You are on page 1of 0

20 Oileld Review

Improving Well Placement with


Modeling While Drilling
Daniel Bourgeois
Ian Tribe
Aberdeen, Scotland
Rod Christensen
Oilexco North Sea Limited
Calgary, Alberta, Canada
Peter Durbin
Ikon Science Limited
Teddington, England
Sujit Kumar
Bogot, Colombia
Grant Skinner
Stavanger, Norway
Drew Wharton
Houston, Texas, USA
For help in preparation of this article, thanks to Adrian Kemp,
Houston.
Drilling Office, ECLIPSE, GeoFrame, InterACT, Osprey,
PERFORM, PeriScope, PeriScope 15 and Petrel are marks
of Schlumberger.
Windows is a registered trademark of Microsoft Corporation.
Increased computing power, the growing capabilities of modeling and simulation
software, and human ingenuity across multiple disciplines are ushering in a new era
in reservoir management. The ability to update reservoir models in real time will lead
to exciting advances in wellbore placement, helping engineers and geoscientists
improve field development.
Sophisticated new LWD tools that help define the
reservoir are being combined with fast reservoir-
modeling software to optimize wellbore place -
ment while drilling. This addition drama t ically
augments the traditional uses of reservoir-
modeling and simulation tools, such as assessing
reservoir performance, forecasting production
and estimating reserves. Now, this combination
helps improve hydrocarbon recovery by showing
drillers where to drill more productive wells.
Furthermore, data acquired while drilling can be
added to the model to provide rapid updates.
Through the years, the E&P industry has
experienced the benefits of establishing a holistic
view of the reservoir. This view is reflected in
modern reservoir modeling and simulation soft -
ware. One of the fundamental roles of these
software tools is to simplify the complex issues
regarding scale, data and uncertainty.
Post-stack seismic data used in modeling
define the interwell reservoir volume and charac -
teristics and represent a static snapshot of the
reservoir. Wellbore data from drilling and well-
logging operations provide detailed near-wellbore
information that can be interpolated away from
the borehole and across the reservoir volume.
Time-lapse, or 4D, seismic volumes are now
used to monitor reservoir changes through time,
examining reservoir dynamics. This often involves
the mapping of seismic attributes derived from
amplitude, phase and frequency content to high -
light changes in the reservoir from one survey to
the next.
Models and simulators help in assessing
and predicting reservoir performance and in
identifying production problems. Although the
terms model and simulation are often used
interchangeably, in the E&P business, there are
important differences between them. Models, or
conceptual models, attempt to represent actual
systems and are largely static, but can be
updated with new information. Simulators, or
simulation models, attempt to describe how a
system changes over time. Despite their
differences, both reservoir models and fluid-flow
simulators help engineers and geoscientists
develop successful drilling plans, make
completion choices, determine workover plans
and formulate secondary-recovery strategies.
The success of these applications relies on the
accuracy of the reservoir models.
In the last decade, drilling capabilities, along
with MWD and LWD technological advances,
have largely outpaced the industrys ability to
manage and rapidly exploit real-time data in
modeling and simulation. Breakthroughs in
drilling include accurate bit placement using a
variety of new technologies, such as rotary
steerable systems coupled with advanced LWD
systems.
1
Extended-reach, multilateral and
geosteering technologies have increased the
ability to contact more of the reservoir with
complex wellbores. Tremendous volumes of high-
quality data are now acquired with modern MWD
and LWD tools. Data can be transmitted to
surface and immediately sent to centers of
expertise for real-time interpretation. In many
cases, borehole placement could be further
optimized if the new information could be
quickly integrated into reservoir models while
drilling is still taking place.
1. Chou L, Li Q, Darquin A, Denichou JM, Griffiths R, Hart N,
McInally A, Templeton G, Omeragic D, Tribe I, Watson K
and Wiig M: Steering Toward Enhanced Production,
Oilfield Review 17, no. 3 (Autumn 2005): 5463.
2. Bryant I, Malinverno A, Prange M, Gonfalini M, Moffat J,
Swager D, Theys P and Verga F: Understanding
Uncertainty, Oilfield Review 14, no. 3 (Autumn 2002):
215.
3. Bratton T, Cahn DV, Que NV, Duc NV, Gillespie P, Hunt D,
Li B, Marcinew R, Ray S, Montaron B, Nelson R,
Schoderbek D and Sonneland L: The Nature of
Naturally Fractured Reservoirs, Oilfield Review 18,
no. 2 (Summer 2006): 423.
4. Ali AHA, Brown T, Delgado R, Lee D, Plumb D, Smirnov N,
Marsden R, Prado-Velarde E, Ramsey L, Spooner D,
Stone T and Stouffer T: Watching Rocks Change
Mechanical Earth Modeling, Oilfield Review 15, no. 2
(Summer 2003): 2239.
Winter 2006/2007 21
Crucial to developing this application are
new software tools that enable a multidiscip linary
approach to model building and updating,
allow faster simulation using updated models
and help asset teams evaluate risk as models
and proposed well designs change with
new information.
This article investigates advances in reservoir
modeling and simulation and their potential to
improve wellbore placement. The contrasting
roles of modeling in the past, present and future
are briefly discussed, along with the visionary
concept of simulation while drilling (SiWD). We
demonstrate how rapid model updating has
already helped operators place their wells more
successfully in the North Sea. Next, we describe
a recent test of SiWD capabilities and the
improvements that are required for further
advancement. Finally, we examine the potential
applications of real-time modeling and simulation.
Moving Modeling Forward
One of the many challenges in developing a
model is to strike a balance between the risk of
high uncertainty and the cost and time needed to
improve accuracy. When creating and main -
taining an optimized model, reservoir engineers
must consider data quality, quantity and
uncertainty. The timing
and frequency of incorpo-
