SPE 113918

Insurance Value of Intelligent Well Technology against Reservoir

Uncertainty
E. A. Addiego-Guevara, SPE, and M. D. Jackson, SPE, Department of Earth Science and Engineering, Imperial
College London, UK; and M. A. Giddins, SPE, Schlumberger
Copyright 2008, Society of Petroleum Engineers

This paper was prepared for presentation at the 2008 SPE/DOE Improved Oil Recovery Symposium held in Tulsa, Oklahoma, U.S.A., 1923April2008.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers 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 SPE copyright.


Abstract
Significant challenges remain in the development of optimized control techniques for intelligent wells, particularly with
respect to properly incorporating the impact of reservoir uncertainty. Most optimization methods are model-based and are
effective only if the model can be used to predict future reservoir behavior with no uncertainty. Recently developed schemes,
which update models with data acquired during the optimization process, are computationally very expensive.
We suggest that simple reactive control techniques, triggered by permanently installed downhole sensors, can enhance
production and mitigate reservoir uncertainty across a range of production scenarios. We assess the implementation of an
intelligent horizontal well in a thin oil rim reservoir in the presence of reservoir uncertainty, and evaluate the benefit of using
two completions in conjunction with surface and downhole monitoring. Three control strategies are tested. The first is a
simple, passive approach using a fixed control device to balance inflow along the well, sized prior to installation. The second
and third control strategies are reactive, employing intelligent completions that can be controlled from the surface. The
second strategy opens or closes the completions according to well water cut and flow rate and individual downhole rate and
phase measurements obtained from a surface multiphase flowmeter and alternating zonal well tests. The third strategy
proportionally chokes the completions as increased completion water cut is measured using downhole multiphase
flowmeters.
A cost-benefit analysis demonstrates that reactive control strategies always yield a neutral or positive return, whereas a
passive, model-based strategy can yield negative returns if the reservoir behavior is poorly understood. While reactive control
strategies enhance production and mitigate reservoir uncertainty, they may not deliver the optimum possible solution.
Proactive control techniques, which additionally incorporate data from downhole reservoir-imaging sensors, may yield near-
optimal gains.

Introduction
Intelligent (or smart) wells are equipped with downhole sensors to monitor well and reservoir conditions and with valves to
control the inflow of fluids from the reservoir to the well (Robison 1997). This combination of monitoring and control
technology has the potential to significantly improve oil recovery (Algeroy et al. 1999; Glandt 2005). However, considerable
challenges remain in the formulation of control strategies to operate the valves during production, particularly when there is
uncertainty associated with the reservoir description.
Inflow control to a well can be passive or active (Jansen et al. 2002; Kharghoria 2002). Passive control may be
effective if the reservoir geology and drive mechanisms are well understood so that inflow can be predicted with confidence
using reservoir and well models, and if the predicted inflow does not change significantly with time during production. The
well can then be configured so that hydrocarbon production (or some other objective function) is maximized, by optimizing
the inflow profile along the well using fixed control devices sized prior to installation (e.g., Brekke and Lien 1994; Permadi
et al. 1997).
Active control is facilitated by the adjustable inflow control valves (ICVs) installed in intelligent wells. The settings of
these valves can be varied to optimize the inflow profile along the well in response to monitoring data obtained from
downhole sensors and to the predictions of reservoir and well models. Active control can be either reactive or proactive
(Kharghoria et al. 2002; Yeten et al. 2004; Aitokhuehi and Durlofsky 2005; Ebadi and Davies 2006). Reactive strategies
change the settings of ICVs in response to adverse changes in flowsuch as the arrival of unwanted fluidsmeasured within
the well or the adjacent reservoir.
2 SPE 113918
Proactive (also termed defensive) strategies change the settings of ICVs in response to changes in flow measured or
predicted in the reservoir at some distance away from the well. The advantage of proactive control is that potential problems,
such as the approach of unwanted fluids, can be mitigated before they impact production from the well. The management of
water or gas displacement fronts to prevent early breakthroughby balancing inflow along the length of a wellis a typical
example of proactive control (Armstrong and Jackson 2001; Sinha et al. 2001; Jansen et al. 2002; Kharghoria et al. 2002;
Braithwaite et al. 2004; Glandt 2005; Ebadi and Davies 2006).
The challenge for both reactive and proactive control strategies is to determine the optimal response of the ICVs at the
well. This is a difficult problem, as the optimal response is typically well- and reservoir-specific and varies through time
(Brouwer and Jansen 2004; Yeten et al. 2004; Aitokhuehi and Durlofsky 2005; Naus et al. 2006). Even for a single well, it is
not clear how to balance the often conflicting demands of short-term production optimization against the long-term
optimization of recovery or net present value (NPV) (Naus et al. 2006). The problem is even more challenging for cases
involving multiple wells, or if the location and number of ICVs is included in the overall optimization problem (Yeten and
Jalali 2001).
Despite these challenges, a range of control techniques has been used to optimize production from simulated intelligent
wells which shows that both reactive and proactive control strategies may add significant value in many reservoir and
production settings (Brouwer et al. 2001; Yeten and Jalali 2001; Brouwer and Jansen 2004; Yeten et al. 2004; Aitokhuehi
and Durlofsky 2005; Elmsallati and Davies 2005; Elmsallati et al. 2005; Sarma et al. 2005a; Ebadi and Davies 2006; Naus et
al. 2006). However, a common problem is that these control techniques rely on the predictions of reservoir and well models
to identify the optimal well response. Yet reservoir models are always uncertain to some degree. It is risky to develop a
control strategy based on the predictions of a model that is unlikely to capture the true reservoir behavior. Even history-
matched models can lack predictive value, especially at the temporal and spatial resolution required to optimize inflow to a
well (e.g., Tavassoli et al. 2004; Carter et al. 2006).
Recent studies recognize the problem of incorporating reservoir uncertainty in the optimization workflow (Brouwer et
al. 2004; Yeten et al. 2004; Aitokhuehi and Durlofsky 2005; Sarma et al. 2005b; Naus et al. 2006). They suggest that the
reservoir model should be periodically updated using data measured at the well, and that the updated model should be used to
identify the optimum response of the well over a fixed time interval, prior to the next model update. However, even these
updated models may have limited predictive value, leading to poor control decisions. Moreover, if the models are highly non-
unique, it is necessary to optimize over an ensemble of models, which is computationally expensive (Aitokhuehi and
Durlofsky 2005). As yet, model-based optimization techniques have only been applied to simulated production cases. They
have not been demonstrated in a real field application.
Incorporating reservoir uncertainty is still a challenge in the formulation of active control strategies to operate intelligent
wells. Yet one significant advantage of these wells should be that they can respond to, and mitigate, the adverse impact on
production of unexpected behavior resulting from an uncertain understanding of the reservoir. For example, it is very
common for wells to experience earlier than predicted breakthrough of displacing fluids such as water or gas, caused by
geological heterogeneity not captured in the reservoir model. In a reactive control strategy, the completion at which
breakthrough has occurred can be identified using downhole sensors and flow into that completion reduced or shut-off. In a
proactive strategy, the encroaching fluids need to be detected before they arrive at the well (Kharghoria et al. 2002; Bryant et
al. 2004; Jackson et al. 2005; Saunders et al. 2006).
In this paper, we apply a pragmatic approach to the management of intelligent wells. Our aim is to investigate whether
simple reactive control strategies, based on a feedback loop between ICV settings and surface or downhole measurements,
can enhance production and mitigate reservoir uncertainty if they are designed to work across a range of production
scenarios. It is well known that inflow control can add value in production scenarios where the displacement front reaches the
well at different times along the well length. This can occur, for example, during commingled production from multiple
reservoir layers or compartments, as a result of permeability heterogeneity within a single reservoir, or because of coning or
cresting of the displacement front towards the well (Brouwer et al. 2001; Sinha et al. 2001; Jansen et al. 2002; Kharghoria et
al. 2002; Brouwer and Jansen 2004; Elmsallati et al. 2005; Glandt 2005; Ebadi and Davies 2006; Naus et al. 2006). We
propose simple reactive strategies to control early breakthrough in cases such as these.
We assess the implementation of an intelligent horizontal well in a high-productivity, thin oil rim reservoir in the
presence of reservoir uncertainty. Early water or gas breakthrough is a significant risk in reservoirs of this type, and it has
been shown in several studies that intelligent completions can add value by balancing inflow along the well (Sinha et al.
2001; Jansen et al. 2002; Elmsallati et al. 2005). However, these studies used model-based optimization techniques rather
than the simple feedback approach adopted here; moreover, they did not consider the impact of reservoir uncertainty.
We consider three controlled production strategies against an uncontrolled base case. The first control strategy is a
simple passive approach using a fixed control device (FCD) to balance inflow along the well, sized prior to installation. The
second and third control strategies are reactive, employing two ICVs which can be controlled from the surface. The second
strategy opens or closes the ICVs according to well water cut and flow rate and individual downhole rate and phase
measurements obtained from a surface multiphase flowmeter and alternating zonal well tests. The third strategy
proportionally chokes the ICVs as increased completion water cut is measured using downhole multiphase flowmeters.
SPE 113918 3
A cost-benefit analysis demonstrates that reactive control strategies always yield a neutral or positive return, whereas a
passive strategy can yield negative returns if the reservoir is poorly understood. Simple reactive control strategies enhance
production and mitigate reservoir uncertainty, but may not deliver the optimum possible solution.

