Professional Documents
Culture Documents
Understanding Loss Mechanisms
Understanding Loss Mechanisms
Summary
Mud losses are frequently observed when drilling in depleted formations. This is because of the decrease in the minimum in-situ stress
during depletion. As a result of this decrease, the lost-circulation pressure—or fracture gradient (FG)—decreases, and the operational
mud-weight window shrinks. Losses in such formations are often observed when drilling through sand/shale sequences. Preventing and
curing losses requires a sound understanding of loss mechanisms. In this study, we investigate several mechanisms that might be
responsible for the elevated risk of mud losses in differentially depleted sand/shale sequences. Numerical models of synthetic cases rep-
resentative of lost-circulation scenarios in a high-pressure/high-temperature (HP/HT) field in the North Sea, under normal faulting con-
ditions, are set up using the finite-element method. The simulations reveal that shear displacement at the horizontal sand/shale
interfaces is unlikely to cause losses. On the other hand, shear displacement and losses might be induced at high-angle sand/shale inter-
faces, such as those found near faults. In addition, faults introduce an extra complexity to the stress distribution and stress-path coeffi-
cients. Stress anisotropy near faults might increase during depletion, making both lost-circulation issues and borehole-stability
problems worse. The zone most prone to lost circulation in a faulted formation is located in depleted sand adjacent to the fault. The loss
mechanism here is because of drilling-induced fractures (DIFs). In addition, depletion itself might induce fractures in shale adjacent to
the depleted sand and located across the fault from it. Such fractures might then serve as escape paths for the drilling fluid during infill
drilling. The loss mechanism here is caused by pre-existing, depletion-induced fractures. These findings are in agreement with field
observations. A noncircular (elliptic or irregular) borehole cross section is found to reduce the fracture-initiation pressure (FIP). An
irregular borehole cross section is, thus, another possible mechanism behind irregular loss patterns observed in depleted fields. The
results from the study are important for establishing best practices when drilling in depleted formations.
Introduction
An important measure for increasing oil and gas recovery is infill drilling. For mature fields, drilling and completion of new wells can
be challenging because of a limited (or nonexistent) operational mud-weight window. The operational window is defined by an upper
and a lower bound. The lower bound is usually based either on the pore pressure or on the borehole-stability limit, whichever is higher.
The upper bound is an estimate of the maximum pressure the wellbore wall can be exposed to without experiencing mud loss (called
lost-circulation pressure or FG).
Lost circulation is experienced in all types of reservoirs. However, depleted reservoirs are more prone to this drilling problem
because of reduced pore pressure and reduced in-situ stress. With decreasing stress, the fracture-propagation pressure (FPP) decreases,
and it becomes easier to open and drive a DIF. Losses can be mitigated by reducing the mud weight, but this is not always possible
without inducing other problems. In particular, in the case of differential depletion (varying the pore-pressure gradient and FG along
the well), reducing the mud weight below the FG in one layer (e.g., depleted sand) might bring it below the lower mud-weight limit in
another, less depleted, layer (e.g., undepleted shale).
Estimating the lower and upper bound of the operational window can be challenging because of unknown input data (rock proper-
ties, in-situ stresses, the amount of depletion) and inherent rock heterogeneity. As a result, operators are forced to live with uncertainty.
This uncertainty can be further aggravated by complex lithology—for instance, faults, folds, and lithological boundaries. There is no
clear industrywide consensus or agreed-upon method for predicting the lost-circulation pressure or, equivalently, the FG. For cases with
a wide drilling window, this does not represent a problem, and FG based on the minimum stress (or the FPP) might provide a suffi-
ciently wide operational window. However, when drilling in depleted reservoirs where differential depletion in different layers has
occurred, constraining FG more precisely becomes a crucial issue. Many operators use the Kirsch equation for the hoop stress to esti-
mate the FG in such cases. However, the Kirsch solution often underestimates the FG observed in the field and in the laboratory tests.
This is because of many factors—for example, fracturing is a process, and FG is not an intrinsic bulk property of the rock that can be
measured reliably (Andrews et al. 2016). Unlike the minimum stress, which is an intrinsic bulk property of the formation that can, theo-
retically, be measured consistently from case to case, formation breakdown pressure (FBP) is a system property dependent upon many
factors, including particle-size distribution in the drilling fluid and fluid losses into the porous matrix. Furthermore, there are mecha-
nisms other than induced fracturing that can lead to lost circulation; for example, losses in highly permeable formations might occur
with poor muds when the bottomhole pressure exceeds the formation pore pressure. In such cases, in-situ stresses play no role. Simi-
larly, losses into naturally fractured or heavily faulted formations are largely controlled by the natural-fracture/fault system, including
fracture orientation, fracture density, frature roughness, and distribution of fracture aperture. In depleted formations, uncertainties in the
stress path and pore-pressure distribution add to this complexity.
Copyright V
C 2019 Society of Petroleum Engineers
This paper (SPE 195605) was accepted for presentation at the SPE Norway One Day Seminar, Bergen, Norway, 14 May 2019, and revised for publication. Original manuscript received for
review 20 March 2019. Revised manuscript received for review 27 August 2019. Paper peer approved 31 August 2019.
in the near-wellbore region. This is further enhanced by the lack of filter-cake deposition on low-permeability rocks such as shales.
Both might contribute to losses when the fracture aperture has become sufficiently large.