rating new data into a model
have an impact on the uses of models
and simulations such as forecasting
production, estimating reserves or planning
the fields development. For example, when
incorporating critical near-wellbore LWD data,
updates would need to be frequent to benefit the
drilling of the subject well. While the model is
being created and updated, uncertainty needs to
be assessed.
2
Reservoir modeling occurs at many levels;
there are models within models. Geologic models
focus mainly on geologic-layer thickness,
depth and extent, but also include faultsa
source of reservoir discontinuity and compart -
mentalization. Seismic and borehole data often
provide the bulk of information from which to
build and update a geologic model, including
formation boundaries or layers. With data from
well logs and cores, petrophysical models
describe forma-
tion lithologies and
reservoir properties,
such as porosity, perme-
ability and fluid content.
This same information gives
geoscientists an appreciation
of the variability within the
reservoir. As reservoir complexity
increases, for example, in naturally fractured
or heterogeneous reservoirs, the relationship
between porosity and permeability systems
becomes more difficult to model.
3
As the industry moves to drilling in more
challenging environments, mechanical earth
models (MEMs) become vitalmost notably to
avoid subsurface drilling hazards.
4
In addition,
PVT models are used to portray fluid properties
across a range of phase-changing reservoir
conditions. These models require input pri-
marily from laboratory measurements, so rapidly
updating this information in reservoir models
may prove difficult.
Also, rigorous fluid-flow simulation helps
describe complex multiphase fluid-flow phenom -
ena in the reservoir. Production simulators also
consider flow behavior outside the reservoir, such
as phase slipping in the wellbore.
Reservoir models and simulators have
contributed to the oil and gas industrys improved
understanding and success in increasingly
complex reservoirs. Nonetheless, building, main -
taining and updating models are time-consuming
processes, which may involve numerous person -
nel across several disciplines. Recent changes
in modeling methods and tools have made it
possible to update models while drilling to
influence drilling operations.
Contrasting Approaches
The established roles of modeling and simulation
are to predict reservoir performance, forecast
production and estimate reserves. Modeling and
simulation are also commonly performed to
determine completion and workover effec tive -
ness and diagnose productivity problems by
comparing actual production to predicted
production, especially in horizontal wells.
Moreover, fluid-flow simulation is crucial for
developing infill drilling plans and formulating
secondary-recovery strategies. While these
important tasks do not necessarily require rapid
decision making, accuracy is paramount to
reduce uncertainty. One way to reduce
uncertainty while drilling is to incorporate the
most recent information as quickly as possible.
To exploit new information while a well is being
drilled, improvements were needed in several
areas, including modeling and simulation
software and hardware, and MWD- and LWD-data
acquisition and delivery.
In the past, computer processor speed had
limited the ability to quickly and frequently
update models and run simulationsespecially
full-field simulation that often takes weeks of
computer and personnel time. Other factors have
inhibited model building and updating.
Throughout much of the 1980s and 1990s, the use
of MWD and LWD data in modeling was
inefficient, primarily because acquisition
technologies and modeling and simulation
software were not properly integrated. In
addition, the process often was not automated
and required human expertise.
Most models were built in discipline silos.
Some disciplines placed a higher priority on
modeling because they used models more
regularly and benefited more often from their
use. Drillers used models less frequently, and
as a result, their models were not optimized
to solve issues related to drilling. This has
changed; a general lack of cross-disciplinary
integration has given way to the multi -
disciplinary asset-team approach, immersive
visualization of the reservoir and real-time data
delivery (above left).
Model Well Paths
The Brenda field in the North Sea produces oil
from a system of channelized turbidite sandstones
(left). Individually, its reservoir sands are
frequently too thin to be explicitly resolved by
seismic methods, complicating exploitation
efforts. Two 3D seismic datasets and information
from 13 wells were available for field appraisal;
these were used to generate a reservoir model.
Modeling fluid flow in the reservoir using
ECLIPSE reservoir simulation software suggested
a four-well development program would be
needed to optimize reserve recovery. To locate oil-
22 Oileld Review
>
Immersive visualization. Early visualization technologies were primarily used to interpret 3D seismic
volumes. Today, the emphasis is on collaboration across multiple disciplines for visualization in well,
reservoir and field management, including well placement.
>
Location of the Brenda field in the North Sea. Potential drilling targets in
the Brenda field were identified using an advanced seismic preprocessing
technique, a high-resolution velocity model, prestack seismic imaging and
elastic-impedance analysis. Three wells have been completed in the Upper
Balmoral sandstones, and production-test results have been encouraging.
A fourth well drilled into the reservoir has been cased and is awaiting the
drilling of the horizontal leg.
km
miles 0
0 100
100
Aberdeen
Edinburgh
Bergen
Stavanger
NORWAY
UK
Brenda
Winter 2006/2007 23
bearing sands within the field, the operator,
Oilexco, and Ikon Science used 3D seismic
processing techniques with prestack data,
including specialized elastic impedance compu-
tations and amplitude variation with offset
(AVO) analysis.
During 2006, Oilexco drilled three production
wells and started a fourth well in the Brenda
field. This four-well project targeted three
sandstones in the Paleocene Upper Balmoral
member of the Montrose group. In the first three
