You are on page 1of 6

Analysis of an Annular Pressure Buildup

Failure During Drill Ahead


P.D. Pattillo, SPE, B.W. Cocales, SPE, and S.C. Morey, SPE, BP America

Summary
While drilling ahead in salt on the Pompano A-31 (Gulf of
Mexico) wellbore below the 16-in. shoe, the rotary stalled
abruptly, and the drillpipe simultaneously became stuck. Subsequent calipers indicated the 16-in. casing to be deformed onto the
drillpipe at approximately 250 ft.
To the authors knowledge, this situation represents the first
instance of an annular pressure buildup (APB) failure during drilling. APB is typically associated with extremes of temperature change
resulting from production operations. In the current instance, the
temperature change is solely from circulating drilling fluid.
This paper provides a detailed post-analysis of the 16-in. casing
failure as follows:
An overview of the conventional casing design for this well,
indicating that normal operating conditions should not have resulted in a failure.
A review of the failure event, substantiated by field measurement and photographs.
An investigation of APB, associated with inadvertently closing the 16-in. casing annulus at the surface, demonstrating the
magnitude of the possible resulting thermally induced pressure load.
Consideration of an alternate failure mode, column buckling,
to demonstrate that this failure mode was not active in this wellbore.
A finite-element model of the drillpipe/16-in. casing/20-in.
casing trio indicating that, in addition to the collapse of the 16-in.
casing and subsequent sticking of the drillstring, the outer 20-in.
casing was also damaged by the event.
The latter item is particularly important because damage to
outer strings can be easily overlooked if concentration is on the
collapsed casing and drillpipe fish. A mechanical caliper confirms
damage to the outer 20-in. string.
Conclusions in the paper center on confirmation of the postulated failure mode by field measurement and recommendations for
avoiding such an event.
Introduction
While drilling ahead at 9,132 ft. on the Pompano A-31 wellbore
(Vioska Knoll 989 A-31, OCS-G-6898) below the 16-in. shoe
(Fig. 1), the rotary stalled abruptly, and the drillpipe simultaneously became stuck. An increase of mudflow out of the flowline
surged over the gumbo buster.
Approximately 500-psi pressure was bled off the 16-in. by
20-in. annulus. The fluid initially recovered from the 16-in. by
20-in. annulus was a clear, 10-ppg brine, but later changed to a
10.5-ppg, synthetic-based mud composition tantamount to the
fluid placed above the cement top in the 16-in. annulus.
Tripping the drillstring out of the hole required 60 to 100 kips
at each 838-in. HWDP tool joint and 150 kips at the uppermost
1412-in. stabilizer. Recovery of subsequent stabilizers and the remaining seven joints of HWDP was aided with drilling jars.
A caliper-logging tool set for a maximum diameter of 1134 in.
tagged an obstruction at 253 ft. Logging upward, the caliper
showed damage from 242 to 253 ft. The wellbore was temporarily
abandoned with four cement plugs. Subsequent to plugging the
well, pressure testing indicated communication between the 16and 20-in. casing strings.

Copyright 2006 Society of Petroleum Engineers


This paper (SPE 89775) was first presented at the 2004 SPE Annual Technical Conference
and Exhibition, Houston, 2629 September, and revised for publication. Original manuscript
received for review 4 June 2004. Paper peer approved 1 May 2006.

