Sei sulla pagina 1di 14

SPE/IADC 92588

Application of Real-Time Wellbore Stability Monitoring on a Deepwater ERD Well

J.A.Greenwood, Halliburton Sperry-Sun, A.Brehm, GeoMechanics International; E. van Oort, D.R Algu and Y.E. Volokitin,
Shell E&P Americas

Copyright 2005, SPE/IADC Drilling Conference


This paper was prepared for presentation at the SPE/IADC Drilling Conference held in
Amsterdam, The Netherlands, 23-25 February 2005.
This paper was selected for presentation by an SPE/IADC 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 SPE,
IADC, their officers, or members. Electronic reproduction, distribution, or storage of any part
of this paper for commercial purposes without the written consent of the Society of
Petroleum Engineers or the International Association of Drilling Contractors is prohibited.
Permission to reproduce in print is restricted to an abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of
where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836,
Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
Good geomechanical modeling can provide valuable
information for the efficient design and drilling of
wellbores. Incorporating real-time wellbore stability
monitoring during drilling can reduce the associated risks,
especially for deepwater extended-reach wells. This paper
presents the preparation, delivery, and outcome of the field
trial for a real-time wellbore stability monitoring service
delivered at Shell Exploration and Production Companys
office in New Orleans.
Three key objectives were set for the field trial: (1) to
develop the processes to incorporate real-time wellbore
stability into the current operations center monitoring
provision, (2) to provide frequent updates of the wellbore
stability model using a geomechanical modeling technique
that was independent of the operators own methods, and (3)
to monitor and verify the geomechanical model based upon
the drilling experience enabling proactive decision making
during drilling. For the operators asset team, the main
objective was to reduce trouble time and make execution of
the well successful.
Ram Powell VK 913 A-9 well was chosen as a candidate
for the field trial. The Ram Powell tension-leg platform is
located in 3200 feet of water in the Eastern Gulf of Mexico.
A-9 was planned as an extended-reach exploration and
would have the highest angle at the shallowest depth in the
field. A geomechanical model for the prospect had already
been created using the operators own well-established
methodology. This pre-drill model was transferred into the
service companys software, and the real-time model was
calibrated to generate as close to the same output as
possible. After verifying the real-time model using the
drilling experience on the closest offset wells, the 24 hr realtime stability monitoring commenced. The real-time
geomechanical monitoring encompassed pore pressure
prediction, rock property calculations from formation

evaluation tools, wellbore trajectory updates, and the use of


surface and downhole drilling data to verify the
geomechanical model.
Integration of the real-time wellbore stability monitoring
contributed to the successful drilling and casing of this
deepwater extended-reach well. The trial resulted in a
greater understanding of the geomechanics of the field. The
trial also resulted in a better understanding of procedures for
maximizing the value of real-time data and of associated
monitoring services, services that will be incorporated in
future Shell E&P wells.
Introduction
Shell Exploration and Production Company, in co-operation
with Halliburton Sperry-Sun, established a real-time
Operations Center (OC) within the operators office in New
Orleans.1 The operations center is regarded as enabling a
multidisciplinary workspace that seamlessly integrates all
aspects of the companys well construction activities.
Halliburton Sperry-Sun and GeoMechanics International
(GMI) had cooperatively developed a real-time enabled
version of new geomechanical analysis software, and
approached the operations center with a proposal to
complement and enhance the existing well construction
process. The proposal was accepted, and a trial was initiated
to establish the feasibility of real-time wellbore stability
monitoring within the operations center environment. An
extended-reach (ERD) well design was chosen that would
become the shallowest ERD well attempted on the prospect.
Field Trial Outline
The ultimate goal of the trial was to enable the real-time
update of the geomechanical model output using
measurements from the actual conditions encountered
during drilling and, crucially, to provide the updates within a
time frame and in a format that allowed proactive decisionmaking while drilling. To meet this goal, the trial had three
central objectives. The first was to develop the processes
that would allow the incorporation of real-time wellbore
stability monitoring into the existing operations center wellconstruction structure. The second was to provide frequent
updates of the wellbore stability model, which in this
instance would use techniques that were independent of the
operators own methods. The third was to monitor and
verify the geomechanical model based upon the drilling
experience to enable proactive decision-making during
drilling.

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

SPE/IADC 92588

From the operator s perspective, the main driver for the


trial was to complement existing in-house modeling
capabilities with the service company s real-time
geomechanical modeling and monitoring service. This and
the independent verification of the two geomechanical
models were seen as contributing to the success of the ERD
well.

for borehole stability modeling employed on A-9 both by


the operator and by the service company. What made A-9
even more challenging was that the G sand target came in
1500 feet shallower than in all previous Ram Powell wells.
This fact combined with the high step-out necessitated that
A-9 penetrate shallow sediments at a 76 angle, something
that had not been attempted before.