Equinor’s experience is that loss initiation events do not typically occur when drilling through homogeneous sands—even when
these are heavily depleted compared with less-permeable bounding lithologies. For reservoir drilling, the expectation is that if losses do
occur, they will likely start in weak zones, such as shale interbeds and shale immediately adjacent to depleted sands (in addition to natu-
rally weak/fractured zones). This experience is also reported elsewhere (Alberty and McLean 2001) and supported by observations that
DIFs observed on image logs are more prominent in shales/silts than in sands. This suggests that the wellbore’s resistance to lost circu-
lation when drilling sands with good muds can be much higher than the far-field stresses in the depleted layers. This is consistent with
the understanding that fracturing and lost circulation are processes and not intrinsic properties of the rock. Commonly used theoretical
models assume that FG is a function only of in-situ stress (e.g., Kirsch). However, numerous field and laboratory observations show
that this is also highly related to fluid properties and to interactions between drilling fluid and formation rock. DEA 13 (1985–88) and
GPRI experiments (2000) showed that the FBP/FPP is significantly affected by the base fluid and solids content of the drilling fluid.
The testing reported by Morita et al. (1990) shows that the back-calculated tensile strength fitted to FG on the basis of the application of
linear poroelasticity (Kirsch) was much higher than the directly (laboratory) measured tensile strength. Nonlinear elasticity could not
account for the difference. The results were explained by introducing a stable (short) fracture growth with high fracture-extension pres-
sure. Particles in this short fracture effectively shield the fracture from the fluid pressure initially. We found that sealing of such narrow
fractures (natural or induced) depends on Young’s modulus, wellbore size, and drilling fluid. This effect is not included in the Kirsch
equations. There are many studies confirming the influence of particles on FG and attempting to quantify this (Morita et al. 1996; van
Oort and Vargo 2008; Guo et al. 2011; van Oort et al. 2011; Jin et al. 2013; van Oort and Razavi 2014; Razavi et al. 2015; Andrews
et al. 2016; Feng et al. 2016; Chellappah et al. 2018; Chimara et al. 2018).
“Weak point” is often used as a generic term for fractures, faults, altered zones, and bed boundaries. A relevant question to ask is
whether or not the presence of such a weak point along a wellpath necessarily means certain losses. Equinor’s experience suggests not
(drilling across apparently similar geology transitions in several wells has shown different experiences), and the hope is that better
understanding of the loss mechanisms might help answer this. Our current understanding is that weak zones might be sites that have
“fracture starter geometry”—that is, they represent open “points of attack” where the well pressure can get “purchase” on the rock, and
fracturing can start (unlike a circular hole where there is no initial point of purchase). This means that the FG/loss-initiation pressure is
more likely closer to the minimum stress (with less contribution from potential hoop stresses around a perfect circular hole). They
might be natural weak points (such as naturally fractured formations or coals with open cleats) or might have been created by the
mechanical work from forces at bit/bottomhole assembly when making the hole. This might simply be the result of the drilling mechan-
ics of going from one lithology to another (side forces when hitting hard stringers leading to a noncircular hole), or it might result from
general excessive mechanical and hydraulic forces in mechanically weak formations that give breakouts/washouts, for example. Sup-
porting experience for this postulate comes from drilling experience where we find that lost circulation tends to occur at the bit; that is,
weak zones are not “weak” once they have been drilled successfully and are behind the bottomhole assembly. Another potential expla-
nation for lost-circulation tendencies at lithology boundaries can be related to the redistribution of stresses that can occur there. This
paper attempts to quantify these various mechanisms/effects.
Understanding loss tendencies at weak zones is critical for drilling in depleted reservoirs. Drilling windows in depleted sands can be
extended far beyond the traditional FG (on the basis of FPPs in the depleted zone) by using the extra “strength” provided by the well-
bore and specialized particles. Such a strategy relies on the wellbore “strength” from hoop stresses and from particles isolating far-field
stresses in “deeper” formations from higher wellbore pressures. Integrity against losses then relies on a thin shell around the wellbore
(cm to dm in scale), where mechanisms such as increased hoop stress and good particle blocking of small incipient-induced fractures/
natural microfractures, give increased resistance to wellbore pressures. However, as mentioned previously, if the integrity of the well-
bore is lost (e.g., because of the presence of weak zones), then pressures required to stop static losses can be dictated by the lower far-
field stresses in the connecting depleted sands. If this occurs while high-pressure permeable sands are exposed in the wellbore
(differential-pressure case), then there is the possibility for a kick and a serious well-control event with an underground blowout. Differ-
ent strategies exist to mitigate such well-control risk (Fambon and Joffroy 2008; Ay et al. 2016).
Loss Classification
One method of classifying losses is according to their severity: seepage losses, partial losses, severe losses, and total losses. However,
this method does not give much information about the loss mechanisms or the quantitative properties giving rise to lost circulation.
Classifying losses according to their mechanisms (high-permeability matrix, vugular formations, natural fractures, DIFs, or a combina-
tion) might give better presumptions to cure the losses (Lavrov 2016a):
• Losses to the porous matrix can be common in sand and gravel and occur when high-permeability rock is being exposed to the
drilling fluid. Such losses can be relatively easy to control with suitable mud specifications, and the losses will stop when as the
filter cake builds up on the borehole wall as drilling proceeds.
• Losses in vugular formations are characterized by total losses and occur in limestone, dolomite, and rock salt. The vugs, or caverns,
might vary in size and might be connected to each other.