242

The 16-in. casing was cut at 1,399 ft and pulled. The remaining
16-in. riser stub, along with the Dril-Quip hanger running tool,
were recovered to 1,485 ft. A caliper was then run in the 20-in.
casing, thereby indicating damaged casing from 253 to 280 ft.
(maximum inside diameter 20.387 in. at 272 ft.) (See Fig. 2.).
Figs. 3 and 4 illustrate the condition of the 16-in. casing recovered from the vicinity of the obstruction. The ovalized cross
section appears adjacent to a window (presumably) worn by milling before recovery.
Pertinent evidence gathered during the initial investigation and
recovery of the ovalized 16-in. casing includes the following:
The drillstring stalled abruptly, indicating an instantaneous event.
There is no evidence from prefailure drilling parameters of an
out-of-the-ordinary occurrence.
The recorded circulating temperatures180F bottomhole
and 168F at the flowlineare high (particularly the latter).
The 16-in. annulus valve was closed while drilling the 1412in. by 1712-in. hole section.
The depth of the failure is shallow. Given 10.5-ppg drilling
fluid outside the 16-in. casing, an evacuated collapse differential is
0.052 psi/ppg-ft by 250 ft by 10.5 ppg136 psi, as compared to
the API collapse rating for this casing of 1,480 psi.
For future reference, the Pompano A-31 casing run before the
failure is listed in Table 1.
Conventional Casing Design
The term conventional casing design in this context refers to a
design in which the integrity of a single-tubular string is investigated without consideration of the interaction of the string with
other tubulars. Loads that are associated, for example, with trapped
pressures in an annulus between the target string and its neighbors
are not considered.
Fig. 5 summarizes the pertinent external pressure design factors
for loads typical of an intermediate casing string. The current
string is designed for the lost-circulation load case in which mud
used to drill the next hole section is allowed to drop until it reaches
a hydrostatic balance with pore pressure in the open hole. The
depth and corresponding pore pressure used are those that result in
the most severe collapse load case. The evacuation load case is
also shown for information because this load scenario is also used
by operators. Pertinent depths in the figure are 1,486 ft, the section
crossover in the 16-in. string (see Table 1) and 2,549 ft, the depth
to which the mud column falls in the lost circulation load case.
Although the lower portion of the 16-in. casing, the 97-lb/ft
N80 run below 1,486 ft, has a safety factor less than unity under
the evacuation-load case, the failure in this string was in the shallower, 84-lb/ft P110 casing. Collapse safety factors for this shallower segment of the 16-in. casing are consistently higher than 1.5.
The implication of this information is that, barring an unexpectedly low-collapse resistance of the subject casing, conventional
collapse loads should not have failed this string at 242 to 280 ft. To
investigate the possibility that the casing opposite this interval has
a lower-than-expected collapse performance, companion joints to
the failed casing were collapse tested, with the results displayed in
Table 2. As indicated in the table, the collapse resistance of the
companion joints is higher than the API minimum collapse rating
of 1,480 psi upon which the safety factors in Fig. 5 are based.
Temperature Modeling
Of the evidence outlined in the previous section, one outstanding
feature is the high-surface circulating temperature during drill
ahead. An attempt was made to match the circulating temperatures
with a commercial thermal simulator.
December 2006 SPE Drilling & Completion

Fig. 2Log-measured displacements of 20-in. casing.

Fig. 1Schematic of Pompano A-31 at time of failure, all depths


in ft RKB.

Unfortunately, the capabilities of the software are insufficient


to model the fluid system of the drilling rig. As an alternative, the
surface circulating temperature (inlet to drillpipe) was adjusted to
the point where the observed circulating temperature of 168F was
predicted by the software. Input data for the model are summarized
in Table 3.
Fig. 6 presents the results of the thermal prediction, indicating
not only the profile outside the drillstring, but also, for later reference, the temperature in the 16- by 20-in. annulus.
Annular Pressure Buildup
In well design, APB refers to the pressure change in a fluid in a
closed annulus. The phenomenon is particularly relevant to off-

Fig. 3Photograph of recovered, collapsed 16-in. casing.


December 2006 SPE Drilling & Completion

shore wells in which annuli can be trapped by terminating a casing


string(s) at the mud line. The ABP phenomenon, however, can
occur in any annulus that is not vented (Halal and Mitchell 1994;
Ellis et al. 2002).
Consider a liquid completely filling a closed container. When
the temperature of the fluid is increased, it attempts to expand in
accordance with its coefficient of thermal expansion. This volume
change is countered by the rigidity of the container. Resistance to
the free expansion of the fluid induces a pressure increase. According to the rigidity of the enclosing walls, this pressure increase
induces a corresponding change in the dimensions of the container.
Equilibrium ensues, involving changes in both the fluid and the
container. The incremental annular pressure accompanying a
change in temperature is, therefore, a function of the following:
Mechanical and thermal properties of the annular fluid.
Flexibility of the confining boundary.
Temperature increase.
The Present Case. The 16- by 20-in. annulus in the current well
constitutes a possibly closed container. First, as mentioned in the
introductory comments, the 16-in. annulus valve was closed while
drilling the 1412- by 1712-in. hole section. Secondly, the targeted
top of cement for the 16-in. casing is close to the 20-in. casing
shoe. Error in cement calculations or application, barite settling in
the drilling fluid above the cement top, or failure to maintain
stability of any open hole above the cement could individually, or
in concert, contribute to closure of the 16- by 20-in. annulus.