Description of Target Well


Ram Powell VK 913 A-9 well was planned as an extendedreach exploration well to the G sand prospect. The G sand is
a near-field exploration prospect delineated by a bright
seismic amplitude approximately three miles from Ram
Powell TLP. Ram Powell is situated in the Eastern Gulf of
Mexico, in water depth 3000 ft deep, Fig. 1. In addition, the
well azimuth was in a direction previously un-drilled by
wells in the prospect.
The well was designed to drill out of the 20-in. casing
shoe and, using a 12 -in. x 20-in. assembly, build angle to
33. After a 16-in. liner was set, a second hole section using
a 10 -in. x 17-in. assembly would continue to build hole
angle to 76. A 13 -in. liner would then be set, and a third
hole section using a 9-in. x 14-in. assembly would
maintain the hole angle at 76. An 11 -in. liner was to be
set, and the final hole section would reach TD using a 9 in. x 10 -in. assembly, again maintaining the hole angle at
76. A 7 -in. production liner was to be set at TD (for
further details see ref. 2).
Maintaining the correct mud weight to prevent hole
collapse and to avoid lost circulation was crucial to the
well s success. The real-time geomechanical monitoring
service allowed the operator to be confident that drilling
problems due to wellbore instability could be avoided, and
that hole conditions for running casing and liners would be
optimal. An additional risk of penetrating sand whilst still in
the 9-in. x 14-in. section was also identified in the predrill modeling. There was a possibility that the sand was
hydrocarbon-wet and had significant structural relief with an
estimated pore pressure higher than could be safely
contained by the 13 -in. liner shoe. The design philosophy
employed and experience obtained from drilling and
completion operations are described in more detail by Algu,
et al. 2

Operations Center Description


As stated, the trial was conducted using the capabilities of
the New Orleans operations center. The operations center
staffing can be broadly divided into (1) technical
support/engineering staff and (2) 24/7 monitoring expert
staff. The technical support/engineering staff is involved in
all parts of the planning process, assisting asset teams in
well design. To improve efficiency and reduce risk, they
can also make recommendations on additional planning
work or additional data-capture during drilling.
The monitoring is performed by optimization specialists.
Their main focus is to scrutinize the incoming data for
warning signs of impending hazards, and to compare the
actual drilling parameters to those predicted by the pre-drill
models generated by the technical support/engineering
section. The monitoring staff had established lines of
communication to the well site to intervene if conditions
were identified that could have a detrimental effect on the
drilling operation and lead to value erosion.
Pore pressure prediction and geomechanical analysis are
fundamental in the design of any well. Within the current
well construction structure of the operations center, this
analysis is provided by the asset petrophysicist and the
operator s internal excellence teams for both pore pressure
and wellbore stability. Wellbore stability analysis is
performed using the operator s own geomechanics software.
Monitoring 24/7 during drilling consists of the use of the
pre-drill geomechanics limits and the use of pore pressure
specialists on a call-out basis. From the leak-off test results,
upper circulating pressure limits are set that include a safety
margin that triggers immediate action at the well site to
manage the ECD or static pressures within the wellbore.

Drilling History on Ram Powell


Prior to A-9, ten wells have been drilled from the Ram
Powell TLP in the timeframe of 1996 to 1999. Employment
of horizontal wells has allowed the operator to reduce the
number of wells from the twenty originally planned to ten
currently.3 Design of Ram Powell prospect is described by
Lester, Lanier, et al.4 Phase II drilling operations, which
commenced in 2004 with the A-9 well, will continue with
several in-fill and step-out wells.
The Ram Powell A-13 well was of special interest in
planning for the A-9 well. A-13 was also an ERD well,
drilled in 1999 as the last well of the original development
drilling campaign. The well had a similar inclination (74)
and reached (26,000 ft MD). Although highly successful in
the end, A-13 reached its objective only after a sidetrack,
due to combination of borehole instability and hole cleaning
problems. The A-13 well was an important calibration point

Trial Preparation
The trial was set-up in a series of meetings with drilling and
petrophysical teams. The results of pre-drill modeling were
presented and were also checked by the operator s
technology center experts.
Monitoring Requirements. The wellbore stability (WBS)
specialists would work alongside the 24/7 monitoring staff
and receive input from them on the drilling conditions and
any potential drilling problems. Two WBS specialists
working a 12-hour rotational shift pattern would operate the
software. An initial trial period consisting of monitoring the
10 -in. x 17-in. and 9-in. x 14-in. sections was agreed.
At the conclusion of this evaluation period, an assessment
would be made by the operator to determine if the trial was
bringing benefit to the operation. If benefit was the case, the
service would be continued over the final 9 -in. x 10 -in.
hole section.