Model Formulation
Reservoir Model
The simple, conceptual reservoir model used in this study represents a thin oil rim in a sandstone reservoir containing
interbedded, laterally discontinuous shale barriers that reduce the effective vertical permeability (Fig. 1a; Table 1). A zone of
high vertical permeability represents a region where the shale barriers are absent. This conceptual model was used in an
earlier study (Raghuraman et al. 2003) and is a much-simplified representation of a thin oil rim reservoir in Indiana, U.S.A.
(Bryant et al. 2002). The base case reservoir model has a zone of high vertical permeability which is 700 ft wide and an
active aquifer of 4,500 stb/d/psi. These parameters are later varied in an uncertainty analysis.
The reservoir model measures 6,000 (length) by 3,100 (width) by 100 (thickness) ft and contains 2,500 active grid-
blocks. The top of the reservoir is located at a depth of 6,000 ft and the oil-water contact at a depth of 6,100 ft. The initial
reservoir pressure is 3,600 psia (Table 1). A major simplification is that we neglect the effect of the gas cap; the model
contains only water and undersaturated oil. Strictly speaking, the model represents a thin oil column underlain by an active
aquifer rather than a thin oil rim. However, neglecting the gas cap does not affect the main conclusions and future work will
include both gas and water influx towards the production well. For simplicity, no capillary pressure effects are included and
we assume that the oil and water phases are incompressible.


Well completions 1 & 2 Well completions 1 & 2 Well completions 1 & 2 Well completions 1 & 2
Completion 1 Completion 1 Completion 2 Completion 2
Tubing Tubing
Casing Casing
Annulus flow Annulus flow
ICV 1 ICV 1 ICV 2 ICV 2
Along hole distance Along hole distance 7000 ft 7000 ft 8000 ft 8000 ft 8900 ft 8900 ft
A
c
r
o
s
s

h
o
l
e
,

f
t
A
c
r
o
s
s

h
o
l
e
,

f
t
6000 ft 6000 ft
Tubing flow Tubing flow
Completion 1 Completion 1 Completion 2 Completion 2
Tubing Tubing
Casing Casing
Annulus flow Annulus flow
ICV 1 ICV 1 ICV 2 ICV 2
Along hole distance Along hole distance 7000 ft 7000 ft 8000 ft 8000 ft 8900 ft 8900 ft
A
c
r
o
s
s

h
o
l
e
,

f
t
A
c
r
o
s
s

h
o
l
e
,

f
t
6000 ft 6000 ft
Tubing flow Tubing flow
Inflow from grid blocks
P
f
P
f
tubing segments
annulus
segments
Valve
Segment
Node inlet pressure (P
n
) Node pressure drop (P
n
)
Segment n
Inflow from grid blocks
P
f
P
f
tubing segments
annulus
segments
Valve
Segment
Node inlet pressure (P
n
) Node pressure drop (P
n
)
Segment n
Inflow from grid blocks
P
f
P
f
tubing segments
annulus
segments
Valve
Segment
Node inlet pressure (P
n
) Node pressure drop (P
n
)
Segment n
(a)
(b)
(c)


Figure 1. (a) Cross-section of the reservoir model showing the high-permeability, shale-free zone. The model is modified from an
earlier study (Raghuraman et al. 2003). (b) Implementation of two inflow control valves (ICVs) that control the flow rate of each
completion independently. (c) Corresponding network of well segments used to model the intelligent completions. The well flow
model uses the multisegment well function of the flow simulator.

Reservoir and Well Flow Simulation
Production was simulated using a horizontal well with a length of 2,800 ft and a tubing ID of 6 in. (Fig. 1; Table 2). The
horizontal section of the well was located 35 ft below the top of the oil column. We expected the approaching water front to
break through earlier at the heel of the well, partly because of higher pressure drawdown at the heel and partly because the
4 SPE 113918
zone of high vertical permeability allows water influx from the underlying aquifer (Fig. 1a). Consequently, the well was
completed in two sections, one close to the heel and the other close to the toe, so inflow along the well could be balanced
(Fig. 1b). Selecting the number and location of completions is itself an optimization problem and we have chosen here to use
a pragmatic approach based on basic reservoir and production engineering considerations. This may not yield the optimal
configuration.
Production was simulated using a commercial reservoir and well flow simulator (Schlumberger 2007). Production
scenarios utilized a multisegment well model (Holmes et al. 1998; Holmes 2001), which divides the wellbore and any lateral
branches into one-dimensional segments, including representation of control devices as separate segments (Fig. 1c). Each
segment has its own set of independent variables to describe the local flow conditions.

Table 1. Reservoir Model Properties

Parameter Amount Unit
Length 6000 ft.
Width 3100 ft.
Height 100 ft.
Top 6000 ft.
Oil water contact 6100 ft.
Initial pressure 3600 psi
NTG 1.00 [-]
GRV 331.30 MMbbl
Vertical permeability (Low) 10 mD
Vertical permeability (High) 500 mD
Horizontal permeability 500 mD
Connate water Saturation 0.25 [-]
Water end-point relative permeabilty 0.80 [-]
Residual oil saturation 0.25 [-]
Oil end-point relative permeabilty 0.80 [-]
Oil viscosity 0.34 cp
Oil density 49.94 lb/cft
Water viscosity 0.50 cp
Water density 62.43 lb/cft
Oil in place 57.1 MMbbl
Mobile pore volume 41.4 MMbbl
Max. recovery factor 72.5 %
Field
Reservoir
Dimensions
STOIIP
Fluid
properties
Reservoir
Properties
Rock
Properties


Production Strategy
Production was simulated under fixed surface liquid rate control of 10,000 stb/day, with a minimum bottomhole pressure
(BHP) limit of 1,500 psia. As an additional production constraint, the well was shut in once the water cut, measured at the
wellhead, exceeded 25%. This low water-cut limit represents field cases with restricted water production handling facilities
(e.g., Arnold et al. 2004). It also allows us to compare the performance of different conventional and intelligent production
strategies before a workover would typically be implemented in a conventional well to reduce water production.
We considered four different downhole well configurations. The first consists of two conventional completions and
represents the uncontrolled base-case against which the controlled production cases are compared. The second has one
fixed FCD installed downhole in the completion closest to the heel of the well, which is sized to balance inflow along the
length of the well. The third and fourth well configurations are equipped with downhole sensors and one ICV in each
completion, to measure and react against unfavorable downhole flow changes (Fig. 1b). These configurations employ either
simple on/off ICVs or more sophisticated ICVs that open and close proportionally to the water cut measured at the
completion.

Fixed flow control
The fixed flow control strategy employs one fixed control device installed in the completion closest to the heel of the well
(Fig. 1a). Flow control devices regulate the inflow profile along a horizontal well by imposing an additional pressure drop
between the sand face and the tubing (Fig. 2a). This pressure drop is adjustable prior to deployment (e.g., Brekke and Lien
1994; Permadi et al. 1997; Jansen et al. 2002). We modeled the FCD as a subcritical valve (Fig. 2b), located between the
SPE 113918 5
annulus and tubing sections of the completion (Fig. 1b). The valve was sized by simulating production over a relatively short
period of 12 months using the base case reservoir model (Fig. 1a). The optimum size was selected by maximizing an
objective function given by

WWPT C WOPT WNPV
W fixed
= (1)

where WOPT denotes the total oil production, WWPT the total water production, and C
W
is a cost conversion factor with a
value of 0.04 BOE/bbl water (Table 2). The optimal FCD setting obtained from the base-case model remained fixed in all
reservoir realizations.