• Losses caused by natural fractures are among the most-challenging types because losses will depend on many parameters: the
wellbore pressure, the hydraulic aperture of the fracture, the drilling-fluid rheology and composition, the leakoff through the frac-
ture walls, and buildup of the filter cake inside the fracture.
• Losses caused by induced fractures occur when the bottomhole pressure exceeds the FBP. Understanding fracture initiation and
propagation is essential to estimate pressures for this type of loss. The FBP is a system response and potentially depends upon
many parameters: in-situ stress, well orientation, rock tensile strength, borehole diameter, stiffness and permeability of the rock,
and drilling-fluid particle content and rheology, and also on the condition and geometry of the wellbore (e.g., pre-existing cracks,
irregular hole geometry).
Thus, there is inherent uncertainty in the estimation of FG. Lost-circulation pressures can range from pore pressure, FPP, and
fracture-reopening pressure to FBP for different wellbore geometries and lithologies. As such, FG is governed not only by in-situ
stresses but also by the wellbore condition, drilling-fluid rheology, elastic properties of the rock, and the presence, orientation, and aper-
ture of natural fractures.
In this work, we consider the influence on the FG of varying rock properties, different borehole geometries, and stress redistribution
that can occur at lithology boundaries. We do not consider the effects of particles on FG. The study includes several synthetic cases
analyzed with the finite-element method. It should be emphasized that, although these cases are representative of the conditions that
have been observed at sand/shale lost-circulation zones in the North Sea, none of these cases aims to represent a specific lost-
circulation event.
Drilling Through Horizontal Shale/Sand Interfaces: Can Losses Be Caused by Shear Stress and
Shear Displacement at the Interface?
Drilling a wellbore brings about stress changes in the near-wellbore area. It seems plausible that these changes could induce failure or
displacement at sand/shale interfaces, eventually leading to mud losses. To verify this conjecture, a finite-element simulation was per-
formed with ABAQUS (Dassault Systèmes 2013) where a borehole was drilled into a sequence of three horizontal layers: shale, sand,
and shale. There were two sand/shale interfaces in the model: upper and lower. The scenario simulated drilling a vertical well into a
depleted reservoir. The rock properties and the reservoir drawdown correspond to an Norwegian Continental Shelf (NCS) asset. Horizon-
tal layering was assumed, which allowed the use of a 2D axisymmetric model. The model geometry is shown in Fig. 1. Only half of the
model is shown, and the borehole is located on the left. The formation extended 12 m from the borehole axis in the horizontal direction
to eliminate possible artifacts caused by the finite size of the model. The height of the model was 12 m, and each shale layer was 5 m
thick. The middle layer was sandstone, and was 2 m thick. Drilling was simulated by advancing the borehole in five simulation steps:
1. From the top of the model to the middle of the top shale layer (7.5 minutes)
2. From the middle of the top-shale layer to the bottom of the top shale (7.5 minutes)
3. From the bottom of the top shale to the middle of the sandstone layer (3 minutes)
4. From the middle of the sandstone layer to the bottom of the sandstone (3 minutes)
5. From the bottom of the sandstone to the middle of the bottom-shale layer (7.5 minutes)
The borehole radius was equal to 0.1 m. A Biot poroelastic model was assumed for the constitutive model for both lithologies. Thus,
possible failure was not modeled explicitly. Instead, we analyzed the stresses that develop as the drill bit passes the interface, and we
compared the shear stresses at the sand/shale interfaces with the shear resistance at those interfaces. The latter is simply the sand/shale-
friction coefficient times the normal effective stress at the interface. Simulations were performed without filter-cake buildup—that is, a
conservative scenario. A filter cake would prevent the pore pressure at the interface from building up, which, consequently, would stabi-
lize the interface.
Shale 5m
Sand 2m
Shale 5m
0.1 m
12 m
Borehole axis
S, S12
(average: 75%)
+4.438×106
+2.717×106
+9.952×105
–7.264×105
–2.448×106
–4.170×106
–5.891×106
–7.613×106
–9.334×106
–1.106×107
–1.278×107
–1.450×107
–1.622×107
Fig. 2—Distribution of shear stress (Pa). The “shelf” on the left indicates the bottom of the borehole as the drill bit crosses
the upper shale/sand interface in the model. The arrow indicates the location where the shear stress on the interface is equal to
1.5 MPa. The effective vertical stress at the same location is equal to 61 MPa, which prevents shear displacement on the interface.
Similar results were obtained at the lower interface. For example, as the drill bit was located at the upper sand/shale interface, the
maximum shear stress at the lower interface was 0.5 MPa. As the drill bit was located at the midheight of the sand layer, the maximum
shear stress at the lower interface was 1.9 MPa. As the drill bit was located at the lower sand/shale interface, the maximum shear stress
at the lower interface was 3.6 MPa. The effective normal stress at the lower interface varied between 60 and 65 MPa during the simula-
tion, thus preventing slip at the lower interface.
An additional simulation was performed, whereby the borehole was drilled in the same model but without prior depletion. It turned
out that the effective normal stress at the interfaces was, in this case, approximately four times lower than when drilling after depletion.
However, this lower effective normal stress was still sufficient to prevent slip. Indeed, the results suggest that depletion might, in fact,
have a stabilizing effect on horizontal interfaces in terms of possible shear displacement.