Fig. 4Photograph of recovered, collapsed 16-in. casing.


243

Of further importance to the APB issue is that a synthetic,


rather than water-based, mud was used to drill the 1712-in. hole
section. This switch in mud composition distinguishes the current
1712-in. hole section from all but one of its counterparts on the
Pompano platform. The 1712-in. hole section of an earlier well was
also drilled with synthetic mud. This earlier operation, however,
ran a 1338-in., 72-lb/ft, N80 casing (2,670-psi collapse resistance),
rather than a 16-in., in the 1712-in. hole.

Fig. 5Conventional collapse design factors, 16-in. casing.

244

Modeling Results. A drill-ahead scenario, using data described


in Table 4, was used to calculate APB effects on the 16- by
20-in. annulus.
The synthetic fluid present in the 16- by 20-in. annulus was
modeled as an oil-based mud because of limitations in the modeling software.
With these inputs, the predicted incremental pressure caused by
annular pressure buildup is 3,423 psi (9.6-bbl potential expansion
volume). Compared to the collapse rating of the 16-in., 84-lb/ft,
P110 casing (1,480 psi), this value is more than sufficient to initiate collapse (assuming a roughly equal density gradient from
inside and outside fluids).
As previously mentioned, one distinguishing fact in the current
scenario is the decision to switch from water-based to synthetic
mud. If the previously shown calculation is repeated with 11.6-lb/
gal, water-based mud in the 16- by 20-in. annulus, the ensuing
pressure caused by APB is 2,395 psi (6.1-bbl potential expansion
volume), which still exceeds the API collapse rating of the inner
string. The following points are important:
The API collapse rating is conservative, but even using a
more modern prediction (Tamano et al. 1985) and targeting average rather than minimum collapse, the rating of the 16 in. is 2,080
psi, still less than the predicted incremental APB pressure.
Both the oil-based and water-based calculations assume a
completely fluid-filled annulus. If any portion of the 16- by 20-in.
annulus is void (e.g., gas-filled), that void can serve as an accumulator for its volume equivalent of fluid expansion. This reduces
the corresponding expansion pressure. The effect of a void/gas cap
in the 16- by 20-in. annulus is illustrated in Fig. 6. With an oilbased mud, a gas cap of 4 bbls reduces the APB pressure from
3,423 psi to 1,995 psi. A void of approximately 5.5 bbl reduces the
incremental APB pressure lower than the API collapse rating and
suggests no failure. The volume capacity of the upper portion of
the 16- by 20-in. annulus is 0.0921 bbl/ft. Thus, 5.5 bbl corresponds to 60 ft.
There is a significant difference between the oil-based and
water-based results, allowing ample room (e.g., in the presence of
a partial gas cap, as previously mentioned) for failure with the

December 2006 SPE Drilling & Completion

Fig. 6Temperature prediction from drill-ahead thermal simulation.

synthetic fluid, but not with a water-based fluid. Returning to


Fig. 7, and using the Tamano et al. (1985) equation to define
collapse, a gas cap with a volume less than 0.81 bbls (8.8 ft.)
produces a collapse with either WBM or OBM, a gas cap with a
volume between 0.81 and 3.76 bbls (40.8 ft.) produces a collapse
with OBM, but not with WBM, and a gas cap of greater than
3.76 bbls is sufficient to avoid a collapse with either annular
fluid. The extent of any possible gas cap in the 16- by 20-in.
annulus is unknown.
Modeling the Collapse Damage
To fully understand the observed damage to the tubulars recovered
from Well A-31, a 2D plane-strain finite-element model of the
failure point was constructed. The model (Fig. 8) consists of concentric cross sections, starting with the inner string, of 658-in., 27.7
lb/ft (ID5.901 in.), S135 drill pipe by 16-in., 84.0 lb/ft
(ID15.010 in.) P110 casing by 20-in., 133 lb/ft (ID18.730 in.)
56 casing.
The mesh is symmetrical along the boundaries of the first quadrant. In fact, the inner-tube and outer-tube meshes are axisymmetric. The middle tube, however, representing the 16-in. casing is
assigned a slight ovality to induce collapse when the 16-in. tube is
subjected to external pressure. All casing elements are four-node,

Fig. 7Variation of annular pressure buildup with void in 16in.20-in. annulus.