SPE/IADC 92588 APPLICATION OF PREDICTIVE REAL-TIME WELLBORE STABILITY MONITORING ON A DEEPWATER ERD WELL

Communications Protocols. It was decided during the


course of these meetings that, for the purposes of the field
trial, a modified communications protocol would be
implemented that allowed the petrophysicist to validate the
output of the real-time geomechanics software. Once
validated, the petrophysicist would communicate with the
on and offshore drilling teams. The Wellbore Stability
specialists would not communicate directly with the rig, but
would communicate directly with the operations
petrophysicist.
Software and Hardware Infrastructure. Surface and
downhole sensor measurements from two other service
providers were supplied for the A-9 well; these
measurements were transferred in real time into an offshore
database using the WITS data transfer protocol. The
offshore database computer is connected to the operations
center network, and an exact copy of the offshore database
is then created on a server located in a secure room within
the operations center. The monitoring staff then displays and
analyzes the drilling and the downhole parameters contained
on the onshore server. An additional capability to view the
information using a web-based interface was not utilized as
part of the trial.
Two additional software packages were required for the
purposes of the trial: one for pore pressure prediction and
one for the geomechanical analysis. Both of the analysis
software packages had been designed to connect directly to
the database, receiving selected data inputs the instant they
are received by the database. Once the calculations and
analysis of the received data are complete, all of the input
data, intermediate calculation values, and calculated results
are transmitted back to the database into predefined records.
Such data transfer allows the entire analysis to be
completely accessible to anyone with access to the database
display functionality. The benefit, is that this approach
avoids the historical problem of creating a dead-end for the
analysis and results, with the only person viewing the results
being the software operator. Additionally, the present
approach allows for a more complete evaluation and
discussion of the analysis between provider experts and
operator experts without having to generate and transfer
hardcopy output.
The pore pressure and geomechanical software packages
were connected to the offshore database using the operations
center network connection. The WBS experts monitored the
output on separate workstations connected to the onshore
database. The output was not utilized by the normal
operations center monitoring staff during the trial.
The geomechanical software was successfully tested
within the operations center on a well prior to commencing
the trial to establish the software functionality and to ensure
that it did not interfere with any of the software systems on
the network.
Geomechanical and Pore Pressure Modelling
To meet the field trial s first objective, the differences
between the operator s and service company s analysis
techniques and software packages were discussed with the
operator s expert geomechanics team and the asset
petrophysicist. Both the operator and service company had

their own different methods for creating these models and


implementing them within their own software packages.
As the operator had already created a geomechanical
model for the field that had been used in several of the offset
wells, it was agreed that there was no requirement for the
service company to generate a completely new
geomechanical model. Rather than focusing on the
differences in the modeling techniques, it was realized that
by approaching the problem using the same input
information but different calculation methods, a greater
confidence level would be created in the pre-drill model if
both methods produced results that were similar.
The existing operator s model was transferred into a
format that could be used in the real-time geomechanics
software. Once the transfer was complete, an additional
calibration of the real-time geomechanical model was
performed to assure that results were produced that both the
operator and service company had confidence in.
Definition of a Geomechanical Model. A geomechanical
model consists of three main components:
The magnitude of the formation pore pressure
The magnitude and orientation of the three subsurface
principal stresses, Sv, Shmax and Shmin
The mechanical and physical properties of the rocks
being drilled

Accurate modeling of these three components is


essential to predicting the mud weights required to maintain
an acceptable level of wellbore stability.
Pore Pressure Modelling. The operator s approach was to
combine proprietary offset well pressure and sonic data to
create a basin-specific set of compaction trend curves. The
compaction trends were then applied to offset sonic data
together with 3-D seismic to identify pressure cell
boundaries and then extrapolate the pressure regime along
the well path 5. The operator s pre-drill pore pressure plot is
shown in Fig. 2. An interesting finding from the pore
pressure analysis is that the highest pore pressure gradient
was predicted to be 2000 ft shallower than the reservoir.
This high-gradient zone was identified as one of the most
critical zones in well planning because, additionally, the
A13 well encountered multiple borehole trouble incidents
across this zone.
The service company calculations were performed using
Eaton s 6 method on the sonic, resistivity, and Dc exponent
measurements. The calculation also used the equivalentdepth method on density data. Offset data were provided for
four wells: 912-1, 912-3, 956-1 and A13 ST1. The 912-1,
912-3, and 956-1 offset wells were vertical or close to
vertical. Only the 912-1 and 956-1 wells contained
measurements within the normally pressured sequence of
formations. The 912-3 well measurements began within an
overpressured sequence. The A13 ST1 was a deviated well
but contained data only over a depth range analogous to the
final section of the A-9 well.
The initial analysis by the service company was
performed as a blind test on each well individually. This
analysis would allow the veracity of the calculations to be