Table 2. Operating and Development Costs




0
2000
4000
6000
8000
10000
12000
14000
0.00 1.00 2.00 3.00 4.00 5.00 6.00
Valve Constriction Size [in]
C
o
m
p
l
e
t
i
o
n

l
i
q
u
i
d

f
l
o
w

r
a
t
e

[
r
b
/
d
]
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
P
r
e
s
s
u
r
e

d
r
o
p

[
p
s
i
a
]
Flowrate
Pressure Drop
Pos. 11 (fully open)
Pos. 1 (closed)
2
3
4
5
6
7
8
9
10
reservoir pressure
sandface pressure
Distance from heel
Pressure
reservoir pressure
sandface pressure
Distance from heel
Pressure
tubing pressure
P
f
reservoir pressure
sandface pressure
Distance from heel
Pressure
reservoir pressure
sandface pressure
Distance from heel
Pressure
tubing pressure
P
f
(a)
(b)

Figure 2. (a) Flow control devices (FCDs) regulate the inflow profile along a horizontal well by imposing an additional pressure drop
P
f
between the sand face and the tubing. (b) FCDs were modeled as subcritical valves with fixed settings that ranged from fully
open (position 11) to closed (position 1). The plot shows the simulated completion liquid flow rate, at reservoir conditions, for the
valve installed in completion 1 (Fig. 1), and the pressure drop across it. Readings were taken after the first month of production.
Well model kept under bottomhole flowing pressure control of 3,540 psia during the test.
Parameter Amount Unit
Length 6000 ft.
Width 3100 ft.
Height 100 ft.
Top depth 6000 ft.
Oil water contact 6100 ft.
Initial pressure 3600 psi
GRV 331.30 MMbbl
Vertical permeability (Low) 10 mD
Vertical permeability (High) 500 mD
Horizontal permeability 500 mD
Connate water saturation 0.25 [-]
Water end-point relative permeabilty 0.80 [-]
Residual oil saturation 0.25 [-]
Oil end-point relative permeabilty 0.80 [-]
Oil viscosity 0.34 cp
Oil density 49.94 lb/cft
Water viscosity 0.50 cp
Water density 62.43 lb/cft
Oil in place 57.1 MMbbl
Mobile pore volume 41.4 MMbbl
Max. recovery factor 72.5 %
Field
Reservoir
Dimensions
STOIIP
Fluid
properties
Reservoir
Properties
Rock
Properties
6 SPE 113918

On/off flow control
The on/off flow control strategy employs two ICVs that can switch between open and closed (on/off) positions (Jackson-
Nielsen and Tips 2003). We opened or closed these ICVs based on interpretations of wellhead water-cut measurements and
downhole flow measurements (Fig. 3). The technical capability to implement this operational strategy is available today.
Measurements of water cut at the well-head trigger a smart workover, which differs from a conventional workover
because the ICVs allow the intelligent completions to be opened or closed remotely from the surface. Production from other
producing zone(s) is not interrupted and a workover rig is not required, so these cash-intensive operations are avoided (Table
2). Operating on/off ICVs allows the flow rate of each phase through each completion to be determined through well testing
by exception and online well tests, in which each completion is analyzed in isolation by selectively opening individual
ICVs and measuring wellhead pressure and multiphase flow rates (Akram et al. 2001; Gai 2001; Paino et al. 2004).
A smart workover is triggered when the wellhead water cut exceeds an arbitrarily low value of 8%. Production is
stopped for one month while a zonal well test is conducted to determine the water production of each completion. If water
production is greater in the completion located at the heel of the well, then this completion is closed; if water production is
greater from the completion at the toe, then it is assumed that choking can no longer improve recovery so both completions
are opened and the well is produced until it violates one of the production constraints. When the smart workover is complete,
the well is reopened and not checked again for six months. Note that production from the well is reduced during these
workovers.

Start
production
WWCT >
Trigger
Value
Stop well for 1
month
(workover)
ICV1 worst
performing
completion
ICV2 worst
performing
completion
Shut ICV1
Open both
ICVs
Open well
(end
workover)
Wait 3
months
(planning)
Dont check
well again in
next 6
months
Continue
production
Start
production
WWCT >
Trigger
Value
Stop well for 1
month
(workover)
ICV1 worst
performing
completion
ICV2 worst
performing
completion
Shut ICV1
Open both
ICVs
Open well
(end
workover)
Wait 3
months
(planning)
Dont check
well again in
next 6
months
Continue
production
[1] [1] [1] [1]
[2] [2] [2] [2]
Yes
Yes Yes
No
[3] [3] [3] [3]
No
(a)
(b)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
0 100 200 300 400 500 600 700 800
Time [days]
C
u
m
u
l
a
t
i
v
e

O
i
l

P
r
o
d
u
c
t
i
o
n

[
M
M
S
T
B
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
e
l
l

a
n
d

I
C
V

W
a
t
e
r

C
u
t

[
-
]
Total Oil Production [MMbbl] Well water-cut [-] ICV 1 water-cut [-] ICV 2 water-cut [-]
[1]
[3] [2]
[1]

Figure 3. (a) Workflow for the on/off control strategy and (b) corresponding well and completion watercut. A smart workover [1] is
scheduled when the well water cut exceeds 8% [1]. The workover controls flow into each completion [2]. Production continues after
one month [3] and the well status is not monitored again for six months. Well production terminates when the flow conditions
violate the production constraints. ICV1 is located at the heel of the well and ICV 2 at the toe (Fig. 1a).

Proportional flow control
The proportional control strategy employs more technically advanced intelligent completions equipped with continuously
variable ICVs that have an infinite number of valve settings (Williamson et al. 2000), and multiphase downhole
flowmeters that can measure the flow rate of each fluid phase in each completion (Kragas 2003a, 2003b; Webster et al.
2006). A simple feedback loop chokes back production from each downhole completion as the measured completion water
cut increases using a relationship of the form

+ = 1 ,
1
. max .
c
o c
WCT
WCT
B A P P (2)

SPE 113918 7
where P
c
is the pressure drop across the ICV for a given choke setting, P
o
is the pressure drop when the ICV is fully open,
WCT is the completion water cut, and A, B and c are constants that are specific to the production case (Yeten et al. 2002).
The values of these constants can be optimized using a simulation model, although the resulting values are unlikely to be
optimal if the reservoir does not behave as predicted. We selected values using the base case reservoir model (Fig. 1a) and
maximizing Equation (1). These values remain fixed in all reservoir realizations (Fig. 4).
It should be noted that this control strategy is not yet commercially available. Real-time downhole multiphase flow
measurements have not been applied in closed-loop workflows; currently, an operator interprets these measurements and
production control decisions are made separately at a later stage (Webster et al. 2006).

1.0 1.2x10
5
1.00
0.5 2.6x10
5
-5x10
4
c B A
1.0 1.2x10
5
1.00
0.5 2.6x10
5
-5x10
4
c B A 0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 100 200 300 400 500 600 700 800
Time [days]
O
i
l

P
r
o
d
u
c
t
i
o
n

R
a
t
e

[
M
S
T
B
/
D
A
Y
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
e
l
l

a
n
d

I
C
V

W
a
t
e
r

C
u
t

[
-
]
ICV 1 Oil Production [Mbbl/day] ICV 2 Oil Production [Mbbl/day] Well water-cut [-] ICV 1 water-cut [-] ICV 2 water-cut [-]
[1]
[3]
[2]
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
0 100 200 300 400 500 600 700 800
Time [days]
P
r
e
s
s
u
r
e

D
r
o
p

a
c
r
o
s
s

C
o
m
p
l
e
t
i
o
n

[
p
s
i
a
]
Pressure Drop ICV 1 [psia] Pressure Drop ICV 2 [psia]
[1]
[3]
[2]
0
20000
40000
60000
80000
100000
0.00 0.05 0.10 0.15 0.20 0.25
Completion Water Cut [-]
P
r
e
s
s
u
r
e

D
r
o
p

M
u
l
t
i
p
l
i
e
r

[
-
]
SPRD<60 psia
SPRD>60 psia
(b)
(c)
(a)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 100 200 300 400 500 600 700 800
Time [days]
O
i
l

P
r
o
d
u
c
t
i
o
n

R
a
t
e

[
M
S
T
B
/
D
A
Y
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
e
l
l

a
n
d

I
C
V

W
a
t
e
r

C
u
t

[
-
]
ICV 1 Oil Production [Mbbl/day] ICV 2 Oil Production [Mbbl/day] Well water-cut [-] ICV 1 water-cut [-] ICV 2 water-cut [-]
[1]
[3]
[2]
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
0 100 200 300 400 500 600 700 800
Time [days]
P
r
e
s
s
u
r
e