As a next step in our study of mud-loss mechanisms in “weak spots,” we investigated whether slip at horizontal sand/shale interfaces
could be caused by the extra load from the weight on bit (WOB). In the simulation previously presented, the normal stress acting on the
bottom was assumed to be equal to the normal stress acting on the side wall of the borehole (mud weight). In reality, there is some extra
load on the bottom, in addition to that created by the mud column; this extra load is acting at the contact spots between the rock and the
cutters (in polycrystalline-diamond-compact bits) or the teeth (in roller-cone bits).
To incorporate the WOB effect into our finite-element model, we replaced the action of individual cutters at multiple contact spots
with equivalent distributed normal stress on the hole bottom. The total normal stress applied to the hole bottom is now equal to
Pf A þ WOB WOB
Peq ¼ ¼ Pf þ ; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ð1Þ
A A
where Pf is the bottomhole fluid pressure and A is the area of the hole bottom (borehole radius of 0.1 m was used in all simulations).
Simulations were performed with WOB varying from zero to 30 t. It was found that within the studied range of WOB, the shear stress
acting at sand/shale interfaces either does not change appreciably or it actually decreases. In particular, the shear stress induced at the
upper sand/shale interface as the drill bit was crossing that interface was found to decrease with WOB. Thus, with the particular choice
of rock properties used in the simulation, the effect of WOB was in fact to somewhat stabilize the interfaces. Even if the effect had
been the opposite, its magnitude would most likely be the same as in Fig. 3, where the shear stress is changed only by 25% when the
highest WOB of 30 t is applied. Such change, though appreciable, would still be insufficient to activate shear displacement at the inter-
face because WOB does not affect the effective normal stress on the interface. However, we note that the simulations here are for a
very idealized case, and we have not modeled the effects of side loads from the bottomhole assembly.
2.0
Shear Stress (MPa)
1.5
1.0
0.5
0
0 10 20 30
Weight on Bit, t
Fig. 3—Shear stress at the upper sand/shale interface at the moment when the drill bit crosses that interface vs. WOB.
The negligible effect of WOB can be easily explained if we estimate the contribution the WOB makes to the total stress applied on
the hole bottom. Given the mud pressure of 80 MPa and the borehole radius of 0.1 m, the extra load created on the hole bottom by a
WOB of 30 t constitutes only 20% of the total load. At shallow depths, where the mud pressure is lower, this percentage would be
higher. But the WOB used to drill shallower rocks is also usually lower, and the borehole diameter is larger.
Thus, the extra stresses caused by WOB cannot be held responsible for shear displacement along horizontal sand/shale interfaces.
can be much more complex. Boreholes might cross nonhorizontal interfaces, and the total normal stress on interfaces might also be
altered during depletion by the arching effect (Fjær et al. 2008). To study these effects and how they affect lost circulation, a more-
realistic-looking numerical model was constructed. In this model, a shale/sand/shale sequence included a normal fault (Fig. 4). The
dimensions of the model are given in Fig. 4. The model was 2D, and plane-strain conditions were applied, which corresponds to the
fault extending infinitely in the direction normal to the page in Fig. 4. The fault was represented with a layer of “soft gouge” along its
middle part and with a “stiff gouge” in the uppermost and lowermost parts. The properties of the stiff gouge were equal to the properties
of shale, except for Poisson’s ratio, which was higher. The Biot effective-stress coefficient, a, was close to unity for all rocks in the
model. The rock properties are given in Table 1. Gravity was not included in the model. This was done intentionally, to make the inter-
pretation of the results more transparent. Gravity would only make a constant additive contribution to the stress field.
“Stiff” gouge
Shale, 20 nd Shale, 20 nd
Sand, 20 md 100 m
San
d, 2 1 km
0m
d
Shale, 20 nd
e
goug
“Soft” Shale, 20 nd
y (2)
e
“Stiff” goug
x (1) 2 km
Fig. 4—Geometry of numerical model containing a sand/shale sequence with a normal fault.
The following initial and boundary conditions were applied: a constant normal total stress of 92 MPa was applied at the top bound-
ary. Zero normal displacement was applied at the bottom boundary and at the left-hand boundary, as is common in geomechanical
models. A normal displacement of –0.3 m was applied at the right-hand boundary. A pore pressure of 75 MPa was initially applied
throughout the model, in both the sand and the shale lithologies.
These initial and boundary conditions resulted in the following initial total stresses in the sand layer located in the hanging wall
(Point A in Fig. 5, top panel): Vertical stress is 89.8 MPa; the horizontal stress in the plane of the figure is 84.3 MPa; and the horizontal
stress normal to the page (i.e., the stress acting parallel to the fault-strike direction) is 80 MPa. These values are roughly representative
of the stress conditions prevailing in an NCS field where mud losses at weak zones have been experienced during infill drilling.
In the sand layer located in the footwall (Fig. 5, bottom panel), the initial total stresses in bulk sand (Point A) were as follows: Verti-
cal stress is 93.3 MPa; horizontal stress in the plane of the figure is 83.2 MPa; and the horizontal stress normal to page (i.e., the stress
acting parallel to the fault-strike direction) is 80.6 MPa. These values, again, are roughly representative of the stress conditions prevail-
ing in an NCS field.