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

established, and the results compared to the operator s


estimates on these wells. The analysis would also identify if
any spatial differences existed in normal compaction trends
(NCT) across the prospect.
No significant differences existed within the slopes of
trends applied to normally compacted zones in the two wells
where the normally compacted sequence contained
measurements. One potential calibration point existed in
shallow sand on the 912-1 well; it was assumed that the pore
pressure in this permeable formation was in balance with the
pressures in the impermeable shale. The NCTs established
on the 912-1 well were then applied to the other wells with
minor adjustments in position but not slope to account for
shifts in log response.
Once the initial analysis was concluded, the results of
the service company analysis were compared to the
operator s predictions, Fig. 3 and Fig. 4. The predictions
matched well, with the only main difference identified being
the frequency of the data employed. The operator employed
data points every 100 ft and used a pressure-cell approach,
and the service company analysis used the frequency of the
input log data to calculate the raw pore pressure results.
Principal Stress Tensor. Central to an accurate
geomechanical model is the determination of the magnitude
and orientation of the three sub-surface principal stresses.
These stresses are the vertical stress, Sv (also more
commonly known as the overburden), and two horizontal
stresses, Shmin and SHmax. The relative magnitude and
orientation of these stresses have a large impact on the
wellbore stability calculations.
For the A-9 well, the overburden increased more rapidly
with the step-out because of diminishing water depths along
the well profile. The overburden provided by the operator
for the A-9 well was calculated from a regression based on
depth-below-mudline. The operator s results were used to
back-calculate the interval densities required by the service
company s software for the overburden calculation.
Overburden calculations were not adjusted during the
drilling of the well in order to maintain consistancy
throughout the different calculation methods, namely
between the two pore pressure prediction methods and two
geomechanical models.
Geomechanical models used horizontal stress inputs as
calculated by the operator. The approach was to take the
lower bound of observed leak-off tests as an approximation
of minimum horizontal stress. Repeat leak-off tests and
leak-off tests in horizontal wells were given highest weights
when fitting the relationship.
Available caliper and sonic log anisotropy data
suggested the absence of large differences between
maximum and minimum horizontal stresses. Virtual absence
of faulting in the field confirms this suggestion. The
maximum horizontal stress was approximated as SHmax=
0.8*Shmin+0.2*Overburden.
Azimuth of the maximum stress was taken at 300-330,
parallel to the faults observed to the north of the platform.
Rock Mechanical Properties. Rock mechanical properties
used in the modeling were established by the operator s
expert group.

SPE/IADC 92588

Explanation of Different Geomechanical Approaches.


The operator s approach 7 is to use a finite element program
that permits the calculation of elastic and plastic stresses and
strains around a borehole subject to various loading and
boundary conditions. The calculated stresses and strains in
turn are used to estimate optimum mud weight for minimum
risk of borehole instability problems caused by plastic
deformation and failure of the formation.
Given the mechanical properties of the formation, the
program calculates the maximum allowable plastic strain
when the formation is yielding but not yet failing.
Exceeding this critical plastic strain is used as a failure
criterion.
Creation of the Real-Time Geomechanical Model.
Once the geomechanical inputs provided by the operator had
been entered into the service company s software, it was
important to illustrate that the real-time model could
produce nearly identical results, despite any differences in
methodology.
The service company adjusted the real-time model using
several different Uniaxial Compressive Strength (UCS) rock
strength correlations and other rock property settings, all of
which were used to try to match drilling experience in offset
wells. To accomplish this aim, a blind study was conducted
on the A13 well with only the input parameters to the
operator s geomechanical model being supplied. Using the
drilling history and mud weights from A13, the service
company constrained the real-time model results so that they
matched the drilling experience on A13.
Fig. 5 shows the software output of predicted wellbore
failure versus depth for A13. This comparison is based on
the geomechanical model that was ported into the software
from the operator and on the actual mud weights used while
drilling the well. When the predicted failure reaches
magnitudes greater than the allowable breakout width,
drilling problems are expected, since higher hole angles are
typically harder to clean and can tolerate lower amounts of
failure. The black line on Fig. 5 (that varies between 90 and
30 degrees) represents the allowable breakout width. This
range is based on the current wellbore deviation at that
depth. Well A13 experienced most of the problems
including tight hole, stuck pipe, and pack-offs between the
depth interval from 9,000 to 11,000 ft TVD. This experience
is consistent with the output of the software and therefore
lends more confidence to the geomechanical model. The
result of the final model selected by the service company as
the appropriate one matched the operator s results with only
minimal differences in recommended mud weights, Fig. 6.
Once both parties had considered the model acceptable,
further discussion of the mud weight schedule for A9
occurred based on the modeling results. Fig. 6 illustrates
that the planned mud weights (brown) were lower at some
depths than both software packages predicted as being
required. This difference lead to some concerns, so further
analysis was performed to determine the amount of risk
using these mud weights based on the uncertainty in the
input parameters for the geomechanical model. A
Quantitative Risk Assessment (QRA) 8, 9 was performed
over several depth intervals. Uncertainties in the input
parameters were assigned, as shown in the distribution plot
Fig. 7.