wells, the true vertical depths ranged from 6,000
to 6,500 ft [1,829 to 1,981 m], whereas the total
measured depths reached 13,700 ft [4,176 m].
Total reservoir thickness in these wells has
ranged from 40 to 60 ft [12 to 18 m]. The top sand,
the UB3, is usually of good quality but thin, and
presents a difficult target to hit and stay within
while drilling. The lower sand, the UB1, is also of
good quality but is sometimes below the oil/water
contact. The middle unit, the UB2, is thicker and
more shale-prone, and is not a primary reservoir
target everywhere in the field.
Seismic data were used to define the optimal
landing point from which to start the horizontal
portion of the wellbore. Given the challenging
reservoir target, the relatively low resolution
from seismic data, local dip variations of the
reservoir top and the long horizontal wells used
to exploit the reservoir, depth accuracy while
drilling was a major concern. More specifically,
depth errors between the bit and the model had
to be resolved prior to landing the borehole near
the proposed target. The targets were in areas
defined by seismic imaging as exhibiting low
elastic impedance, a good indicator of hydro -
carbon accumulations in the Brenda field.
To resolve depth errors, an Oilexco operations
geologist would establish the actual drilling
depths of two markersthe tops of the Sele and
Lista formationslying just above the Upper
Balmoral sands, then compare the drilling depths
to seismically determined depths and adjust the
drilling path accordingly. Oilexco required that
wells hit the reservoir while the wellbore
was nearly horizontal89 deviationto ensure
optimal location of the well path within the
reservoir. Immediately beneath the top of the
reservoir, the 12
1
2-in. casing was set. A successful
landing was paramount for the ultimate drilling
objectiveto maximize contact with the most
promising reservoir. The horizontal leg was
drilled using an 8
1
2-in. bit.
As the wells were drilled, real-time borehole-
survey data transmitted uphole were delivered to
Oilexco in Aberdeen, using the InterACT real-
time monitoring and data delivery system, and
then sent electronically to Ikon Science in
London. These new data were incorporated into
the Petrel seismic-to-simulation software model
so that the current bit location could be
displayed with respect to the desired targets in
the model. Using these displays, the operations
personnel in the Aberdeen Oilexco office could
send proposed well-path changes back to the rig
to optimize the landing of the well into the
reservoir in preparation for drilling the horizontal
portion of the well (below).
>
Petrel well planning and visualization of Brenda field D3 well. The top map, used by the operations geologist for landing the well, shows the reservoir structure
contours in black and white. Areas of low elastic impedance are shaded in light blue. Existing wells are blue and show the top of the reservoir as orange dots.
The proposed D3 horizontal well path is shown as light green, and the actual D3 horizontal well path is in red. A Universal Transverse Mercator (UTM) grid on
the map allows direct manual plotting of well coordinates. The bottom image shows the proposed D3 horizontal well path (light green) in 3D, together with
existing wells (blue and red) that show formation tops. Pink spheres indicate the top of the Balder formation, light blue spheres show the top of the Sele
formation, yellow spheres mark the top of the Lista formation, and orange spheres identify the top of the reservoir. The orange target boxes define the landing
point and the XY limits of the horizontal path, and the white outlines surround areas of low elastic impedance that indicate a high probability of commercial
hydrocarbons. These areas are draped onto the contoured 3D surface of the reservoir top. The arrow in the bottom right corner shows the north direction.
Once the wellbores were successfully landed
in the top of the reservoir, Oilexco needed a more
precise way to evaluate the reservoir sandstones
and to locate the nonproductive shale immedi -
ately around the wellbore. To accomplish this,
Oilexco used the Schlumberger PeriScope 15
directional, deep imaging while drilling tool
(above). The PeriScope 15 tool is a deep-reading
electro magnetic resistivity device that
determines the direction and distance to bed
boundaries by showing conductivity contrasts.
With a transmitter-receiver spacing of 96 in.
[244 cm], the tool has the theoretical capability
to detect boundaries up to 15 ft [4.6 m] from the
borehole. However, the actual resolved distance
depends on the resistivity of the surrounding
and adjacent beds and the complexity of the
geologic layering.
The PeriScope tool acquired data from
around the borehole and successfully identified
the reservoir ceiling and the presence of zones
of lower quality within the reservoir, helping
Oilexco to fine-tune the drilling of the horizontal
wellbore. Adjustments were then made in
steering the wellbore to maximize wellbore
length within high-quality reservoir while
maintaining the largest possible standoff
distance above the oil/water contact at the base
of the reservoir.
Azimuthal polar plots generated from the
PeriScope tool results show the bit position with
respect to nearby bed boundaries, allowing the
24 Oileld Review
>
PeriScope curtain section for the Brenda field D1 well. This curtain-section display of Brenda Well D1 was used by the geosteering team to optimize
the placement of the 1,800-ft [549-m] horizontal well leg (red, from left to right) in complex geology. It enabled them to stay predominantly within 10 ft
[3 m] below the top of the Upper Balmoral reservoir. The lighter colors represent higher resistivity sandstone, and darker colors indicate lower resistivity
shale. Without the PeriScope information, much of the proposed well path (blue-green) would have strayed into the shale, making the last half of the
horizontal leg nonproductive. The image also highlights a low-resistivity zone from 10,750 to 11,050 ft MD. Missing the reservoir in this section allowed
drillers to optimize pay-zone contact in the high-resistivity zones.
1
1.27
1.61
2.04
2.58
3.27
4.15
5.25
6.66
8.44
20.70
13.56
17.19
21.79
27.61
35
T
r
u
e