D
r
o
p

a
c
r
o
s
s

C
o
m
p
l
e
t
i
o
n

[
p
s
i
a
]
Pressure Drop ICV 1 [psia] Pressure Drop ICV 2 [psia]
[1]
[3]
[2]
0
20000
40000
60000
80000
100000
0.00 0.05 0.10 0.15 0.20 0.25
Completion Water Cut [-]
P
r
e
s
s
u
r
e

D
r
o
p

M
u
l
t
i
p
l
i
e
r

[
-
]
SPRD<60 psia
SPRD>60 psia
(b)
(c)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 100 200 300 400 500 600 700 800
Time [days]
O
i
l

P
r
o
d
u
c
t
i
o
n

R
a
t
e

[
M
S
T
B
/
D
A
Y
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
e
l
l

a
n
d

I
C
V

W
a
t
e
r

C
u
t

[
-
]
ICV 1 Oil Production [Mbbl/day] ICV 2 Oil Production [Mbbl/day] Well water-cut [-] ICV 1 water-cut [-] ICV 2 water-cut [-]
[1]
[3]
[2]
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
0 100 200 300 400 500 600 700 800
Time [days]
P
r
e
s
s
u
r
e

D
r
o
p

a
c
r
o
s
s

C
o
m
p
l
e
t
i
o
n

[
p
s
i
a
]
Pressure Drop ICV 1 [psia] Pressure Drop ICV 2 [psia]
[1]
[3]
[2]
0
20000
40000
60000
80000
100000
0.00 0.05 0.10 0.15 0.20 0.25
Completion Water Cut [-]
P
r
e
s
s
u
r
e

D
r
o
p

M
u
l
t
i
p
l
i
e
r

[
-
]
SPRD<60 psia
SPRD>60 psia
(b)
(c)
(a)

Figure 4. (a) Oil rate and water cut and (b) downhole pressure drop curves for the two intelligent proportional completions. In this
example, the pressure drop across ICV1 increases as the completion water cut increases, while the liquid production decreases [1].
ICV2 starts to close when the watercut in this completion increases [2]. (c) Pressure drop multiplier vs. completion water-cut
measurement at given values of A, B, and c (Equation 3). The continuous (green) curve in Fig. 4c corresponds to the behavior of the
proportional intelligent completions during normal flow. A safety release mechanism opens both valves when the pressure drop
across them is greater than a triggering value of 60 psia [3]; the dotted (orange) curve in Fig. 4c represents the flow behavior of the
completions at this stage. ICV1 is located at the heel of the well and ICV 2 at the toe (Fig. 1a).

Objective Function
Our aim is to estimate the incremental return generated by the extra investment for each control strategy. We use a
customized objective function to measure and compare their performance at the end of simulated production, given by

cinit W G
W WWPT C WGPT C WOPT WNPV + = (3)

where WOPT is the total oil produced, C
G
is a gas value conversion factor, WGPT is the total gas produced, C
W
is a water cost
conversion factor, WWPT is the total water produced, and W
cinit
is the initial well cost. The resulting value of the well WNPV
is calculated in units of equivalent barrels of oil at surface conditions (BOE). This approach allows us to perform a
comparative analysis without embarking on complex economic calculations. We assume that the technology cost
depreciation rate over time is of similar magnitude to a correspondent discount rate (inflation).
We also define an additional parameter, termed the Incremental Return generated by Extra Investment (IREI), to
compare the performance of each control strategy per unit incremental cost. The IREI term is calculated in units of equivalent
barrels of oil gained for each extra equivalent barrel of oil invested. For example, an IREI of 10 yields a profit of 10
equivalent barrels of oil for each extra equivalent barrel of oil invested.
8 SPE 113918
We modified the operating costs (Table 2) from earlier studies (Algeroy et al. 1999; Raghuraman et al. 2003; Arnold et
al. 2004; Poel and Jansen 2004). The initial investment expenditure for the conventional horizontal well is summarized in
Table 3. Different downhole control strategies incur distinct extra development costs, shown in Table 4. The proportional
control strategy is the most expensive, while the fixed strategy is the cheapest.

Table 3. Initial development cost (W
cinit
term in Equation 3) for the base (uncontrolled
production) development. Costs are calculated using average values converted into
barrels of oil equivalent (BOE) units.


Development Costs
Uncontrolled [$MM]
# Wells 1 9.00
Drilling one
horizontal well
Zones/well 2 0.66
Sand screen
per zone
Packers/well 3 0.09 per packer
Well Monitoring Surface 0.10 per well
Well Control Surface 0.15 per well
Total [$MM] 10.8
Total [MMBOE] 0.22
Total costs
analysis
Well components
Well
Configuration



Table 4. Relative development cost (W
cinit
term in Equation 3) and configuration of the four different
production scenarios. Costs are calculated using average values converted into barrels of oil equivalent
(BOE) units.


Costs
Uncontrolled FIXED ON/OFF PROPORT. [$MM]
FCD (fixed) - - - 0.05 per valve
ICV (on/off) - - - 0.10 per valve
ICV (continuous) - - - 0.15 per valve
Deployment
Extra installation
days
- 2 3 4 0.50 per day
Well Monitoring
Permanent
downhole
sensors
- - - 0.15 per well
On/Off - - - 0.10 per well
Proportional - - - 0.15 per well
Total [$MM] 10.8 11.9 12.5 13.3
Total [MMBOE] 0.22 0.24 0.25 0.27
Extra Cost [%] - 9.7% 15.7% 22.6%
Development
Downhole
Control Devices
Total costs
analysis
Downhole Well
Control
Well components




Base Case Model Predictions
We begin by simulating production from the base case reservoir model (Fig. 1a) and comparing the different control
strategies. As expected, the conventional, uncontrolled production strategy yields early water breakthrough at the completion
closest to the heel of the well (Fig. 5a), partly because of the higher drawdown at the heel and partly because of the higher
vertical permeability adjacent to the heel (Fig. 1a). The well is shut in after approximately 2 years of production when the
water-cut limit is reached, leaving significant volumes of bypassed oil in the reservoir (Fig. 6a). Similar behavior has been
observed in numerous studies of oil rim production via horizontal wells (Sinha et al. 2001; Jansen et al. 2002; Kharghoria et
al. 2002; Ebadi and Davies 2006; Naus et al. 2006).
Each of the three controlled production strategies yields higher oil recovery and lower water production than the
uncontrolled case (Fig. 5b). In the fixed control strategy, water breakthrough is delayed because the FCD installed in the
completion closest to the heel of the well balances inflow along the well, yielding a more uniform displacement front (Fig.
SPE 113918 9
6b). In the reactive control strategies, early water breakthrough occurs at the completion closest to the heel; it is identified
and the corresponding ICV either closed or choked back to reduce water production. In all three controlled production cases,
the target oil rate is maintained despite choking back inflow, although the well BHP falls to compensate for the reduced
productivity of the well.

0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 100 200 300 400 500 600 700 800
Time [days]
C
u
m
u
l
a
t
i
v
e

o
i
l

p
r
o
d
u
c
t
i
o
n

[
M
M
S
T
B
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
a
t
e
r

C
u
t

[
-
]
Well water-cut [-] Total Oil Production [MMbbl] ICV 1 water-cut [-] ICV 2 water-cut [-]
(a)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
0 100 200 300 400 500 600 700 800
Time [days]
C
u
m
u
l
a
t
i
v
e

o
i
l

p
r
o
d
u
c
t
i
o
n

[
M
M
S
T
B
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
a
t
e
r

C
u
t

[
-
]
Well water-cut [-] Total Oil Production [MMbbl] ICV 1 water-cut [-] ICV 2 water-cut [-]
(a)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
0 100 200 300 400 500 600 700 800 900
Time [days]
C
u
m
u
l
a
t
i
v
e

o
i
l

p
r
o
d
u
c
t
i
o
n

[
M
M
S
T
B
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
a
t
e
r

C
u
t

[
-
]
Oil Prod. 'Uncontrolled' Oil Prod. 'FIXED' Oil Prod. 'ON/OFF' Oil Prod. 'PROP.'
Water-cut - 'Uncontrolled' Water-cut - 'FIXED' Water-cut - 'ON/OFF' Water-cut - 'PROP.'
(b)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
0 100 200 300 400 500 600 700 800 900
Time [days]
C
u
m
u
l
a
t
i
v
e

o
i
l

p
r
o
d
u
c
t
i
o
n

[
M
M
S
T
B
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
a
t
e
r

C
u
t

[
-
]
Oil Prod. 'Uncontrolled' Oil Prod. 'FIXED' Oil Prod. 'ON/OFF' Oil Prod. 'PROP.'
Water-cut - 'Uncontrolled' Water-cut - 'FIXED' Water-cut - 'ON/OFF' Water-cut - 'PROP.'
(b)

Figure 5. (a) Oil production, total and segment water cut for the uncontrolled production strategy in the base case reservoir model.
(b) Oil production and water cut for the uncontrolled and controlled production strategies in the base case reservoir model.

