Block 22/11. Proc. Offshore
Europe 89, Aberdeen, Sept. SPE 19240.
19. Mogi, K. (1967). Effect of the intermediate principal
stress on rock failure. J. Geophys. Res., Vol.72.
pp.5117-5131.
20. Takahashi, M. and Koide, H. (1989). Effect of the
intermediate principal stress on strength deforma-
tion behaviour of sedimentary rocks at the depth
shallower than 2000m. Proc. Int. Symp. ISRM-SPE,
Pau, France, Aug., pp.19-26.
21. Green, G.E. and Bishop, A.W. (1969). A note on the
drained strength of sand under generalised strain
conditions. Geotechnique, Vol.19, pp.144-149.
22. Paslay, P.R. and Cheatham, J.B. (1963). Rock
stresses induced by flow of fluids into boreholes. Soc.
Petrol. Engrs. J., Vol.3, No.1, March, pp.85-91.
23. AadD~Jy, B.S. (1987). Continuum mechanics analy-
sis of stability of inclined boreholes in anisotropic
rock formations. Ph.D Thesis, Norwegian Institute
of Technology, University of Trondheim.
24. Westergaard, H.M. (1940). Plastic state of stress
around a deep well. J. Boston Soc. of Civ. Engrs.,
Vol.27, No.1, Jan., pp.1-5.
25. Veeken, C.A.M., Walters, J.V., Kenter, C.J. and
Davies, D.R. (1989) Use of plasticity models for
predicting borehole stability. Proc. Int. Symp.
ISRM-SPE, Pau, France, Aug., pp.835-844.
26. Sulem, J. and Vardoulakis, I. (1988). A new
approach to borehole stability based on bifurcation
theory. Proc. 6th Int. Conf. Num. Meth. Geomech.,
Innsbruck, pp.1929-1935.
27. Guenot, A. (1987). Contraintes et ruptures autor
des forages petroliers. Proc. 6th ISRM Congr., Rot-
terdam, pp.109-118.
IADC/SPE 19941 M.R. McLEAN, M.A. ADDIS 9

28. Klein, R.J. and McLean, M.R. (1988). ·Application 33. Duncan Fama, M.E. and Brown, E.T. (1989).
of wellbore stability to a horizontal UK land well. Influence of stress dependent elastic moduli on
Con{. on Applied Rock Engng., Newcastle, Jan., plane strain solutions for boreholes. Proc. Int. Symp.
pp.117-126. ISRM-SPE, Pau, France, Aug., pp.819-826.

29. Bratli, R.K. and Risnes, R. (1981). Stability and
failure of sand arches. Soc. Petrol. Engrs. J., Vo.21,
No.3, April, pp.236-248.
30. Vardoulakis, I., Sulem, J. and Guenot, A. (1988).
Borehole instability as bifurcation phenomena. Int.
J. Rock Mech. Min. Sci. & Geomech. Abrst., No.25,
pp.159-170.
31. Sulem, J. and Vardoulakis, I. (1989). Simplified
bifurcation analysis of deep boreholes in rocks with
microstructure. Proc. Int. Symp. ISRM-SPE, Pau,
France, Aug., pp.845-852.
32. Haimson, B.C. and Herrick, C.G. (1989). Borehole
breakouts and in-situ stress. Proc. 12th Ann.
ETCE-ASME Drilling Symp., Houston, pp.17 -22.
34. McLean, M.R. (1988). Analysis ofwellbore stability.
Ph.D Thesis, University of London.
35. Klein, R.J. and Barr, M.V. (1986). Regional state of
stress in Western Europe. Proc. Int. Con{. on Rock
Stress and Rock Stress Measurements, Stockholm,
Sept., pp.33-44.
36. Brown, E.T. (1981). Rock characterisation testing
and monitoring. ISRM Suggested Methods. Perga-
mon, Oxford.
37. Child, A.J. and Cocking, D.A. (1989). Drilling
simulator improves drilling performance. Oil & Gas
J., Aug. 28, pp.41-47.

TABLE 1 Categorization of Peak-Strength Criterion.

Function of 0"1, cr2 & 0"3 Function of 0"1 & 0"3 only

Category A Category B
Linear Criterion
e.g. Drucker-Prager [4] e.g. Mohr-Coulomb
Category C Category D
Non-Linear Criterion
e.g. Pariseau [5] e.g. Hoek-Brown [6]

TABLE 2 Selection of Published Wellbore Stability Models.

Source Basic Model Behaviour Additional Features

Bradley [13] Linear Elasticity

Paslay & Cheatham [22] Linear Elastic Allowance for fluid flow
Aadnfl)y [23] Linear Elastic Allowance for elastic aniso-
tropy
Westergaard [24] Elasto-plastic
Veeken et al [25] Elasto-plastic Incmporates hardening and
softening behaviour
Sulem & Vardoulakis [26] Rigid-plastic, strain-harden-
ing
Santarelli [17] Stress dependent elasticity Includes pre-peak yielding

269
 SPE 19 94 1'

TABLE 3 Predicted and Actual Failure Pressures of Hollow Cylinders (after

Santarelli [17]).