SPE/IADC 92588 APPLICATION OF PREDICTIVE REAL-TIME WELLBORE STABILITY MONITORING ON A DEEPWATER ERD WELL

Monte Carlo simulations were run on 10,000 different


simulations to generate the probability of success plot
shown in
Fig. 8. The result was that there was a less than a 50%
chance of having no wellbore stability problems with the
current planned static mud weight.
A sensitivity analysis was also performed on this QRA,
with results shown in Fig. 9. The figure illustrates the
impact of the uncertainty in each input parameter. In this
case, as is the case in most Gulf of Mexico studies, the pore
pressure and rock strength have the highest impact on the
results.
Based on the results of this analysis (including the
differences between the pre-drill model and the real-time
updates), the operator was prepared to make changes to the
mud weight schedule during drilling.
Real-Time Monitoring
As discussed in previous sections, the real-time pore
pressure and geomechanics software was installed and the
network and database connections established within the
operations center. For the 10 5/8-in. x 17-in. and 9 5/8-in. x
14 -in. hole sections, two service company WBS
specialists operated the software and monitored the output.
The calculated wellbore collapse and fracture pressure mud
weights were displayed against downhole mud pressure
measurements including ECD and pumps-off minimum,
maximum, and average pressures. An example of the
calculated output is presented in Fig. 10. Mechanical
drilling parameters such as torque and drag, pick-up and
slack-off weights, and indications of overpull were
monitored by the 24/7 operations center monitoring staff.
10 5/8-in. x 17-in. Section. During the drilling of this
section, there were no obvious signs of wellbore instability;
the static mud weights and ECD pressures remained within
the calculated real-time model boundaries. Real-time pore
pressure predictions remained close to the predrill estimates.
The operator s standard practice dictates that red and
yellow ECD limits be established as safety margins below
the LOT/FIT measurement for the hole section so as to
avoid lost circulation events. Drilling ECD did increase
above the yellow limit at 7622 ft MD and above a revised
yellow limit at 8589 ft. In both instances, circulation of
cuttings out of the hole reduced the ECD to - or below - the
yellow limits. At 9074 ft MD, a sweep was pumped to assist
in hole cleaning; and the ECD climbed to 11.23 ppg EMW
with double the volume of cuttings at the shakers at bottoms
up. The increase in cuttings observed during the sweep and
subsequently during backreaming out of the hole was
attributed to cuttings bed build up in the 17-in. hole section
at 75o. It was not seen as an indication of borehole
instability.
9 5/8-in. x 14 -in. Section. This section again showed no
obvious signs of wellbore instability during the drilling of
the section. The lower-than-expected Shmin value of 12.0 ppg
obtained during the leak-off test (LOT) placed additional
constraints on the upper ECD limit. ECDs during drilling
rose above the yellow limit at 12286 ft and 12382 ft where
the hole was circulated for 15 minutes each time to reduce

ECD from 11.8 to 11.7 ppg. At 12544 ft, the string was
pulled to the casing shoe, without problems, in order to
replace a cable on mud pump #1. After running in hole and
drilling ahead, the ECD again rose above the yellow limit at
12764 ft and 13147 ft; and the hole was circulated to reduce
cuttings load.
At 14017 ft, connection gasses were observed, and the
mud weight was increased in response. The real-time pore
pressure predictions, based upon the sonic and density data,
were adjusted based on the connection gas information.
These predictions forecast a steadily increasing pore
pressure to TD of the section where the calculated pressures
from the overlays indicated an 11.4 ppg EMW pore pressure
Fig. 12. The calculated, borehole collapse pressures derived
from the real-time sonic data placed the collapse pressure
very close to the pore pressure, as shown in Fig. 10.
Specifically, the real-time pore pressures and UCS rock
strengths were regarded as too high, as there was no
evidence of instability during drilling.
Two factors influenced the pore pressure predictions.
The first was the real-time monopole sonic data which
produced a variation of 2 to 3 ppg EMW in the raw
calculations from the pore pressure overlays. The second
was adjusting the normal compaction trends so that the raw,
calculated pore pressures values were 0.2 ppg below the
mud weight over the interval of the connection gas. After
the memory data itself became available, the pore pressures
were adjusted down to 10.8 ppg at the TD of the section,
Fig. 13, placing the allowable breakout limit at 11.45
ppg. The pumps-off minimum pressures from the real-time
data were 11.41 to 11.45 ppg over the last 200 ft of the
interval, indicating the possibility of some wellbore breakout. However, pressures remained within manageable limits.
Real-Time Sonic Data. From the onset of the trial and even
prior to that in the well planning stage, it was recognized
that sonic logs in Ram Powell, like in many areas of the
GOM, provide the most accurate information about
formation pore pressure and mechanical properties. Hence,
it was planned to use primarily sonic real-time data for
updating the models. However, during drilling, two large
problems were identified, i.e., noise in the data and acoustic
anisotropy.
Effects of noisy sonic logs are illustrated in Fig. 10,
where updates of the upper and lower pressure boundaries
calculated from real-time sonic data are shown. Sonic is
shown in blue in the rightmost track. Due to low signal
levels, the downhole logic of MWD tools was often
inadequate and computed interval transient times were too
fast, 95 s/ft instead of the expected 120 s/ft for the
example in Fig. 10. Fast acoustic velocity was interpreted by
the software as a more competent rock (higher UCS). The
use of this interpretation reduces the calculated collapse
pressure, giving the impression that the borehole stability
mud weight is less. This can be seen in the leftmost track.
The pre-drill estimates of pore pressure and stability mud
weight are shown respectively in red and blue dashed lines.
As can be seen, approximately 1.2 lbs/gal margin above
pore pressure is required to keep the wellbore stable. The
values computed from real-time logs are shown in solid red
and blue lines. Due to seemingly hard rock (from sonic