Depletion was applied either in the hanging wall or in the footwall by reducing the pore pressure, Pp, from 75 to 25 MPa over a
10-year period. The pore-pressure distributions at the end of the 10-year period are shown in Fig. 5. Differential depletion is evident in
Fig. 5. Stress changes were observed at the monitoring points (A, B, and C in Fig. 5); the values of FIP were calculated; and the stress-
path coefficients were calculated for the vertical stress rv, the maximum horizontal stress rH, and minimum horizontal stress rh. We
denote these stress-path coefficients as cv, cH, and ch, respectively. A stress-path coefficient c is defined here as the ratio of the stress
change to the pore-pressure change at the same location.
FIP is often used in drilling engineering to estimate the upper mud-weight limit. The lost-circulation pressure is then assumed equal
to the FIP. FIP in sand was calculated here assuming a vertical well and perfect filter cake. In this case, the FIP can be estimated as fol-
lows: FIP ¼ 3rh – rH – Pp as per the nonpenetrating Kirsch equations (Fjær et al. 2008; Feng et al. 2016). This equation, which is
widely used, assumes a perfect filter cake. This is rarely the case when the borehole has just been drilled. In reality, in a high-
permeability rock such as reservoir sandstone, two competing processes are in play: filter-cake buildup and pore-pressure diffusion. If
the former process is fast enough, the preceding equation provides a sensible estimate for FIP. If not, the FIP will likely be lower
because the pore pressure in the near-wellbore region will be higher than in the rest of the reservoir. Equations are available to calculate
FIP for a fluid-penetrating case (including poroelastic effects), but these have not been used here.
POR
(average: 75%)
+8.570×107
+8.065×107
+7.559×107
+7.053×107
+6.547×107
+6.041×107
+5.535×107 A
+5.029×107 25 MPa B
+4.523×107 C
+4.018×107 75 MPa
+3.512×107
+3.006×107
+2.500×107
POR
(average: 75%)
+8.635×107
+8.124×107
+7.613×107 C
A 25 MPa
+7.102×107 75 MPa B
+6.590×107
+6.079×107
+5.568×107
+5.056×107
+4.545×107
+4.034×107
+3.523×107
+3.011×107
+2.500×107
Fig. 5—Pore pressure distributions (Pa) after depletion of sand in the hanging wall (top) or in the footwall (bottom). A, B, and C
indicate the locations of three points used for stress monitoring.
In shale, the FIP was estimated using the same formula: FIP ¼ 3rh – rH – Pp. This formula in shale is valid only at the very instant
the hole is drilled and exposed to the mud pressure (assuming that drilling induces no volumetric changes in near-wellbore rock). For
subsequent times, pressure diffusion into the rock will gradually change the FIP.
The traditional approach to quantifying the lost-circulation pressure is to use the value of the minimum far-field in-situ stress rh as
the lost-circulation pressure. This approach implies that the borehole has intersected a long, natural fracture, and there is full hydraulic
contact between the borehole and the rest of the fracture. The value of rh was used in this study as an alternative measure of lost-
circulation pressure.
It should be noted that several other approaches are available in rock mechanics for estimating the FIP. For instance, Jin et al.
(2013) suggested estimating the lost-circulation pressure assuming a short initial (natural) crack emanating from the borehole wall. The
lost-circulation pressure obtained with this model rapidly decreases with the length of the initial crack.
• The stress-path coefficient cv was equal to 0.01, 0.21, and 1.08 at Points A, B, and C, respectively. Thus, there was no stress arch-
ing at Point A and some stress arching at Points B and C. It should be noted that stress arching was observed in sand at Point B,
although sand was stiffer than shale in our model (see the explanation for the preceding case). The high value (in excess of unity)
of cv at Point C is caused by the way we calculate the stress-path coefficient: Namely, the local variation of pore pressure is used
in the denominator. At Points A and B, this makes no difference because the local variation in pore pressure is equal to the pres-
sure depletion (50 MPa). At Point C, however, located in shale, the pore-pressure variation is smaller by an order of magnitude.
Therefore, the stress-path coefficients appear larger than unity, although the actual stress variation is relatively small.
• FIP decreased in all locations (A, B, and C) during depletion. The largest decrease was observed at Point B (depleted sand adja-
cent to the fault).
• The Horizontal-stress-path coefficient at Location A was anisotropic: cH < ch. As a result, horizontal stress anisotropy increased
during depletion. This increase of anisotropy might affect both the FIP and the borehole stability of a vertical well.
• At Locations B and C, stress reversal was observed during depletion. At Location B, for example, the minimum horizontal stress
before depletion was acting in the plane of the figure; after depletion, the minimum horizontal stress was acting out of plane—that
is, the principal stress directions became reversed. Thus, DIFs would grow along the fault strike when drilling in sand adjacent to
shale before depletion. At Location C, the opposite was observed: before depletion, the minimum in-situ stress was acting out of
plane (normal to the page in Fig. 7). After depletion, the minimum in-situ stress was acting in the plane of the figure. Therefore,
when drilling in shale adjacent to sand before depletion, DIFs would grow toward the fault. When drilling after depletion, DIFs
would grow parallel to the fault strike direction.
POR
(average: 75%)
+8.570×107
+8.065×107
+7.559×107
+7.053×107
+6.547×107
+6.041×107
+5.535×107 A
+5.029×107 25 MPa B
+4.523×107 C
+4.018×107 75 MPa
+3.512×107
+3.006×107
+2.500×107
Fig. 6—Summary of stress changes during depletion in the hanging wall. The local pressure decrease at Point C was used to cal-
culate the stress-path coefficients at that location, hence their high values.