r
e
s
i
s
t
i
v
i
t
y
,

o
h
m
.
m
1
0
4
0
0
1
0
3
0
0
1
0
5
0
0
1
0
2
0
0
1
0
1
0
0
1
0
6
0
0
1
0
7
0
0
1
0
8
0
0
1
0
9
0
0
1
1
0
0
0
1
1
1
0
0
1
1
2
0
0
1
1
3
0
0
1
1
4
0
0
1
1
5
0
0
1
1
6
0
0
1
1
7
0
0
1
1
7
6,740
6,760
6,780
6,800
6,820
6,840
1,800 1,600
T
r
u
e

v
e
r
t
i
c
a
l

d
e
p
t
h

s
u
b
s
e
a
,

f
t
6,820
6,840
6,860
6,880
6,900
6,920
6,940
1,400 1,200 1,000 800 600 400 200 0
T
r
u
e

v
e
r
t
i
c
a
l

d
e
p
t
h
,

f
t
True horizontal length, ft
>
Creating a mapable surface from PeriScope results. It is possible to
convert the PeriScope results that define the top of reservoir into a mapable
surface using the distances from the borehole to the bed boundaries
calculated from the PeriScope tool and borehole-survey data. In this
example, the data are used to create surface sticks (pink) representing the
boundary identified by PeriScope readings. From this a surface is created
(red with black contour lines). The surface is not shown in areas where it
dips below the wellbore.
10000
11000
11500
10500
Winter 2006/2007 25
geosteering team to make real-time trajectory
adjustments to optimize well placement.
5
The
bed boundaries defined by the PeriScope images
could then be converted to Petrel surfaces and
mapped within the reservoir model (previous
page, bottom).
The use of Petrel software to model the
Brenda field facilitated rapid reservoir
mappingusing geologic and geophysical
dataand well-path design in one software
package running on a standard laptop PC. Petrel
software was instrumental in the Brenda field
drilling workflow because it gave everyone
involved the ability to observe the wellbore
position with respect to the reservoir prior to and
after landing, enabling efficient well-path design
changes, avoiding plugbacks and sidetracks, and
maximizing productivity.
Oilexco has completed three Brenda
production wells, which are currently being tied
into the Brenda manifold. The completion flow
tests for the first three wells exceeded Oilexco
expectations. Their sandface productivity indexes
and normalized flow calculations suggest a
theoretical combined production rate of
44,000 bbl/d [6,995 m
3
/d] of oil.
6
First oil is
anticipated in late 2006 or early 2007.
Operating companies around the world are
increasingly using Petrel software to visualize
the reservoir, make interpretations, evaluate risk
and rapidly update the model while drilling,
allowing them to optimize bit placement and
produce more hydrocarbons. Modeling-while-
drilling workflows have also been successful in
fields offshore Vietnam, India and Malaysia.
Initially, visualization referred to seismic
volume interpretation using high-power
computers and was not connected to reservoir
models. However, to fully understand reservoirs
and fields, a wide variety of data must be
analyzed and multiple disciplines must be
involved. With Petrel software, a more
comprehensive workflow is now possible using
low-end Windows PCs at the desktop and in
collaborative multidisciplinary environments.
This allows asset teams to visualize, evaluate and
assess complex relationships in 3D, and through
time to better understand risk and uncertainty in
multiple scenarios and more accurately predict
production behavior.
Modeling Ahead
Specialized tools within Petrel software are
tailored for modeling-while-drilling applications.
For example, the Petrel Process Manager
facilitates fast data loading and model updating
by establishing an automated workflow. This
reduces decision-making and cycle time, and
saves time and money. Well paths can be designed
and updated using the Petrel Well Design tool,
increasing drilling efficiency and bit-placement
accuracy. The integrated workflow also can model
log responses ahead of the bit along the proposed
well path. Generating modeled petrophysical
responses ahead of the bit helps asset teams
understand the reservoir more fully and lets them
choose the optimal well path in 3D, reducing
uncertainty in complex settings.
While some of these capabilities are here
today in limited use, more widespread usage may
be imminent. Many advances have enabled the
move to modeling while drilling. For example,
supervisory control and data acquisition
(SCADA) systems, which have been in place for
many years, allow immediate access to downhole
data and control of downhole hardware. In
addition, the new generation of reservoir simu -
lators, which exploit faster, more sophisticated
processors, has increased the computational
power available to asset teams. Reservoir models
are now truly multidisciplinary tools that evolve
as new reservoir or field information is acquired,
such as new 3D seismic volumes, well logs, core
data, well-test data or production-history
information. Petrel softwares unique structure
and functionality, coupled with its PC
compatibility, facilitate integrated workflows in
geology, geophysics, well engi neering and
reservoir engineering (above).
Most reservoir models incorporate porosity and
permeability only within reservoir sections and
ignore the effects of overburden. MEMs contain