All three controlled production strategies yield higher values of WNPV (Equation 3) compared to the uncontrolled case
(Fig. 7). The highest value of WNPV is obtained using proportional reactive control, which is the most sophisticated control
strategy. The increase in WNPV is mainly due to a significant increase in oil recovery relative to the uncontrolled case, with
some additional benefit from reduced water production. However, the WNPV of the least sophisticated, fixed control strategy
is similar, due to a more modest increase in oil recovery, a significant decrease in water production, and lower initial costs.
The on/off reactive control strategy yields the largest decrease in water production but also the lowest increase in oil
recovery, so it has a lower WNPV than either the fixed or proportional control strategies.
Although the proportional control strategy yields the highest WNPV, the fixed control strategy yields the highest IREI.
This is because the costs associated with installing the more sophisticated reactive control equipment are significantly higher
(Table 4). The lowest IREI is obtained from the on/off control strategy, because this incurs significant additional installation
costs (although lower than for the proportional case [Table 4]) without yielding a comparable increase in WNPV.
Consequently, the fixed control strategy appears to be the most attractive option. However, it should be noted that this will
depend upon the value of the oil and gas produced, water handling costs, and the additional installation costs.
This simple analysis suggests that a fixed control strategy is attractive if the device can be optimally sized using a
reservoir simulation model and the optimal size does not change significantly during production. The additional costs
10 SPE 113918
associated with installing an FCD are less than those associated with more sophisticated control options, and recovery can be
improved by balancing inflow along the well and delaying the breakthrough of unwanted fluids. These findings are not
surprising and are included primarily to confirm that our model predictions are in line with those of previous studies (Brekke
and Lien 1994; Permadi et al. 1997; Jansen et al. 2002). However, they are based on the assumption that there is no
uncertainty associated with the reservoir description or behavior.
Oil Saturation [%] 25.0% 75.0%
Q
total
=10000 stb/day Well completions 1 & 2
(b)
(a)
Oil Saturation [%] 25.0% 75.0% Oil Saturation [%] 25.0% 75.0%
Q
total
=10000 stb/day Well completions 1 & 2 Q
total
=10000 stb/day Well completions 1 & 2 Q
total
=10000 stb/day Well completions 1 & 2
(b)
(a)


Figure 6. Cross-section showing oil saturation for the (a) uncontrolled and (b) fixed control development strategies in the base case
reservoir model. Snapshots taken at the same time during production. As expected, in the uncontrolled case (a) water breaks
through earlier at the heel of the well; there is a region of bypassed oil at the toe. In the controlled case (b), the choked heel
completion allows the water front to sweep the oil more uniformly.

Sources of change
25 17 48 IREI [ratio]
15 7 12 GAIN [%]
23 16 10 Extra well cost [%]
-8 -34 -27 Water [%]
15 7 12 Hydrocarbon [%]
8.12
9.13
8.69
9.35
7.50
8.00
8.50
9.00
9.50
Uncontrolled FIXED ON/OFF PROPORTIONAL
T
o
t
a
l

p
r
o
d
u
c
t
i
o
n

(
W
N
P
V
)

[
M
M
B
O
E
]
Sources of change
25 17 48 IREI [ratio]
15 7 12 GAIN [%]
23 16 10 Extra well cost [%]
-8 -34 -27 Water [%]
15 7 12 Hydrocarbon [%]
8.12
9.13
8.69
9.35
7.50
8.00
8.50
9.00
9.50
Uncontrolled FIXED ON/OFF PROPORTIONAL
T
o
t
a
l

p
r
o
d
u
c
t
i
o
n

(
W
N
P
V
)

[
M
M
B
O
E
]
8.12
9.13
8.69
9.35
7.50
8.00
8.50
9.00
9.50
Uncontrolled FIXED ON/OFF PROPORTIONAL
T
o
t
a
l

p
r
o
d
u
c
t
i
o
n

(
W
N
P
V
)

[
M
M
B
O
E
]


Figure 7. Production totals for different downhole control strategies in the base case reservoir model.

Uncertainty Analysis
We now investigate the impact of uncertainty in the reservoir description on the performance of each production strategy.
Two parameters are varied in the reservoir model: the width of the shale-free zone of high vertical permeability and the
aquifer strength. Both of these would be difficult to characterize, especially early in field life. Three discrete values are used
to describe each parameter, resulting in nine reservoir realizations, one of which is the base-case already considered (Fig. 8).
We assume that the uncertain parameters are normally distributed, with the mean and standard deviation obtained
from a reservoir characterization study (Bryant et al. 2002). Each continuous distribution is sampled by the three discrete
values, so that the mean and standard deviation of these discrete values is equal to the original mean and standard deviation
(Raghuraman et al. 2003).

The control strategies remain unchanged from the base case; we do not optimize them for each realization. The base
case reservoir model (Fig. 1a) is assumed to represent the reservoir model used to optimize the control strategies, while the
eight realizations around the base case are assumed to represent a range of possible reservoir behaviors which are not
predicted by the model. Our aim is to determine whether simple passive and reactive control strategies can add value when
the reservoir does not perform as predicted. A performance matrix is used to compare the results (Figs. 911). The x-axis
represents the return generated by the extra investment (IREI), while the y-axis represents the additional WNPV (or gain)
generated by a given control strategy relative to the uncontrolled production case.
We find that three cases show high gains and high IREI (Fig 9); these cases have the widest zone of high vertical
permeability and the strongest aquifer (Fig. 8). The well NPV is increased by up to 30% compared to the uncontrolled
production case, while the IREI is up to 90 BOE/BOE. Water breakthrough at the completion closest to the heel occurs
SPE 113918 11
earliest in these realizations, so there is a greater potential to decrease water production and increase oil recovery using an
inflow control strategy. The highest gain is consistently obtained using proportional control, reflecting the dynamic updating
of the ICV settings in response to downhole measurements of phase flow rates through each completion. However, fixed
control consistently shows the highest IREI, reflecting the lower upfront costs associated with this much simpler technology.
The on/off strategy yields the lowest gain and IREI.
1000 4500 8000
Low (1) Mean (2) High (3)
300
Low
(1)
W1A1 W1A2 W1A3
700
Mean
(2)
W2A1
Base
Case
W2A3
1100
High
(3)
W3A1 W3A2 W3A3
W
i
d
t
h

[
f
t
]
Aquifer Strength [stb/d/psi]
Width of high-permeability zone (dark regions)
Statistics:
Width
Mean = 700 ft
STD = 327 ft
Aquifer strength
Mean = 4500 stb/d/psi
STD = 2858 stb/d/psi
Well completions 1 & 2
(a)
(b)
1000 4500 8000
Low (1) Mean (2) High (3)
300
Low
(1)
W1A1 W1A2 W1A3
700
Mean
(2)
W2A1
Base
Case
W2A3
1100
High
(3)
W3A1 W3A2 W3A3
W
i
d
t
h

[
f
t
]
Aquifer Strength [stb/d/psi]
Width of high-permeability zone (dark regions)
Statistics:
Width
Mean = 700 ft
STD = 327 ft
Aquifer strength
Mean = 4500 stb/d/psi
STD = 2858 stb/d/psi
Well completions 1 & 2 Well completions 1 & 2 Well completions 1 & 2
(a)
(b)

Figure 8. Reservoir models representing uncertainty in (i) the width of the zone of high vertical permeability and (ii) aquifer strength.

-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
-20 0 20 40 60 80 100 120
IREI (return generated by extra investement) [ratio]
G
A
I
N

[
%
]
Loss
Profit
Low Value High Value
Fixed
On/Off
Proportional
High width, high aquifer (W3A3)
High width, medium aquifer (W3A2)
Medium width, high aquifer (W2A3)
Low (1) Med (2) High (3)
L
o
w

(
1
)
W1A1 W1A2 W1A3
M
e
d

(
2
)
W2A1
Base
Case
W2A3
H
i
g
h

(
3
)
W3A1 W3A2 W3A3
Aquifer Strength
W
i
d
t
h

Figure 9. Increase in WNPV (gain) over the uncontrolled production case vs. return generated by extra investment for each
controlled production case in reservoir realizations with a wide zone of high permeability and strong aquifer.