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

values), almost no margin is required. For example, a 76o


inclination well can be drilled through shales with MW
equaling the pore pressure. This expectation is clearly
erroneous and is derived from incorrect input data.
These problems lead to a change of plans. The first was
to use trends from MWD density logs to update pore
pressure predictions. For mechanical properties, a
combination of the predrill model and real-time information
was most effective in determining the expected conditions.
The ability to be able to combine predrill inputs with the
conditions encountered during drilling provided the most
benefit. Specifically, the most benefit came from using the
predrill rock properties where the real-time input data
quality was poor, and combining these rock properties with
the current trajectory and pore pressure.
Sonic Anisotropy. In addition to poor quality monopole
sonic data, there was another, more fundamental obstacle
with using sonic data for mechanical properties predictions.
Acoustic logs from a deviated well drilled through a
laminated mudstone/siltstone sequence may suffer from
anisotropy effects, e.g., see references10. A quick
comparison of post-processed sonic data from A-9 at 76o
inclination with checkshot interval velocities (vertical)
acquired by the Seismic While Drilling tool show that LWD
sonic velocities are ~10% faster for the depth interval
chosen. Comparisons with offset wells, one deviated and
one vertical, confirm the magnitude of anisotropy effect for
the Ram Powell area. This comparison is shown in Fig. 11.
The magnitude of anisotropy effect needs to be known and
corrected for in order to use sonic data for pore pressure and
mechanical properties calculations.
9 -in. x 10 -in. Section. After the conclusion of the first
two sections, the operator agreed to the continuation of the
trial in the final 9 -in. x 10 -in. hole section. Only one
service company WBS expert operated and monitored the
software. This person also worked the opposite 12-hour
period to the asset petrophysicist. After drilling this section
to TD with no indications of instability during drilling,
severe borehole problems were encountered while pulling
out of the hole. These problems were eventually exacerbated
by mechanical failure of the 5-in. drill pipe. This failure, in
turn, led to a fishing operation which was successful.
Borehole Instability at Well TD. In terms of the wellbore
stability analysis, borehole instability at well TD was
probably the most critical event affecting the well. Since
there are some broader lessons that can be taken for this
event, it is described below in more detail.
Fig. 14, shows bit depth vs. time, with color coding and
numbers denoting sequence of events. In the next figure,
Fig. 15, recorded mode data from the downhole pressure
measurement tool is shown for the same operations.
1) Drilling to TD. Fig. 15, illustrates that, with a surface
mud weight of 11.8 ppg, the measured downhole static
densities are 12.15 ppg. As drilling slows down, around ~32
hrs, the downhole static densities drop to 12.05 ppg as
cuttings are being circulated out of the system. The
predicted mud weight required to control WBS was
calculated at 12.15 lbs/gal, with a critical instability point at

SPE/IADC 92588

approximately 11,000 ft TVD. No cavings generated by


wellbore collapse or drilling problems were observed during
this period.
2) Clean-out cycle prior to POOH. Wellbore collapse
was initially controlled by ECDs of 12.7 ppg. Halfway
through the cycle, the surface MW was increased to 12.0
ppg, corresponding to an equivalent downhole density of
12.3 ppg to provide for an extra margin when tripping out of
the hole. Again no cavings or angular cuttings were
observed across the shakers. No drilling problems were
observed.
3) Short trip 10 stands POOH and back to TD. No
problems reported. Swabs down to 11.7 ppg are evident on
recorded mode downhole pressure data, but were not
available in real time; mud telemetry was unavailable, as no
pumping was performed. During this short trip, the BHA
remained within the depth range where predicted stability
mud weight was 11.9-11.8 ppg.
4) Begin POOH from TD. Essentially no problems until
the BHA reached approximately 11,000 ft TVD, the point of
highest required stability mud weight. Large swabs down to
11.6 ppg are evident on recorded mode downhole pressure
data, but again were not available in real time.
5) Backreaming to get BHA past 10,700 ft TVD with no
success. Major overpull and excessive torque were
observed, accompanied by pack-offs; however, no losses
were observed. The decision was taken to trip in to TD,
clean the hole, and raise the surface MW to 12.2 ppg.
6) The second clean-out cycle at TD. Wellbore collapse
was controlled by ECDs of 12.8 ppg. A combination of lovis and hi-vis mud sweeps was pumped in an attempt to
fully clean the hole. During the sweeps, 0.5 inch angular
cuttings were observed coming from the well. The surface
MW was then raised to 12.2 ppg, 12.45 ppg equivalent
Downhole MW.
7) Begin POOH from TD with rotating and circulating.
The recorded MWD density caliper from this pass showed a
good hole with no major zones of breakout. However, only
data up to 11,200 ft TVD was available. No significant
problems were experienced until 5-in. drill pipe parted at a
connection 3500 ft above the BHA. This problem developed
during a relatively minor pack-off, leaving the drilling
assembly and drill pipe in the hole. Drilling staff then
commenced fishing operations, raising the MW to 12.4 ppg
at surface and successfully retrieving the fish. The hole was
smooth with no overpull.
After this post-well analysis, the conclusion drawn was
that borehole instability and not poor hole cleaning was
responsible for problems encountered. The start of the
problems occurred around 11,000 ft TVD where the updated
real-time geomechanical models predicted instability with
the measured mud weights. The second attempt to pull out
of the hole, after raising the mud weight (denoted #7 in Fig.
14), was going well until the mechanical failure of the drill
pipe.
There are two possible causes for the borehole instability
between 10,700-11,000 TVD. The first is the drop in
downhole equivalent static density during a reduction in
penetration rates, causing fewer suspended cuttings to be in
the mud system. The second is the swabbing caused by the
BHA whilst pulling out of the hole.