stress, mechanical rock-property and pore-
pressure predictions from the reservoir to the
surface. Consequently, workflows frequently break
down when the MEM is inadequate or nonexistent.
Knowledge of overburden geomechanics greatly
improves the well-construction process because, in
part, it allows asset teams to assess the risks along
a proposed well path and avoid hazards.
>
From seismic data to simulation. In this example, Petrel software has been
used to visualize production data and perform history-matching, improving
simulation and field development. The upper left image shows the reservoir
simulation model with local-grid refinement around the boreholes. The
upper right shows a horizontal well through the reservoir model with the
seismic volume displayed in the background. The lower left shows the
possible production-profile output from the ECLIPSE reservoir simulator.
Each curve represents a different model realization created for an
uncertainty study. On the lower right is a cumulative production plot at each
borehole. The bubble size represents relative production volume. The Petrel
workflow not only allows experts from multiple domains to meld their
domain-specific information and knowledge into a single model-centric
representation, but it also supports the ability to easily update and visualize
the collective understanding as soon as new information is available. It is
now possible to visualize, evaluate and assess complex relationships in 3D
space and time to better understand risk and uncertainty in multiple
scenarios and to more accurately predict reservoir flow behavior.
5. Chou et al, reference 1.
6. http://www.oilexco.com/news/060622.pdf (accessed
September 29, 2006).
In the past, exploiting the information in
mechanical earth and reservoir models, including
uncertainties, for practical drilling applications
has not been straightforward. However, in 2000,
Schlumberger Cambridge Research in England,
as part of the MoBPTeCh industry consortium
comprising Mobil Oil, BP Amoco, Texaco and
Chevron, completed the development of a
drilling-simulator prototype. This software lays
the groundwork for more automated risk
assessments in difficult drilling conditions.
7
Today, the Schlumberger Osprey drilling risk
prediction software and a Petrel plug-in enable
critical risk assessment and drilling-cost and
drilling-time estimates, in addition to providing a
collaborative link between drillers, geophysicists
and geologists.
8
Following an efficient workflow,
Osprey and Petrel tools allow asset teams to
interactively design well trajectories and to
update well-design plans as the model or
proposed well path is changed (left). Another
advantage of this software is that drilling
engineers can customize the system to incorporate
regulatory and company requirements, as well as
local and historical experience.
The industry is now considering the
possibility of simulating reservoir response to
new wells while drilling them. Along with the
integration of real-time data into models and
rapid model updating, the E&P industry is also
benefiting from faster simulators. This is
especially important when simulating complex
fluid-flow and production behavior in large
reservoirs, because these require large reservoir
models. The need for dynamic evaluation while
drilling intensifies as complexity increases. For
example, simulating while drilling (SiWD) in
three-phase, heterogeneous reservoirs already
affected by nearby producing wells would be
more beneficial than when drilling homogeneous,
single- or two-phase reservoirs with zero dipa
case where using field experience might suffice.
The idea of reservoir simulation while
drilling, or dynamic evaluation, is not new. One
such effort began in 1997 as part of a near-
wellbore modeling project by BP, Schlumberger
GeoQuest, Norsk Hydro and Saudi Aramco.
9
This
early project determined that real-time
optimization of a well path is a true multi -
disciplinary exercise, requiring asset teams to
have a clear understanding of the common goal
and to be prepared for changing scenarios.
Another finding was that real-time model
updating should focus on the near-wellbore
volume, where MWD and LWD data are most
pertinent. Permeability distribution along the
wellbore is critical to the predicted well
26 Oileld Review
>
Saving time and money while reducing risk. Osprey Risk interactive software was
designed to assist asset teams with well planning by providing probabilistic cost, time
and risk assessment while incorporating geological and geomechanical models into
the process. With subsurface targets identified, the well trajectory is designed in the
Petrel Well Design tool (top) and input into Osprey software as a deviation survey.
Earth-model dataat a minimum, pore pressures, fracture gradients and unconfined
compressive strengths from the Petrel MEMare also required input. The software
proposes optimal hole size, bit type, maximum mud weights, and casing sizes, weights
and set depths, taking into account production requirements, wellbore stability and