Two cases show marginal or negative gains and IREI (Fig. 10); these cases have a wide zone of high vertical
permeability and a small aquifer, and vice-versa. Proportional control in these cases consistently yields the highest gain (c. 5
%) and IREI (c. 14 BOE/BOE), although these are much lower than in realizations with a wide high-permeability zone and a
strong aquifer. However, the on/off strategy exhibits a negative gain and IREI in one case, while the fixed strategy exhibits a
negative gain and IREI in the other case. In both cases, the control strategy developed using the base case reservoir model is
suboptimal for the actual reservoir behavior. The fixed control strategy cannot be changed without a conventional workover.
The on/off strategy is reactive only over a relatively long time scale, because phase flow rates through each
completion must be identified using zonal tests. Oil production is reduced during these tests, although the well is not shut in.
Moreover, once the offending completion has been identified, the ICV is closed for six months until the well is tested again.
If the inflow characteristics change during this time, the ICV settings are no longer optimal. Consequently, the on/off reactive
control strategy yields a smaller increase in oil recovery compared to the proportional strategy. Moreover, because the
upfront cost of installing the ICVs is still relatively high, it does not always yield a positive return on investment.
Three cases show negative gains and IREI for the fixed control strategy, but marginal or positive gains for the
proportional control strategy (Fig. 11). These cases have the narrowest zone of high permeability and the smallest aquifer.
Water breakthrough still occurs earliest at the completion close to heel of the well, but the water migrates less rapidly
12 SPE 113918
towards the heel than in the base case model, so breakthrough occurs later than predicted. The on/off control strategy yields a
mixture of marginal positive and negative gains and IREI.

-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
-10 -5 0 5 10 15 20
IREI (return generated by extra investement) [ratio]
G
A
I
N

[
%
]
Loss
Profit
Low Value High Value
Fixed
On/Off
Proportional
Low width, high aquifer (W1A3)
High width, low aquifer (W3A1)
Low (1) Med (2) High (3)
L
o
w

(
1
)
W1A1 W1A2 W1A3
M
e
d

(
2
)
W2A1
Base
Case
W2A3
H
i
g
h

(
3
)
W3A1 W3A2 W3A3
W
i
d
t
h
Aquifer Strength

Figure 10. Increase in WNPV (gain) over the uncontrolled production case vs. return generated by extra investment, for each
controlled production case in reservoir realizations with a wide zone of high permeability and weak aquifer, and vice-versa.

-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
-60 -50 -40 -30 -20 -10 0 10 20
IREI (return generated by extra investement) [ratio]
G
A
I
N

[
%
]
Loss
Profit
Low Value High Value
Fixed
On/Off
Proportional
Low width, low aquifer (W1A1)
Low width, medium aquifer (W1A2)
Medium width, low aquifer (W2A1)
Low (1) Med (2) High (3)
L
o
w

(
1
)
W1A1 W1A2 W1A3
M
e
d

(
2
)
W2A1
Base
Case
W2A3
H
i
g
h

(
3
)
W3A1 W3A2 W3A3
Aquifer Strength
W
i
d
t
h

Figure 11. Increase in WNPV (gain) over the uncontrolled production case vs. return generated by extra investment for each
controlled production case in reservoir realizations with a narrow zone of high permeability and weak aquifer.

The base case reservoir model was used to size the FCD installed at the heel completion in the fixed control strategy.
Consequently, the fixed control strategy chokes production from this completion harder than is necessary to balance inflow
along the well. This yields lower oil recovery than the uncontrolled case, although water production is also lower (e.g., Fig.
12). The fixed control strategy therefore yields negative returns. In contrast, the proportional control strategy reacts only
when water breaks through at the completion. Production is choked back proportionately to the water cut, and although the
relationship between water cut and choke size was optimized to the base case reservoir model, it still yields a sensible control
response if water breakthrough occurs later (or earlier) than expected. The proportional control strategy therefore yields
higher oil recovery (e.g., Fig. 12), and marginal or positive gains and IREI. These gains are small, because inflow control
yields smaller benefits as water breakthrough becomes more evenly distributed along the well (Brouwer et al. 2001; Sinha et
al. 2001; Jansen et al. 2002; Kharghoria et al. 2002; Brouwer and Jansen 2004; Elmsallati et al. 2005; Glandt 2005; Ebadi
and Davies 2006; Naus et al. 2006). The on/off control strategy also yields lower oil recovery than the uncontrolled case in
some production scenarios, yielding a negative return on investment.

SPE 113918 13
0.00
2.00
4.00
6.00
8.00
10.00
12.00
0 200 400 600 800 1000 1200
Time [days]
C
u
m
u
l
a
t
i
v
e

o
i
l

p
r
o
d
u
c
t
i
o
n

[
M
M
S
T
B
]
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
W
a
t
e
r

C
u
t

[
-
]
Oil Prod. 'Uncontrolled' Oil Prod. 'FIXED' Oil Prod. 'ON/OFF' Oil Prod. 'PROP.'
Water-cut - 'Uncontrolled' Water-cut - 'FIXED' Water-cut - 'ON/OFF' Water-cut - 'PROP.'

Figure 12. Oil production and water cut for each production strategy in the medium-width, low-aquifer-strength scenario (W2A1).

Discussion
Of the three controlled production strategies we have investigated, the passive strategy yields the highest return on
investment if the reservoir behaves as predicted, so the control device can be sized prior to installation (Fig. 13). There is a
gain in NPV relative to the base case of uncontrolled production, which is obtained by balancing inflow between the heel and
toe completions of the well to delay water breakthrough, yielding accelerated and increased oil recovery, and reduced water
handling costs. This translates to a high return on investment, because of the relatively low installation cost of an FCD
compared to more sophisticated control strategies.
The proportional control strategy yields the highest gain in WNPV, obtained from significantly improved oil recovery
(Fig. 13). This reflects the continual updating of the ICV settings in response to data acquired downhole. However, the
maximum return on investment obtained using the proportional strategy is less than that obtained using the fixed strategy,
because of the significantly higher upfront costs associated with installing the more complex downhole equipment.
The fixed strategy also yields the highest return on investment if the reservoir does not behave as predicted, but water
breakthrough occurs earlier than in the base case reservoir model. However, if water breakthrough occurs later than
predicted, the strategy yields a negative return on investment. The proportional strategy always yields a neutral or positive
return, regardless of reservoir behavior (Fig. 13). The return is higher if breakthrough occurs earlier, because this is a
favorable scenario for inflow control to yield benefit.
Although passive control yields the highest returns, it is a risky strategy if the reservoir behavior cannot be predicted
with certainty. This is the case in almost all reservoir developments. A reactive control strategy provides a measure of
insurance against reservoir uncertainty, because adverse changes in flow can be detected downhole and inflow changed in
response. Moreover, our results suggest that the response strategy does not need to be tailored to each possible reservoir
realization; as long as it provides a sensible control strategy for the base case model, it is robust over a range of reservoir
uncertainty. An approach such as this is much simpler to apply in practice than model-based control strategies that
incorporate reservoir uncertainty, although these will yield closer to optimal control of the well (Brouwer et al. 2004; Yeten
et al. 2004; Aitokhuehi and Durlofsky 2005; Sarma et al. 2005b; Naus et al. 2006).
Our results suggest that a simple, reactive control strategy can provide insurance against reservoir uncertainty. The
strategy can be optimized on a base case reservoir model, or across an ensemble of models that reflect the uncertain reservoir
description. Because the reservoir will not behave exactly as predicted, the control strategy will not yield the optimal value of
the chosen objective function. However, because it can react to changes in flow measured downhole, the reactive strategy is
less risky than a passive strategy or active strategies driven only by reservoir model predictions. However, it should be noted
that inflow control will not yield benefit in all production scenarios. It is particularly useful when the displacement front
reaches the well at different times along the well length (Brouwer et al. 2001; Sinha et al. 2001; Jansen et al. 2002;
Kharghoria et al. 2002; Brouwer and Jansen 2004; Elmsallati et al. 2005; Glandt 2005; Ebadi and Davies 2006; Naus et al.
2006). A scoping study must be conducted to determine whether control is likely to yield benefit prior to installation of the
well (e.g., Jansen et al. 2002; Glandt 2005; Elmsallati et al. 2005; Ebadi and Davies 2006).