POR
(average: 75%)
+8.635×107
+8.124×107
+7.613×107 C
A 25 MPa
+7.102×107 75 MPa B
+6.590×107
+6.079×107
+5.568×107
+5.056×107
+4.545×107
+4.034×107
+3.523×107
+3.011×107
+2.500×107
Fig. 7—Summary of stress changes during depletion in the footwall. The local pressure decrease at Point C was used to calculate
the stress-path coefficients at that location, hence their high values.
The results are summarized in Fig. 7. It is evident from Fig. 7 that, when drilling a vertical well in such a differentially depleted for-
mation, the largest risk for mud losses is at Location B: in the depleted sand near the fault. The mechanism for mud losses at this loca-
tion is drilling-induced tensile fractures propagating toward the fault (the minimum in-situ stress is the out-of-plane stress acting
parallel to the fault’s strike direction). The lower FIP at the interface fits with the experience of loss-initiation events at such
transition zones.
An interesting observation was made in this simulation regarding the evolution of the minimum effective stress during depletion.
Before depletion, this stress was compressive in the entire model except for two locations where properties of the fault gouge changed
(Fig. 8, top panel). After depletion, an area of tensile minimum effective stress developed in shale adjacent to the depleted sand and
located across the fault from the latter (Fig. 8, bottom panel). Thus, depletion might induce tensile fractures in shale located near an
inclined sand/shale interface. Such fractures might cause mud losses during subsequent drilling. The mechanism of lost circulation, in
this case, would then be depletion-induced rather than DIFs.
S, maximum in-plane
principal (average: 75%)
+3.826×107
+3.507×107
+3.188×107
+2.870×107
+2.551×107
+2.232×107
+1.913×107
+1.594×107
+1.275×107
+9.565×106
+6.377×106
+3.188×106
+0.000×100
–3.619×107
S, maximum in-plane
principal (average: 75%)
+3.500×107
+3.208×107
+2.917×107
+2.625×107
+2.333×107
+2.042×107
+1.750×107
+1.458×107
+1.167×107
+8.750×106
+5.834×106
+2.917×106
+0.000×100
–4.539×107
Fig. 8—Minimum effective stress (Pa) before depletion (top panel) and after depletion (bottom panel). Dark area means compres-
sive effective stress. Colored areas indicate tensile stress. Tensile stress is positive in this figure.
The reason for the tensile effective horizontal stress in the blue area in Fig. 8 (bottom panel) is the stress-equilibrium condition:
Total horizontal stress decreases in the depleted sand, and the total horizontal stress in the adjacent shale must change accordingly,
whereas the pore pressure in shale remains almost unchanged (Lavrov 2016b). It should be noted that in order for this tensile effective
stress to appear, the sand/shale interface must have a sufficiently high angle. No tensile effective stress was observed in the simulation
with depletion in the hanging wall presented previously because the interface (the fault) was almost horizontal at Location C in that
case (Fig. 6).
Fig. 10—Tensile effective hoop stress (Pa) around boreholes with different cross sections and well pressures. Tensile stress is
positive in this figure. Left-hand column: circular hole; middle column: elliptic hole; right-hand column: irregular hole. Top row:
well pressure 70 MPa; middle row: well pressure 78 MPa; bottom row: well pressure 86 MPa. Far-field stress is the same and equal
to 52 MPa in all plots. Pore pressure is 25 MPa in all plots. Dark area means compressive effective hoop stress. Color code indi-
cates zones where the effective hoop stress is tensile and might induce fractures. There is no tensile failure with the circular hole
and well pressures at 70 and 78 MPa, respectively. There is tensile failure in all other cases.
It is evident from Fig. 10 that although borehole pressure of 70 MPa does not create tensile hoop stress around a circular well, it
would do so if the wellbore were elliptic or irregular. Tensile conditions begin at wellbore pressures of 79, 61, and 61 MPa, respec-
tively, for the circular, elliptic, and irregular geometries simulated. Thus, the irregular geometry in this simulation is, in terms of tensile
failure, equivalent to an elliptic borehole with the axis ratio of 1:2. In borehole-stability models, the upper mud-weight limit is usually
calculated assuming a circular cross section. Such predictions would be too optimistic if the cross section is not circular. The effect of
cross-section shape on lost-circulation pressure might be one of the reasons for erratic mud-loss patterns observed in sand/shale sequen-
ces in depleted reservoirs: The lost-circulation pressure is reduced in elliptic and irregular wells. In addition, washouts/breakouts
increase the borehole circumference and thereby the probability of having an initial crack in the wall (this is similar to the “size effect”
well-known in rock mechanics and materials science). The irregular hole shape with correspondingly lower FIP could be a potential
mechanism in explaining why losses tend to begin in weak zones.
Sand (50 md) with 1.0-cm crack and good filter cake 70
Sand (50 md) with 1.0-cm crack and poor filter cake 60
This illustrates the influence of cracks (assumption with Kirsch of the perfectly intact and circular wellbore) and how the extra
“strength” present from the hoop stress can be significantly reduced in wellbores that contain pre-existing cracks (natural cracks). The
results also indicate how a good mud (which forms a good filter cake) can help reduce the negative effect of such cracks on break-
down pressures.