many other factors. Using the available data and Monte Carlo simulation, drilling times
and costs are calculated at key depths for a set of defined probabilities (bottom). This
output can be used as a working operational plan. Significant drilling risks for the
technical design are generated and can be displayed individually or grouped into
categories of fluid gains, mud losses, stuck pipe and mechanical problems. A total risk
index is also computed and can be used to rank scenarios. In the risk display, dark
green indicates low risk, light green shows low to medium risk, yellow means medium
risk, orange indicates medium to high risk and red shows high risk. Osprey Risk and its
plug-in are fast-responding applications so that drilling engineers and geoscientists
can easily change the design and compare results within minutes. This process leads
to reduction of technical risk and highlights where mitigation strategies will be needed
to implement the operational plan.
Winter 2006/2007 27
performance. However, accurately capturing
near-wellbore flow phenomena requires a
smaller grid size and local-grid refinement,
which decreases the processing time step while
increasing processing time. Also, full-field
simulation while drilling was deemed unrealistic
in most cases because of the imposed time
constraints during drilling operations. As part of
this work, software was developed that generated
a reduced near-wellbore model within the full-
field model and served as an early simulation-
while-drilling prototype tool.
Well Performance to Define Well Placement
In 2006, simulation while drilling (SiWD) was
defined as a real-time optimization process to
dynamically improve the design of the trajectory
and the configuration and completion strategy of a
well while it is being drilled.
10
This concept has
become more relevant today with the emergence
of innovative drilling, MWD and LWD tech nologies
that have enabled geosteering and advanced wells
with elaborate trajectories, multiple branches or
both. However, one of the key drawbacks in
drilling these advanced wells was the level of
uncertainty inherent in the initial reservoir
description, including which fluids are present.
This uncertainty accentuated the need for real-
time data collection, integration and interpretation.
SiWD has not been adopted for several
reasons. The industry has only recently realized
the advantages of real-time MWD and LWD data in
well construction, reservoir and simulation
engineering.
11
In addition, the seamless integra -
tion of while-drilling measurements into model ing
and simulation software tools has been difficult.
Until recently, an integrated approach has been
hampered by the lack of a proper platform from
which multiple disciplines could work. Moreover,
there has been a perception that model updating
in the appropriate time frame was not possible.
Finally, to be feasible, complicated workflows
needed for real-time evaluation of multiple well
trajectories and configurations will require
automated optimization.
Today, geosteering involves the interactive
placement of the borehole based on geology and
the desire to contact as much of the reservoir as
possible, with the goal of optimizing the initial
well productivity. While this technique has been
successful, complicated scenarios require a more
rigorous approach to effectively reduce risk. With
todays computing power and the increased
ability to model critical factors while drilling,
experts are envisioning the possibility of
simulating well productivity ahead of the bit.
Critical factors that impact short- and long-term
productivity include well-completion options,
multiphase-flow phenom ena in the reservoir and
in the well, the effects of drawdown at the
wellbore, and pressure and fluid changes in the
reservoir from neighboring production or
injection wells.
However, large, multilayer models make it
difficult, if not impossible, to update, upscale
and perform full-field simulations in time to
impact simultaneous drilling operations. This
problem is tempered by the fact that it is not
essential to examine all regions of the reservoir
equally when evaluating the future performance
of a single well. For example, changes in
reservoir pressure or hydrocarbon saturation at
great distances or in isolated layers might have a
minimal effect on the subject well. There may
also be minimal effects when the near-wellbore
permeability distribution domi nates the analysis.
Here, semianalytical methods can provide fast
and accurate results when modeling uncon -
ventional wells, but are less rigorous
when dealing with multiphase-fluid flow and
reservoir heterogeneity.
12
A closed-loop process was tested to design,
optimize and configure advanced wells in
real time (above). To prove this concept,
Schlumberger reservoir and software experts,
along with Spectrum Consultores, started with a
model based on data from a North Sea field. With
>
A closed-loop process for simulation while drilling. Defined by the