ROCK TYPE FAILURE PRESSURE

Linear-Elastic Santarelli [17] Test Results
Carboniferous 45MPa 108MPa 127MPa
Sandstone
Doddington Sand- 24MPa 34MPa 57MPa
stone
Gebdykes 26MPa 38MPa 52MPa
Dolomite

LOST
CIRCULATION

(a)

HOLE
CLOSURE

Figure 1. Types of Stress Induced Figure 2. In-Situ Stress Field.

Wellbore Instability.

270
. ' SPE 1994 1

Biaxial Stresal
Unear Elastic Hollow Path /
Cylinder Stress Path /
/
120
-Blfx1ir~~~n~h----~-----------~~~~ 50 /~
6 /
/
Uniaxial Stresi / / " ,./
'i' Path / //I / l "
a.
/
. /
I
I
/
/ ~40
!. I I / c:r
,n
/ I .i
u;!!= 80 /
/
1 ,/'
I /Biaxial
Stress Path us~ 30
1
m / I / ()
•t:
.c / I ,. 0
en / I /
~ / I / i
i.c / I / ~ 20
/ I/
CCI / I,/ Peak Strength Data
u 40 / 1.-/ x Drained Triaxial Tests
0 ' I,/
1/ 10
Uniaxial.
6 Undrained Triaxial Tests
Strength -- I/
/ 1/
/ 1,/ Data Points From Santarelli (1987) Uniaxial
/ /#'
,
Strength-L---~--~--..----..-----..---
10 20 30 40 50
40 80 120 160 Mean Effective Stress, p' (MPa)
Octahedral Normal Stress (MPa)

Figure 3. Peak-Strength Data for a Figure 4. Peak-Strength Data for Weald

Gebdykes Dolomite (after Santarelli Clay (after Nakken et al [15]).
[17]).

(.Biaxial
1 Stress Path
40 /
0/ 0 #
Ci' Predicted el
D. 8TaX:TaTsfreng1t1 __ _
::::E
;: 30 Linear-elastic stresses
0 Unlaxla~ at internal wall
Fitted Strength
~
Stress; Or :0
Path i Criteria
c;; i Oe =2.13 P
() 20
i:
i Oz =1.07 P
i p __.
~ I
·~ o =Test Data
c 10

Uniaxial
Strength [._--..,.-----,...-----,,------..------,
10 20 30 40 50
Mean Effective Stress, p' (MPa)

Figure 5. Peak-Strength Data for Weald Figure 6. Typical Loading Adopted in

Clay (after Marsden et al [16]). Hollow Cylinder Testing.

271
 SPE 19 94' f

0.5 1.0 1.5 300

Shear Strain (%)

0.5 • .__ _ _ 50

lc
·e
Ci)
----20
.g
-
Cl)
E
::s
0 --r---,-----r--~:....,__~-
' - - - - 1 0 (Confining Pressure
in MPa)
;g oL--~~--~--~--~
0 1.0 2.0 3.0 4.0
-0.5
Axial Strain (%)

Figure 7. Model Calibration using Figure 8. Triaxial Test Data for a Car-
Uniaxial Test Results (after Sulem & boniferous Sandstone (after Santarelli
Vardoulakis [26]). [17]).

12

• Test Results

2o~~-~-~~-~
0 2.5 5.0 7.5 10.0 12.5
Diameter (em)

(a) (b)

Figure 9. Effect of Borehole Diameter on Breakout Initiation

Stress (after Haimson & Herrick [32]).

272
 SPE. 1994 1

0.0 ROTARY TABLE

0
25m SEA LEVEL
140m

BUILD AT 1.5,30m

\ 18 5/8", 1f, SOOm TVD,

502m MD

13 3/8", 2t, 1300m TVD,

~\,. 1357m MD

CYRU5-1

§§§::= BALDER Fm 9 5/8", 78.t, 2557m TV0,29&6m MD

SELFm
~
ANDREW BUILD AT 3"30m
FORMATION / TD, 2575m TVD, 3726m MD
9cf,2575m TVD,3126m MD
3000 L-..-----.-~-.....,.....-----,----..-----,

0 400 800 1200 1600 2000

Figure 10. Cyrus Well Trajectory and Geological Profile.

140
21
120
Ci 'ii'
~ 19
!_100
"'o

= E 11
! 80 Cit
u; .a:I 15
ca so "0

~ 40
::2
Cit
13
~g' 11
20 :I
:. 9

5 10 15 20 7
Confining Pressure (MPa) 0 5 10 15 20 25 30 35
Confining Pressure (MPa)

Figure 11.Peak-StrengthData for Cyrus Figure 12. Variation ofYoung's Modulus

Reservoir Sandstone. with Confining Pressure.

273
 SPE 19 94 1

-
(U
D.
~
100
Assumed horizontal
a; 90 stress gradient
3: ~
e0 80
:e
~ 70
I~ Unur Elastic

--
Cl)
.c Analysis
60
(U
U)

-
>
U)
e 50
en
a. 40
0
0
J: FEM Anolyslo
Cl) 30
>
=u
:!w 20
E 10
::I
E
·;c
(U
::::E
0.9 1.0 1.1 1.181'.2 1.3
Mud Density (S.G.)

Figure 13. Variation of the Maximum Effective Hoop Stress, a 9', with Mud
Density.

274