Conclusion
Numerical simulations of potential shear failure at horizontal sand/shale interfaces far from faults show that it is unlikely that such fail-
ure occurs while drilling through sand/shale or sand/coal sequences. Shear displacement along interfaces would require an extremely
low friction coefficient (0.1 or lower). Even if shear displacement could develop, losses could only occur if the wellbore pressure was
larger than the normal total stress on the interface, which is the vertical in-situ stress. In the normal tectonic regime, this is quite
unlikely because the vertical stress is the largest principal stress. Nonhorizontal shale/sand boundaries, on the other hand, might be
more prone to shear displacement and lost circulation because the normal stress on such interfaces is lower.
Not surprisingly, it is found that faults introduce an extra complexity into stress distribution and stress-path coefficients. Horizontal
stresses can be anisotropic near faults, and this anisotropy might increase during depletion both in sand layers and in shales adjacent to
faults. This is expected to have consequences for borehole stability. Stress-path coefficients for the minimum and intermediate in-situ
stresses are likely to be unequal near faults. During depletion, FIP in sand might decrease less dramatically than the total minimum
in-situ stress does. The simulations show that the zone most prone to loss initiation is the depleted sand near a fault (a sand/shale inter-
face). The lower FIP at the interface fits with the experience of loss-initiation events at such transition zones. In shale adjacent to
depleted sand across a fault, lost circulation during subsequent drilling can be caused by depletion-induced fractures if the minimum
effective stress during depletion became tensile and in excess of the tensile strength of the shale.
At a given borehole pressure, a borehole of noncircular shape (elliptic or irregular) is found to be more likely to experience tensile
fracturing and mud losses than an ideal circular borehole. In a borehole of elliptic or irregular cross section, tensile fractures are most
likely to appear in the most-concave parts of the cross section. Maintaining a circular borehole cross section will, therefore, reduce the
probability of lost circulation. Irregular borehole shape might contribute to nonconsistent lost-circulation patterns observed in some
fields. Formations subject to washouts and breakouts will be more prone to unpredictable, irregular-loss patterns. This is a plausible
mechanism in explaining why weak zones act as loss-initiation sites—for example, changes in lithology can give side forces during
drilling that can lead to an irregular hole shape. The presence of existing cracks in the wellbore, another deviation from the common
assumption of the ideal circular hole, is also shown to reduce the FG significantly. The results also indicate how muds that form a good
filter cake can help mitigate the negative effect of such cracks on lost-circulation pressures.
The various simulations presented herein predict a large range in potential FG for a case representative of the conditions prevailing
in an NCS field for infill drilling with 500-bar depletion. If FIP and/or FBP is taken to be dimensioning for FG, the simulations show a
potential range from 600 to 790 bar for the given assumptions on, for example, geometry and properties. This range can be extended
even further downward if minimum stress is assumed to dictate loss potential. The work also confirms the understanding that poor
drilling practice (that can create weak points) can directly reduce the FG (in addition to unfavorable equivalent-circulating density-
conditions in the well).
Acknowledgments
The authors are grateful to Equinor for granting permission to publish this paper. Thanks also to Bård Bostrøm for help with simulations
on the effect of microcracks on lost-circulation pressures.
References
Alberty, M. W. and McLean, M. R. 2001. Fracture Gradients in Depleted Reservoirs—Drilling Wells in Late Reservoir Life. Paper presented at the SPE/
IADC Drilling Conference, Amsterdam, The Netherlands, 27 February–1 March. SPE-67740-MS. https://doi.org/10.2118/67740-MS.
Andrews, J. S., Fintland, T. G., Helstrup, O. A. et al. 2016. Use of Unique Database of Good Quality Stress Data to Investigate Theories of Fracture Ini-
tiation, Fracture Propagation and the Stress State in the Subsurface. Paper presented at the 50th US Rock Mechanics/Geomechanics Symposium,
Houston, Texas, USA, 26–29 June. ARMA-2016-887.
Ay, A., Sagberg, T., Gyland, K. R. et al. 2016. Use of Specially Treated Micronized Barite Fluid System with Wellbore Strengthening Approach to Drill
Extremely Depleted Reservoir—HPHT Kvitebjørn Field, Offshore Norway. Paper presented at the SPE Bergen One Day Seminar, Grieghallen,
Bergen, Norway, 20 April. SPE-180055-MS. https://doi.org/10.2118/180055-MS.
Barton, N. 1976. The Shear Strength of Rock and Rock Joints. Int J Rock Mech Min Sci Geomech Abstr 13 (9): 255–279. https://doi.org/10.1016/0148-
9062(76)90003-6.
Chellappah, K., Majidi, R., Aston, M. et al. 2018. A Practical Model for Wellbore Strengthening. Paper presented at the IADC/SPE Drilling Conference
and Exhibition, Fort Worth, Texas, USA, 6–8 March. SPE-189589-MS. https://doi.org/10.2118/189589-MS.
Chimara, G., Calder, A., Amer, W. et al. 2018. Engineered Wellbore Strengthening Application Enables Successful Drilling of Challenging Wells. Paper
presented at the SPE Trinidad and Tobago Section Energy Resources Conference, Port of Spain, Trinidad and Tobago, 25–26 June. SPE-191219-MS.
https://doi.org/10.2118/191219-MS.
Dassault Systèmes. 2013. ABAQUS 6.13 Theory Guide. Vélizy-Villacoublay, France.
Fambon, L. and Joffroy, G. 2008. Successful Development Drilling of an HP/HT Infill Well in a Highly Depleted Reservoir: Case Study. Paper presented
at the IADC/SPE Drilling Conference, Orlando, Florida, USA, 4–6 March. SPE-112708-MS. https://doi.org/10.2118/112708-MS.