frequency of measurements and the speed of optimization, the cyclical
process includes data acquisition and interpretation, model updates,
parameter changes and simulation until an optimal solution is determined;
in the final step, appropriate action is taken.
Action
Propose
Simulate
Update Interpret
7. Booth J, Bradford IDR, Cook JM, Dowell JD, Ritchie G
and Tuddenham I: Meeting Future Drilling Planning and
Decision Support Requirements: A New Drilling
Simulator, paper SPE/IADC 67816, presented at the
SPE/IADC Drilling Conference, Amsterdam, February 27
March 1, 2001.
8. Givens K, Luppens C, Menon S, Ritchie G and
Veeningen D: Geomechanics-Based Automatic
Well-Planning Software Provides Drilling Decision
Support to Asset Teams, paper SPE 90329, presented
at the SPE Annual Technical Conference and Exhibition,
Houston, September 2629, 2004.
9. Be , Flynn J and Reiso E: On Near Wellbore Modeling
and Real Time Reservoir Management, paper
SPE 66369, presented at the SPE Reservoir Simulation
Symposium, Houston, February 1114, 2001.
Be , Cox J and Reiso E: On Real Time Reservoir
Management and Simulation While Drilling, paper
SPE 65149, presented at the SPE European Petroleum
Conference, Paris, October 2425, 2000.
10. Primera A, Perez-Damas C, Kumar S and Rodriguez JE:
Simulation While Drilling: Utopia or Reality? paper
SPE 99945, presented at the SPE Intelligent Energy
Conference and Exhibition, Amsterdam, April 1113, 2006.
11. Aldred W, Belaskie J, Isangulov R, Crockett B,
Edmondson B, Florence F and Srinivasan S:
Changing the Way We Drill, Oilfield Review 17, no. 1
(Spring 2005): 4249.
12. Wolfsteiner C: Modeling and Upscaling of
Nonconventional Wells in Heterogeneous Reservoirs,
PhD Thesis, Stanford University, California, 2002.
the flux-boundary condition technique and local-
grid refinement, the original 600,000-cell model
was reduced to 30,000 cells, thereby simplifying
the model (above).
With a reduced-grid, near-wellbore model,
several different well-path options were simu -
lated over a six-year production period and
compared based on three predicted outputs:
water cut, oil-production rate and gas/oil ratio
(GOR) (above right). Next, using the chosen
optimal well path, the reduced-grid simulation
was tested using a single-processor computer
against the full-field simu lation. Although the
reduced-grid simulation yielded a slightly higher
water-cut prediction over time, the predicted
cumulative oil production and GOR were
comparable (next page, top).
The study proved that SiWD is feasible at
typical North Sea rates of penetrationabout
200 ft/h [61 m/h], depending on MWD and LWD
operational requirements and BHA and bit
configurations. In this study using a North Sea
reservoir model, times for the various workflow
steps were determined to be acceptable for a
typical 10-day horizontal drilling operation in the
field. However, time estimates vary because many
of these steps are dependent on model complex -
ity and size and hardware and software
availability. Data acquisition and transmission
were assumed to occur in real time. The steps
included analysis and interpretation of the new
information; updating the model; gridding
optimization involving perpendicular-bisector
grid processing and local-grid refinement; the
initial new well proposal; and simulation runs
using the near-wellbore model. In this example
in a typical established field, the total estimated
turnaround time was 20 to 30 minutes.
Keeping It Real Time
From an engineering perspective, simplifying
models to enable modeling and simulation while
drilling is not necessarily the answer. However,
simplifying the workflow is always a positive step.
Software tools continue to become faster and
easier to use, connectivity to remote locations is
increasingly reliable and larger data volumes are
being transmitted at higher rates from downhole
tools to end-users as technologies improve.
28 Oileld Review
>
Making simulation faster. Grid-coarsening, local-grid refinement and
boundary-conditioning techniques are used to decrease simulation time
while preserving sufficient resolution of reservoir heterogeneities and
allowing multiple geostatistical realizations. In this North Sea example, the
simulation model of a channelized reservoir (top) is optimized by upscaling
each cell to a given resolution, which is defined by the local geologic
heterogeneities, the degree of fluid-flow activity and the distance from the
subject well (bottom).
>
Evaluating alternative well paths. For three
proposed well trajectories, three inflow-
performance predictions were used to determine
the optimized well path: water cut (top), oil-
production rate (middle) and GOR (bottom). In
this example, the simulation of Trajectory 1-2
(blue) terminates early because the higher water
cut has exceeded the assumed surface water-
handling capabilities. Trajectory 1-1 (green) is
optimal because it shows the largest cumulative
produced-oil volume and results in the highest
net present value.
01/04 12/04 12/05 12/06 12/07 12/08 12/09
Date
0
2
4
6
8
10
12
14
W
a
t
e
r