SPE/IADC 92588 APPLICATION OF PREDICTIVE REAL-TIME WELLBORE STABILITY MONITORING ON A DEEPWATER ERD WELL

Conclusions
The trial sucessfully met the three objectives set out at the
start of the project. In preparing and executing the trial, the
processes were developed that would allow the
incorporation of real-time stability modeling into the
operations center.
The real-time update of the geomechanical model
functioned without problems connected to the operations
center networking and database infrastructure.
Both pore pressure and geomechanical predictions were
verified from the drilling experience. The pore pressure
models correctly predicted the depth range of higher
pressures, and the real-time geomechanical model correctly
predicted the pressures required to keep the borehole stable
and the higher mud weights required. Results of the analysis
were used to enable proactive decision-making during the
drilling of the well.
To effectively deliver real-time geomechanical
monitoring, a greater preparation time is required than most
other well site services. Greater preparation time is required
in order to develop a well constrained geomechanical model
for the prospect.
Effective communication of information is also essential
to the success of such monitoring. Integrating all wellsite
measurements,
calculated
pore
pressures,
and
geomechanical results within a common onshore database
enables this communication. It allows the complete picture
to be presented and decisions to be made with all of the
relevant facts at hand and in a timely fashion.
The break in the 24-hour monitoring that caused the
reduction in ECD as the well reached TD (an event that was
initially unnoticed) could have contributed to the wellbore
stability issues encountered. The drop in cuttings load was
evident from real-time data but not reacted upon,
emphasizing the importance of constant monitoring of all
critical parameters. This experience has led to a
modification of the operations center 24/7 monitoring
communication protocols.
Raising the mud weight was initiated by the rig crew and
supported
per
the
wellbore
stability
model
recommendations. Higher mudweight was instrumental in
recovering the fish.
Actual downhole pressures were lower than the wellbore
collapse pressure in two instances - due to the drop in
cuttings load and due to swabs. Reduced downhhole
pressures led to borehole instability and trouble time. The
ability to obtain the swab pressures from the downhole
pressure tool when circulation resumed would have
provided critical information for the stability analysis.
Having a long 200 ft rigid BHA with two full gauge
stabilizers created significant swab pressures when pulling
out of the hole without circulating. These swabs were
created by the top concentric reamer and could have created
pack-offs at the lower stabilizer. The potential swab
pressures must be taken into account when planning mud
weights for pulling out of a hole.

Acknowledgements
The Ram Powell partnership (BP Americas, ExxonMobil,
and Shell E&P) and Halliburton Sperry-Sun are thanked for
permission to publish this paper. The Shell real-time
operations center staff and Larry Barfield s coffee provided
support throughout the project.
References

1.

Pellerin, N., Kaminski, D., Dunbar, D., Crumhorn, W.,


McColpin, G., Williams, J.: Improving Drilling Decision
Making Via Real-Time Operations Centers SPE/IADC
79893 presented at SPE/IADC Drilling Conference
Amsterdam 19-21 February 2003.
2. Algu, D., Landgrave, S., Esquinance, B., Volokitin, Y.,
Derise, B.: Extended-reach Drilling in the GOM-Ram
Powell Case Study SPE/IADC 92371 presented at
SPE/IADC Drilling Conference Amsterdam 23-25 February
2005.
3. Craig, P.A., Bourgeois, T.J., Malik, Z.A. and Stroud T.B.
(2001): Planning, evaluation and performance of horizontal
well in Ram Powell field, deepwater Gulf of Mexico in T.R
Carr et al., eds. AAPG Methods in Exploration, No. 14, p 95109
4. Lester, G.S, Lanier, G.H, Javanmardi, K., Bernardi, T. and
Halal, A.(1999): in.Ram/Powell Deepwater Tension-Leg
Platform: Horizontal Well Design and Operational
Experiencein., SPE 57069, Offshore Technical Conference,
Houston, 3-6 May 1999
5. Lpez, J.L., Rappold, P.M, Ugueto, G.A. Wieseneck, J.B. and
Cung, K. Vu (2004): Integrated Shared Earth Model: 3D
Pore-Pressure Prediction and Uncertainty Analysis in
Leading Edge , January 2004.
6. Eaton, B.A.: Graphical Method Predicts Geopressures
Worldwide World Oil (May 1976) 51-56.
7. S-W. Wong et al.: Borehole Stability in Shales SPE 24975,
1993.
8. Moos, D., P. Peska, T. Finkbeiner, and M.D. Zoback, 2003.
Comprehensive Wellbore Stability Analysis Utilizing
Quantitative Risk Assessment, Journal of Petroleum Science
and Engineering, Special Issue on Wellbore Stability, B. S.
Aadnoy and S. Ong, eds. 38 (2003) p. 97-109.
9
Ottesen, S., R.H. Zheng, and R.C. McCann, Wellbore
Stability Assessment Using Quantitative Risk Analysis,
SPE/IADC 52864, presented at the SPE/IADC Drilling
Conference in Amsterdam, Holland, 911 March, 1999.
10. Furre, A., and Brevik, I., 1998, Characterization of angle
dependency in sonic logs: 68th Annual Internat. Mtg., Soc.
Expl. Geophys., 68th Annual Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts.