Feng, Y., Jones, J. F., and Gray, K. E. 2016. A Review on Fracture-Initiation and Propagation Pressures for Lost Circulation and Wellbore Strengthening.
SPE Drill & Compl 31 (2): 134–144. SPE-181747-PA. https://doi.org/10.2118/181747-PA.
Fjær, E., Holt, R. M., Horsrud, P. et al. 2008. Petroleum Related Rock Mechanics, second edition. Amsterdam, The Netherlands: Elsevier.
Guo, Q., Feng, Y. Z., and Jin, Z.-H. 2011. Fracture Aperture for Wellbore Strengthening Applications. Paper presented at the 45th US Rock Mechanics/
Geomechanics Symposium, San Francisco, California, USA, 26–29 June. ARMA-11-378.
Jin, X., Shah S. N., Roegiers, J.-C. et al. 2013. Breakdown Pressure Determination—A Fracture Mechanics Approach. Paper presented at the SPE
Annual Technical Conference and Exhibition, New Orleans, Louisiana, USA, 30 September–2 October. SPE-166434-MS. https://doi.org/10.2118/
166434-MS.
Jo, Y. and Chang, C. 2017. Effect of Host Rock Properties on Shear Behavior of Discontinuities in Sandstones. Int J Rock Mech Min Sci 100: 238–249.
https://doi.org/10.1016/j.ijrmms.2017.10.007.
Kostov, N., Gosavi, S. V., Kulkarni, K. et al. 2016. Fracture Modeling for Optimum Wellbore Integrity Enhancement. Paper presented at the Abu Dhabi
International Petroleum Exhibition & Conference, Abu Dhabi, UAE, 7–10 November. SPE-183325-MS. https://doi.org/10.2118/183325-MS.
Lavrov, A. 2016a. Lost Circulation—Mechanisms and Solutions. Oxford, UK: Elsevier.
Lavrov, A. 2016b. Dynamics of Stresses and Fractures in Reservoir and Cap Rock Under Production and Injection. Energy Procedia 86: 381–390.
https://doi.org/10.1016/j.egypro.2016.01.039.
Morita, N., Black, A. D., and Fuh, G.-F. 1990. Theory of Lost Circulation Pressure. Paper presented at the SPE Annual Technical Conference and Exhi-
bition, New Orleans, Louisiana, USA, 23–26 September. SPE-20409-MS. https://doi.org/10.2118/20409-MS.
Morita, N., Black, A. D., and Fuh, G.-F. 1996. Borehole Breakdown Pressure with Drilling Fluids—I. Empirical Results. J Rock Mech Min Sci 33 (1):
39–51. https://doi.org/10.1016/0148-9062(95)00028-3.
Rai, U. B., Kumar, R., Adwani, A. et al. 2016. Drilling Depleting Reservoirs: Case Study from a Bruneian Field on Diminishing Drilling Windows and
Stress Paths. Paper presented at the International Petroleum Technology Conference, Bangkok, Thailand, 14–16 November. IPTC-18949-MS.
https://doi.org/10.2523/IPTC-18949-MS.
Razavi, O., Vajargah, A. K., van Oort, E. et al. 2015. How to Effectively Strengthen Wellbores in Narrow Drilling Margin Wells: An Experimental
Investigation. Paper presented at the SPE Annual Technical Conference and Exhibition, Houston, Texas, USA, 28–30 September. SPE-174976-MS.
https://doi.org/10.2118/174976-MS.
van Oort, E., Friedheim, J. E., Pierce, T. et al. 2011. Avoiding Losses in Depleted and Weak Zones by Constantly Strengthening Wellbores. SPE Drill &
Compl 26 (4): 519–530. SPE-125093-PA. https://doi.org/10.2118/125093-PA.
van Oort, E. and Razavi, O. S. 2014. Wellbore Strengthening and Casing Smear: The Common Underlying Mechanism. Paper presented at the IADC/
SPE Drilling Conference and Exhibition, Fort Worth, Texas, USA, 4–6 March. SPE-168041-MS. https://doi.org/10.2118/168041-MS.
van Oort, E. and Vargo, R. F. 2008. Improving Formation-Strength Tests and Their Interpretation. SPE Drill & Compl 23 (3): 284–294. SPE-105193-PA.
https://doi.org/10.2118/105193-PA.
Therese Scheldt is the project leader in research and technology at Equinor, having previously held positions in petroleum engi-
neering and in drilling and well. She holds an MSc degree in engineering geology/rock mechanics and a PhD degree in rock
mechanics from the Norwegian University of Science and Technology.
Jamie Stuart Andrews is a leading adviser at Equinor, specializing in sand control, fracturing, injection, and geomechanics. He
holds an MA degree in natural sciences and a PhD degree in quantum mechanics from the University of Cambridge. Andrews
serves as a volunteer to the SPE Stavanger Section, Norway.
Alexandre Lavrov is the senior scientist in the Petroleum Department of SINTEF Industry (Norway), where he has worked for more
than 15 years. Lavrov’s main research interests include drilling and cementing, lost circulation, petroleum-related rock mechan-
ics, and numerical modeling in geomechanics. He holds PhD and DSc degrees, both from the Moscow State Mining University.
Lavrov received the 2008 SPE Outstanding Technical Editor Award.