December 2006 SPE Drilling & Completion

reduced-integration plane-strain finite elements. In addition, the


mesh contains contact elements to model possible interaction between the strings.
Typical results from the analysis are summarized in Figs. 9 and
10. Fig. 9 displays successive deformation of the 16-in. casing. In
Fig. 9a, the initial (e.g., manufactured) ovality of the 16-in. casing
is magnified by external pressure in the 16- by 20-in. annulus to
the point that the tube reaches the maximum load it can tolerate
and is in the process of collapsing.
Fig. 9b illustrates the fact that the post-buckled deformation of
the 16-in. casing actually contacts the 20-in. casing first. This
contact (see discussion of Fig. 10) momentarily reinforces the
16-in. cross section. The driving pressure, however, is sufficient to
cause further deformation of the middle tube, leading to its eventual contact with the inner string of drillpipe. The model reproduces both the deformation of the 20-in. casing as well as the
contact and seizing of the drillpipe by the 16-in. casing.
Fig. 10 lends additional detail to the deformed plots in Fig. 9,
in which displacements at the points of initial contact with the
20-in. casing (Point A, Fig. 8) and the drillpipe (Point B, Fig. 8) are
plotted vs. the external pressure applied to the 16-in. casing.
Initially, the slight ovality assigned the 16-in. cross-section
results in both horizontal (Point A) and vertical (Point B) displacements. At a critical value of external pressure, the maximum response of the 16-in. cross section is reached, and collapse occurs.
The post-buckled response of the tube is to suffer additional ovalization under decreasing pressure.
As deformation of the 16-in. casing continues, it eventually
contacts the outer, 20-in. tube. Briefly, the 16-in. casing is reinforced as displacements at Point A become more difficult because
of the stiffness of the 20-in. cross section. Eventually, however,
and at a pressure much less than the initial collapse pressure of the

Fig. 8Undeformed mesh for casing collapse finite-element


model.
245

Fig. 10Displacements of key locations on 16-in. casing. (See


Fig. 7 for definition of points A and B.)

mum and minimum diameters of the 20-in. casing at the time


the post-buckling 16-in. casing just touches the drillpipe. (See
Fig. 9c.) The exact displacements of the 20-in. casing should differ
from these benchmarks because of the following factors:
After initial contact, the 16-in. casing further deforms the
20-in. casing as it deforms around the drillpipe. (See the tails of the
16-in. displacements reported in Fig. 10.)
Once the 16-in. casing is recovered, there is a small elastic
recovery of the 20-in. casing.
Recovery of the 16-in. casing from the wellbore may have
further deformed the 20-in. string.
The intended message of Fig. 2 is that observed damage of the
20-in. casing is consistent with the scenario of the 16-in. casing
collapsing because of APB effects.
A Note on Column Buckling
Early in the failure investigation, the possibility of damage in the
16-in. casing being the result of column buckling was investigated.
To examine the consequences of column buckling, a drill-ahead
load case identical to the one used in the APB analysis was constructed, the only difference being that for the buckling analysis,
the 16- by 20-in. annulus was assumed open. Buckling of a vertical
tubular is governed by the so-called effective force:
Fe = Fz piAi poAo, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (1)

Fig. 9Successive deformation of 16-in. casing from collapse


through contact of outer and inner tubes. Contours are von
Mises intensity.

16-in. tube, there is sufficient vertical displacement for the tube to


contact the inner drillpipe.
Evidence in support of the previously described model is provided by a caliper run in the 20-in. casing following recovery of
the inner 16-in. string. Fig. 2 summarizes the damage to the 20-in.
casing caused by the collapse of the 16-in. casing as caliper-log
measurements of maximum and minimum internal diameter. For
reference, the figure also contains (dashed lines) predicted maxi246