14 SPE 113918
-10.0%
-5.0%
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
30.0%
35.0%
40.0%
-60 -40 -20 0 20 40 60 80 100 120
IREI (return generated by extra investement) [ratio]
G
A
I
N

[
%
]
Fixed
On/Off
Proportional
Loss
Profit
Low Value High Value

Figure 13. Comparison of gain vs. IREI for all downhole control strategies. Desirable values should fall in the upper-right quadrant.

The reservoir model and associated range of uncertainty used in this study are very simple. Moreover, the costs
identified in the economic analysis will not be applicable to many real developments. The actual return on investment
delivered by a given control strategy will be case-dependent, so the values we present are not of direct interest. However,
they demonstrate the flexible control offered by reactive strategies in the presence of reservoir uncertainty. It should also be
noted that we use an overall well water-cut constraint of 25%, at which the well is shut in. In the uncontrolled production
scenario, this leaves more oil in the reservoir than would be the case for a higher water-cut limit. Future studies will
investigate a higher limit.

Conclusions
We investigated three strategies to control production from a horizontal well in a thin oil rim reservoir. We found that a
passive control strategy using an FCD sized prior to installation yielded the highest returns on investment, but can also yield
negative returns and is a risky approach if the reservoir behavior cannot be predicted with certainty. This is the case in almost
all reservoir developments. However, a simple reactive control strategy, using variable ICVs adjusted in response to
downhole measurements of phase flow rates, yielded a neutral or positive return regardless of reservoir behavior.
In this control strategy, the optimal feedback loop between measurement and control is specific to the production case.
However, we used a feedback loop optimized on a base case reservoir model to control the ICVs in all production scenarios.
This suggests that a feedback control strategy that provides a sensible control solution for the base case is likely to be robust
over a range of reservoir uncertainty. However, it should be noted that a reactive control strategy using variable ICVs is not
yet commercially available. Moreover, a control strategy using on/off ICVs and zonal well tests, which can be delivered by
current technology, is too slow to react to changes in flow downhole to yield the same improvements in production.
Our results suggest that a simple, reactive control strategy can insure against reservoir uncertainty. The strategy can be
optimized on a base case reservoir model, or across an ensemble of models that reflect the uncertain reservoir description.
Because the reservoir will not behave exactly as predicted, the control strategy will not yield the optimal value of the chosen
objective function. However, unlike most control techniques, it does account for reservoir uncertainty because the control
decisions are not strongly model dependent. It is also much simpler to apply than model-based control strategies that do
incorporate reservoir uncertainty.
Closing the loop between downhole measurements and control decisions using simple reactive strategies may be one
way to make intelligent well technology more attractive to operators. Reducing the costs associated with downhole valves
and sensors will also yield higher returns. Proactive (or defensive) strategies may yield benefits if they can be incorporated in
a similar feedback loop. As yet, proactive control depends on the predictions of a reservoir model. However, downhole
reservoir imaging techniques, which can monitor fluid flow and saturation changes at a distance from the well, may be used
in a proactive feedback loop if they become more established.
SPE 113918 15
Nomenclature

Symbol Description Units
BOE Equivalent barrels of oil Barrels (bbl)
WNPV Total equivalent-oil production BOE
WOPT Total oil production (at surface conditions) BOE, STB
WGPT Total gas production (at surface conditions) MSCF
C
G
Gas conversion coefficient BOE/MSCF
WWPT Total water production (at surface conditions) STB
C
W
Water conversion coefficient BOE/STB
W
cinit
Initial well cost BOE
IREI Incremental Return (gain) generated by an Extra Investment [-]
P
c
Pressure drop across an ICV for a given choke setting Psia
P
o
Pressure drop across a fully open ICV Psia
A, B, c Coefficients in Equation 2 [-]
WOR Water/oil ratio [-]
WCT Water cut [-]

SI Metric Conversion Factors

ft x 3.048* x10
-1
= m
in. x 2.54* x10
0
= cm
psi x 6.894 757 x10
0
= kPa
bbl x 1.589 873 x10
-1
= m
3
* Conversion factor is exact.