c
u
t
,

%
01/04 12/04 12/05 12/06 12/07 12/08 12/09
Date
12,000
10,000
8,000
6,000
4,000
2,000
0
O
i
l

p
r
o
d
u
c
t
i
o
n
,

b
b
l
/
d
1,400
G
O
R
,

f
t
3
/
b
b
l
1,200
1,000
800
600
400
200
0
01/04 12/04 12/05 12/06 12/07 12/08 12/09
Date
Trajectory 1-1
Trajectory 1-2
Trajectory 2-2
Winter 2006/2007 29
Another indication that the implementation of
modeling and SiWD will increase is seen in the
growing number of resources dedicated to real-
time drilling solutions. Schlumberger Operation
Support Centers (OSCs), for example, are now
distributed worldwide to remotely monitor, model
and control drilling processes. These centers are
staffed with experienced personnel armed with
powerful software to help operating companies
minimize drilling risks and achieve their drilling
objectives in a collaborative setting (above).
While the results have been encouraging so
far, several areas need further work. Well-path
trajectory optimization would be improved with
the development of automated well-path
selection algorithms. For complete optimization,
more consideration of downhole completion
systems is needed to better simulate the inflow
performance. The increased use of artificial
intelligence techniques should continue to be
explored. There remains a need to couple fluid
flow and geomechanics in SiWD. In addition,
integrating surface and subsurface simulations
would improve the accuracy of production
predictions, although this would add a significant
amount of time to the process. Finally, more work
is needed before SiWD in fractured and other
complex reservoirs becomes feasible. Modeling
dual-porosity and -permeability systems and the
complex interaction between fractures and the
matrix is challenging, even without restrictive
time constraints.
The advances in modeling and simulation
software and hardware, coupled with the E&P
industrys increased understanding of complex
reservoirs and complex wells, will create a
more fertile environment for optimizing well
placement while drilling. MG
>
Real-time workflow. Secure, real-time transmission of downhole data from remote drilling sites is accomplished using the InterACT system or third-party
servers (left). Experts at the Schlumberger Operation Support Centers (OSCs) use this timely information and specialized software tools to help operators
monitor and analyze crucial drilling, geological and geophysical data; avoid drilling hazards; hit reservoir targets; and remotely control drilling operations
(center). Throughout the process, a wide range of data, including depth, time, operational and trajectory data (right), are used to update models, run
simulations and identify the appropriate actions.
Rig sensors
Downhole tools
Wellsite data
acquisition,
aggregation and
display
InterACT
data hub
InterACT hub
Schlumberger OSC
Depth data
GeoFrame
Petrel
Operational data
Time data
Trajectory data
Drilling Office
Specialist services
No Drilling Surprises
PERFORM
Geosteering
Remote monitoring
Remote control
Client asset team
>
Comparison of SiWD results to full-field simulation results of Trajectory 1-1. The optimized, near-wellbore simulation produced results similar to that of a
full-field simulation of Trajectory 1-1 production rates over six years. Initially, the reservoir volume described in the near-wellbore reservoir model has
enough energy to match the full-field simulation volume results. However, after the initial three years in the near-wellbore simulation, the absence of
pressure support from the full reservoir volume shows up as relative permeability beginning to dictate fluid movement.
01/04 12/04 12/05 12/06 12/07 12/08 12/09
Date
1
2
3
4
5
6
7
8
W
a
t
e
r

c
u
t
,

%
0
1,400
G
O
R
,

f
t
3
/
b
b
l
1,200
1,000
800
600
400
200
0
01/04 12/04 12/05 12/06 12/07 12/08 12/09
Date
Full-field simulation
Near-wellbore simulation
01/04 12/04 12/05 12/06 12/07 12/08 12/09
Date
2.0E+07
0
C
u
m
u
l
a
t
i
v
e

o
i
l
,

b
b
l
/
d1.6E+07
1.2E+07
8.0E+06
4.0E+06
2.0E+06
6.0E+06
1.0E+07
1.4E+07
1.8E+07

You might also like