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

SPE/IADC 92588

Figures

Fig. 1Map of Ram Powell area showing A-9 well and


calibration wells used in WBS modeling.

Fig. 3Operators offset pore pressures.

A-9 Overburden

-5000

20"

Po
Frac Grad

-6000

Frac Init ~ Sh_min

16"

A-9 Actual FIT/LOT

-7000

Stability MW
-8000

Actual Surface MW

13 5/8"

-9000

11 3/4"
-10000
-11000

G sand

7 5/8"
-12000
8

10

11
12
13
14
15
Equivalent Mud Weight (ppg)

16

17

Fig. 2Pore Pressure and mud weight plot for A-9 well. Circles
represent actual leak-off tests performed on A-9; darker circles
represent fracture initiation pressure. All data represents
downhole pressures converted to equivalent mud weight,
except for actual mud weights used in drilling which are surface
mud densities.

Fig. 4Service company offset pore pressures.

SPE/IADC 92588 APPLICATION OF PREDICTIVE REAL-TIME WELLBORE STABILITY MONITORING ON A DEEPWATER ERD WELL

Fig. 5 Predicted wellbore failure versus depth for A13.

Fig. 6 Geomechanical model.

10

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

Fig. 7 Uncertainty analysis.

Fig. 8 Probability of success.

Fig. 9 Sensitivity analysis.

SPE/IADC 92588

SPE/IADC 92588 APPLICATION OF PREDICTIVE REAL-TIME WELLBORE STABILITY MONITORING ON A DEEPWATER ERD WELL

Stability Mudweight
Pre-drill
Realtime updated
Pore Pressure
Pre-drill
Realtime updated
Equivalent MW, lbs/gal

Realtime GR

Realtime RHOB

Realtime Res

Realtime DT

18

80

us/ft

11

180

Allowable
Breakout width

ECD

10 feet TVD per


minor vertical grid

Dots represent
Pumps off PWD values

Hole Angle
Fig. 10 Real-time update of upper and lower pressure boundaries calculated from real-time sonic data. Sonic is shown in blue in the
rightmost track. Expected value for this depth is 120 s/ft. In the leftmost track, pre-drill estimates (dashed) and real-time updates (solid)
of pore pressure (red) and stability mud weight (blue).

VK 913 A-9, 76 deg

VK 913 A-9 CHECKSHOT

VK 912#3 vertical
VK 914#1, 51 deg

50

60

70

80 90 100 110 120 130 140 150


Interval transit time, us/ft

Fig. 11 Likely effects of anisotropy on shale acoustic velocities - comparison of data along deviated wellbores (A-9, VK 914#1) and
vertical (A-9 Checkshots, VK 912#3).

12

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

SPE/IADC 92588

Depths 250 ft
TVD per
Divison

11.4 ppg
PP EMW
Fig. 12 Real-time pore pressure prediction 9 5/8-in. x 14 -in. section Pore Pressures calibrated to connection gas. Track 1 density
measurements with NCT; Track 2 resistivity measurements with NCT; Track 3 Sonic measurements with NCT; Track 4 raw
calculated pore pressures and interpreted pore pressure (red).

SPE/IADC 92588 APPLICATION OF PREDICTIVE REAL-TIME WELLBORE STABILITY MONITORING ON A DEEPWATER ERD WELL

13

Depths 250 ft
TVD per
Divison

10.8 ppg
PP EMW

Fig. 13 Revised PP Prediction 9 5/8-in. x 14 -in. section Memory Data Track 1 density measurements real-time and recorded,
Track 2 Resistivity measurements real-time and recorded, Track 3 Sonic measurements real-time and recorded, Track 4 Raw
calculated pore pressures real-time and recorded and interpreted pore pressure (Red)

14

J.A.GREENWOOD; A.BREHM, E VAN OORT; D.R.ALGU; Y.E VOLOKITIN

24

48

Time, hrs
72

96

SPE/IADC 92588

120

10500
5

10700
WBS MW

Bit TVD, feet

10900

11100

11300

7
3

11500

11700
11

12

13

14

15

16

Fig. 14 Depth vs. time plot describing sequence of events and borehole instability at well TD. For explanations, see text.

ECD, lbs/gal

14.3
14.1
5
13.9
13.7
13.5 LOP @11 3/4" shoe
13.3
13.1
1
12.9
2
12.7
12.5
12.3
12.1
WBS
11.9MW
11.7 Surface MW
3
11.5
4
11.3
0
24
48
72
Time, hrs

96

120

Fig. 15 Plot of ECDs measured by the downhole pressure tool for the same time frame as in Fig. 14.

Potrebbero piacerti anche