References
Aitokhuehi, I., Durlofsky, L. J. 2005. Optimizing the performance of smart wells in complex reservoirs using continuously updated
geological models. Journal of Petroleum Science and Engineering 48 (34): 254264. doi:10.1016/j.petrol.2005.06.004
Akram, N., Hicking, S., Blythe, P., Kavanagh, P., Reijnen, P., Mathieson, D. 2001. Intelligent Well Technology in Mature Assets. Paper
SPE 71822 presented at Offshore Europe Conference, Aberdeen, Scotland, 48 Sept. doi: 10.2118/71822-MS.
Algeroy, J., Morris, A. J., Stracke, M., Auzerais, F., Bryant, I., Raghuraman, B. et al. 1999. Controlling reservoirs from afar. Oilfield
Review 11 (3): 1829.
Armstrong, A.C.C., and Jackson, M.D. 2001. Management of water breakthrough using Intelligent Well Technology. Proc., Offshore
Technology Conference, Houston, Sept. OTC013284.
Arnold, R., Burnett, D. B., Elphick, J., Feeley, T. J., Galbrun, M., Hightower, M., et al. 2004.Managing WaterFrom Waste to Resource.
Oilfield Review 16 (2): 2641.
Braithwaite, S. R., Mssig, S., Poel, R. v/d, Putten, S. van, Waal, W. v/d, Kass, M. 2004. Toucan Smart Field Development: How to
Generate More Value from Hydrocarbon Resources. Paper SPE 88649 presented at Abu Dhabi International Conference and
Exhibition, Abu Dhabi, United Arab Emirates, 1013 Oct. doi: 10.2118/88649-MS.
Brekke, K., and Lien, S. C. 1994. New, Simple Completion Methods for Horizontal Wells Improve Production Performance in High-
Permeability Thin Oil Zones. SPE Drilling & Completion 9 (3): 205209. doi: 10.2118/24762-PA.
Brouwer, D.R., Jansen, J. D., Starre, S. v/d, Kruijsdijk, C. P. van, Berentsenm C. W. 2001. Recovery Increase through Water Flooding with
Smart Well Technology. Paper SPE 68979 presented at SPE European Formation Damage Conference, The Hague, Netherlands, 21
22 May. doi: 10.2118/68979-MS.
Brouwer, D.R., and Jansen, J.D. 2004. Dynamic optimisation of water flooding with smart wells using optimal control theory. SPE Journal
9 (4): 391402. doi: 10.2118/78278-PA.
Brouwer, D. R., Naevdal, G., Jansen, J. D., Vefring, E. H., Kruijsdijk, C. P. 2004. Improved Reservoir Management through Optimal
Control and Continuous Model Updating. Paper SPE 90149 presented at SPE Annual Technical Conference and Exhibition, Houston,
Texas, 2629 Sept. doi: 10.2118/90149-MS
Bryant, I.D., Chen, M.-Y., Raghuraman, B., Raw, I., Delhomme, J.-P., Chouzenoux, C. et al. 2002. An Application of Cemented
Resistivity Arrays To Monitor Waterflooding of the Mansfield Sandstone, Indiana, U.S.A. SPE Reservoir Evaluation & Engineering 5
(6) 447454. doi: 10.2118/81752-PA.
Bryant, I.D., Chen, M.-Y., Raghuraman, B., Schroeder, R., Supp, M., Navarro, J., Raw, I., et al. 2004. Real-Time Monitoring and Control
of Water Influx to a Horizontal Well Using Advanced Completion Equipped with Permanent Sensors. SPE Drilling & Completion 19
(4): 253264. doi: 10.2118/77522-PA.
Carter, J. N., Ballester, P.J., Tavassoli, Z. et al. 2006. Our calibrated model has poor predictive value: An example from the petroleum
industry. Reliability Engineering and System Safety 91: 13731381. ISSN: 09518320.
Ebadi, F., and Davies, D. R. 2006. Should Proactive or Reactive Control Be Chosen for Intelligent Well Management? Paper SPE
99929 presented at Intelligent Energy Conference and Exhibition, Amsterdam, The Netherlands, 1113 Apr. doi: 10.2118/99929-MS.
Ebadi, F., Davies, D. R., Reynolds, M., Corbett, P. W. M. 2006. Screening of Reservoir Types for Optimisation of Intelligent Well Design.
Paper SPE 94053 presented at SPE Europec/EAGE Annual Conference, Madrid, Spain, 1316 June. doi: 10.2118/94053-MS.
16 SPE 113918
Elmsallati, S. M., and Davies, D. R. 2005. Automatic Optimisation of Infinite Variable Control Valves. Paper IPTC 10319 presented at
International Petroleum Technology Conference, Doha, Qatar, 2123 Nov.
Elmsallati, S.M., Davies, D. R., Erlandsen, S. M. 2005. A Case Study of Value Generation with Intelligent Well Technology in a High-
Productivity, Thin Oil Rim Reservoir. Paper SPE 94995 presented at the SPE Middle East Oil and Gas Show and Conference,
Bahrain, 1215 March. doi: 10.2118/94995-MS.
Gai, H. 2001. Downhole Flow Control Optimization in the Worlds 1st Extended Reach Multilateral Well at Wytch Farm. Paper SPE
67728 presented at SPE/IADC Drilling Conference, Amsterdam, Netherlands, 27 Feb1 Mar. doi:10.2118/67728-MS.
Glandt, C. A. 2005. Reservoir Management Employing Smart Wells: A Review. SPE Drilling & Completion 20 (4): 281288. doi:
10.2118/81107-PA.
Holmes, J. A., Barkve, T. and Lund, O. 1998. Application of a Multisegment Well Model to Simulate Flow in Advanced Wells. Paper SPE
50646 presented in SPE European Petroleum Conference, The Hague, Netherlands, 2022 October. doi:10.2118/50646-MS.
Holmes, J. A. 2001. Modelling Advanced Wells in Reservoir Simulation. Journal of Petroleum Technology 53 (11): 5460, 66. doi:
10.2118/72493-MS.
Jackson, M. D., Saunders, J. H., Addiego-Guevara, E. A. 2005. Development and application of new downhole technology to detect water
encroachment towards intelligent wells. Paper SPE 97063 presented at the SPE Annual Technical Conference and Exhibition, Dallas,
Texas, U.S.A., 912 Oct. doi: 10.2118/97063-MS.
Jackson-Nielsen, V. B., Tips, T. R. 2003. Case Study: First Intelligent Completion System Installed in the Gulf of Mexico. SPE Drilling &
Completion 18 (1): 5058. doi: 10.2118/81928-PA.
Jansen, J.D., Wagenvoort, A. M., Droppert, V. S., Daling, R., Glandt, C. A. 2002. Smart Well Solutions for Thin Oil Rims: Inflow
Switching and the Smart Stinger Completion. Paper SPE 77942 presented at the SPE Asia Pacific Oil and Gas Conference and
Exhibition, Melbourne, Australia, 810 Oct. doi: 10.2118/77942-MS.
Kharghoria, A., Zhang, F., Li, R., Jalali, Y. 2002. Application of Distributed Electrical Measurements and Inflow Control in Horizontal
Wells under Bottom-Water Drive. Paper SPE 78275 presented at the European Petroleum Conference, Aberdeen, United Kingdom,
2931 Oct. doi: 10.2118/78275-MS.
Kragas, T.K., Bostick, F. X., Mayeu, C., Gysling, D. L., Spek, A. M. 2003a. Downhole Fiber-Optic Flowmeter: Design, Operating
Principle, Testing, and Field Installations. SPE Production and Facilities 18 (4): 257268. doi: 10.2118/87086-PA.
Kragas, T. K., Johansen, E. S., Hassanali, H., Costa, S. L. 2003b. Installation and Data Analysis of a Downhole, Fiber Optic Flowmeter at
Mahogany Field, Offshore Trinidad. Paper SPE 81018 presented at SPE Latin American and Caribbean Petroleum Engineering
Conference, Port-of-Spain, Trinidad and Tobago, 2730 Apr. doi:10.2118/81018-MS.
Naus, M. M., Dolle, N., and Jansen, J. D. 2006. Optimization of Commingled Production Using Infinitely Variable Inflow Control Valves.
SPE Production & Operations 21 (2): 293301. doi: 10.2118/90959-PA.
Paino, W. F., Tengah, N.H., Woodward, M. I., Snaith, N., Salleh, H., Browne, M., 2004. Using Intelligent Well Technology to Define
Reservoir Characterization and Reduce Uncertainty. Paper SPE 88533 presented at the SPE Asia Pacific Oil and Gas Conference and
Exhibition, Perth, Australia, October. doi:10.2118/88533-MS.
Permadi, P., Wibowo, W., Alamsyah, Y., Pratomo, S. W. 1997. Horizontal Well Completion with Stinger for Reducing Water Coning
Problems. Paper SPE 37464 presented at the SPE Production Operations Symposium, Oklahoma City, Oklahoma, U.S.A., 911
March. doi: 10.2118/37464-MS.
Poel, R. v/d, and Jansen, J. D. 2004. Probabilistic analysis of the value of a smart well for sequential production of a stacked reservoir.
Journal of Petroleum Science and Engineering, 44 (12): 155172. doi:10.1016/j.petrol.2004.02.012.
Raghuraman, B., Cout, B., Savundararaj, P., Bailey, W. J., Wilkinson, D. J. 2003. Valuation of Technology and Information for Reservoir
Risk Management. SPE Reservoir Evaluation & Engineering 6 (5): 307316. doi: 10.2118/86568-PA
Robison, C. E. 1997. Overcoming the Challenges Associated With the Life-Cycle Management of Multilateral Wells: Assessing Moves
Towards the "Intelligent Completion. Paper SPE 38497 presented at the Offshore Europe conference, Aberdeen, UK, 912 Sept. doi:
10.2118/38497-MS.
Sarma, P., Aziz, K., Durlofsky, L. J. 2005a. Implementation of Adjoint Solution for Optimal Control of Smart Wells. Paper SPE 92864
presented at SPE Reservoir Simulation Symposium, The Woodlands, Texas, 31 Jan2 Feb. doi: 10.2118/92864-MS
Sarma, P., Durlofsky, L. J., Aziz, K. 2005b. Efficient Closed-Loop Production Optimization under Uncertainty. Paper SPE 94241 presented
at SPE Europec/EAGE Annual Conference, Madrid, Spain, 1316 Jun. doi: 10.2118/94241-MS
Saunders, J. H., Jackson, M. D., Pain, C. 2006. A new numerical model of electrokinetic potential response during hydrocarbon recovery.
Geophys. Res. Lett. 33. L15316. doi:10.1029/2006GL026835.
Schlumberger. 2007. ECLIPSE Reference Manual 2007.1.
Sinha, S., Kumar, R., Vega, L., Jalali, Y. 2001. Flow Equilibration towards Horizontal Wells Using Downhole Valves. Paper SPE 68635
presented at the SPE Asia Pacific Oil and Gas Conference and Exhibition, Jakarta, Indonesia, 1719 April. doi: 10.2118/68635-MS.
Tavassoli, Z, Carter, JN, King, PR. 2004. Errors in history matching. SPE Journal 9 (3): 352361. doi: 10.2118/86883-PA.
Webster, M., Richardson, S., Gabard-Cuoq, C., Fitzgerald, J. B., Stephenson, K. E. 2006. Well Surveillance with a Permanent Downhole
Multiphase Flowmeter. SPE Production & Operations 21 (3): 388393. doi: 10.2118/90024-PA.
Williamson, J. R. Bouldin, B., Purkis, D. 2000. An Infinitely Variable Choke for Multi-Zone Intelligent Well Completions. Paper SPE
64280 presented at the SPE Asia Pacific Oil and Gas Conference and Exhibition, Brisbane, Australia, 1618 Oct. doi:10.2118/64280.
Yeten, B., Jalali, Y. 2001. Effectiveness of Intelligent Completions in a Multiwell Development Context. Paper SPE 68077 presented at the
SPE Middle East Oil Show, Bahrain, 1720 March. doi: 10.2118/68077-MS.
Yeten, B. Durlofsky, L.J. Aziz, K. 2002. Optimization of Smart Well Control. Paper SPE 79031 presented at the SPE International Thermal
Operations and Heavy Oil Symposium and International Horizontal Well Technology Conference, Calgary, Alberta, Canada, 47
Nov. doi:10.2118/79031-MS.
Yeten, B. Brouwer, D.R. Durlofsky, L.J., and Aziz, K. 2004. Decision analysis under uncertainty for smart well deployment. Journal of
Petroleum Science and Engineering, 43 (34): 183199. doi: 10.1016/j.petrol.2004.02.013.