where a negative value of Fe indicates buckling. Thus, lowering po


(e.g., removing the potential for APB by prohibiting generation of
thermally induced pressure in this annulus) reduces Fe, increasing
the likelihood of excess buckling.
Fig. 11 summarizes the results. Only the upper section of the
16-in. casing (e.g., the 84 lb/ft, P110 section) is analyzed, because
the upper and lower sections of the string are separated in terms of
axial load by a landing ring at the crossover. A worst-case scenario, that assumes that all the weight of the upper section is
landed on the landing ring, is used in the analysis. The entire 16-in.
casing string is buckled, as evidenced by the negative effective
stress (Fe / [AoAi]). However, nowhere is the von Mises equivalent stress (maximum in cross section is displayed), which includes
the effect of bending caused by buckling, in excess of the yield
stress (110,000 psi) of the casing at the failure point.
Finally, and as further evidence that column buckling is not the
failure mechanism in this well, buckling and its associated deformations should be more severe in the lower sections of the uncemented length of the 16-in. casing. (See Fig. 11.) The failure in this
well occurred near the surface.
December 2006 SPE Drilling & Completion

Nomenclature
Ai tube internal cross-sectional area, [L2], in.2
Ao tube external cross-sectional area, [L2], in.2
Fe effective force, [ML/T2], lb.
Fz axial force, [ML/T2], lb.
pi internal pressure, [M/(LT2)], psi
po external pressure, [M/(LT2)], psi
References
Ellis, R.C., Fritchie, D.G., Jr., Gibson, D.H., Gosch, S.W., and Pattillo,
P.D. 2004. Marlin Failure Analysis and Redesign; Part 2Redesign.
SPEDC 19 (2): 112119. SPE-74529-PA. DOI: http://www.spe.org/
elibrary/servlet/spepreview?id74529-PA.
Halal, A.S. and Mitchell, R.F. 1994. Casing Design for Trapped Annular
Pressure Build-Up. SPEDC 9 (2): 107114. SPE-25694-PA. DOI:
http://www.spe.org/elibrary/servlet/spepreview?id25694-PA.
Tamano, T., Mimake, T., and Yanagimoto, S. 1985. A New Empirical
Formula for Collapse Resistance of Commercial Casing. Nippon Steel
Technical Report 26.

Fig. 11Stress state in 16-in. casing caused by column buckling.

Conclusions
1. Damage to the 16-in. casing in the subject well is caused by
collapse associated with APB during drill ahead.
2. The 16- by 20-in. annulus was closed by (a) closing the annulus
valve at the upper end, and (b) either cement, formation (e.g.,
wellbore stability) or barite settling at the lower end.
3. To the best of our knowledge, this is the first-known instance of
an APB-related collapse caused by drilling, rather than production, thermal loads. The primary source of pressure is the relatively high surface circulating temperature.
4. The variation in thermal behavior between an oil-based and
water-based fluid, possibly coupled with the presence of a gas
cap in the 16-in. by 20-in. annulus, can explain why this problem was not encountered on previous wellbores.
5. Collapse of the 16-in. casing not only seizes the inner drillpipe,
but also ovalizes the outer 20-in. casing. The latter effect is
confirmed by a mechanical caliper.
6. Excess bending stress caused by column buckling during drill
ahead is not the failure mechanism in this well.

December 2006 SPE Drilling & Completion

SI Metric Conversion Factors


ft 3.048*
E01
m
in. 2.54*
E+00
cm
psi 6.894 757
E+00kPa 0.01 bar*
*Conversion factor is exact.

Phillip D. Pattillo is a Distinguished Advisor with BP America in


Exploration and Production Technology. e-mail: pattilpd@bp
.com. Since 1972, he has worked in the areas of multiphase
flow and tubular and rock mechanics. Pattillo holds a BS degree in mechanical engineering, an MS degree in engineering
science from Louisiana State U., and MS and PhD degrees in
engineering science from the U. of Notre Dame. Brett Cocales
is a senior deepwater drilling engineer with BP America in Exploration and New Developments. In the industry since 1989,
he has worked primarily in drilling and completion operations in
GoM deepwater, the shelf, and land. Cocales holds a BS degree in petroleum engineering and MBA degrees from Montana Tech and the U. of Montana. Steve Morey is a tubular
technology specialist with BP in Houston In the industry since
1978, he has worked in drilling and completions in GoM deepwater, the shelf, and the L48. Since 1989, he has consulted on
tubular technology for BP worldwide. Steve holds a BS degree
in mechanical engineering from the U. of New Orleans.

247