You are on page 1of 22

SPE-200746-MS

A Rapid, Simple, Portable Tool to Design and Analyse the Value of Inflow
Control Devices ICD and Autonomous Inflow Control Devices AICD

Bona Prakasa, Gina Corona, and Clifford Allen, Halliburton

Copyright 2020, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Norway Subsurface Conference originally scheduled to be held in Bergen, Norway, 22 April 2020. Due to COVID-19
the physical event was postponed until 2 - 3 November 2020 and was changed to a virtual event. The official proceedings were published online on 2 November 2020.

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
The criticality and a means of modelling latest generation wells employing Inflow Control Device (ICD) and
Autonomous Inflow Control Device (AICD) completions providing solutions to the production constraints
encountered in multi-reservoir contact wells is presented in this paper. ICD-AICD wells enable a better-
distributed production/injection from/into different reservoir zones/layers by adding a passive and or
autonomous downhole restriction. ICD-AICD technology thereby enables more efficient field development
by improving sweep efficiency, potentially reducing the number of production/injection wells required and
maintaining well longevity and profitability.
Modelling the completion interaction with the reservoir is critical to determine the incremental value
of applying such technology. This being achieved by coupling the wellbore model, which includes the
lower completion, to the reservoir properties. The modelling technique, a grid-based numerical well/
reservoir simulation, necessitates knowledgeable personnel and detailed, sometimes uncertain, geological
data. The multiple model iterations required when using ICD-AICD technology further increases the
reservoir simulators' computing time.
A new tool was developed to design and analyse ICD-AICD efficiency, bridging between a simple steady-
state inflow snapshot evaluation and a full dynamic reservoir simulation. The new model uses the proven
classical method of fractional flow, combined with the performance of specific flow control devices installed
in the well. This allows fast forecasting of the production profile and oil recovery, as well as optimizing
ICD-AICD configuration at the well design stage.
The proposed workflow is implemented as either a production forecast or a diagnostic tool for an ICD-
AICD well in an immiscible gas or aquifer-supported oil field. It provides one missing link between
today's ICD-AICD design workflows and the long-term value evaluation of a specific design when a full
reservoir simulation is either not available or too time consuming. The method's transparency and ease of
implementation of its algorithms can make it a useful tool for well and reservoir engineers to design and
analyse the value of ICD-AICD devices.
2 SPE-200746-MS

Introduction
A large amount of oil is being produced from heterogeneous reservoirs by the means of waterflooding.
Carefully designed flood pattern and well placement, as well as well control, are required to delay water
breakthrough, optimise sweep efficiency, and maximize oil recovery.

• Waterflood analysis and forecast models, e.g. Dykstra-Parsons (DP) (Dykstra and Parsons 1950)
method, or Buckley-Leverett (BL) (Buckley and Leverett 1942) are used routinely to aid such
design.
• Inflow Control Devices or Autonomous Inflow Control Devices (ICD-AICDs), i.e. completions
capable of controlling fluid flow at the reservoir sandface, are a proven and widely used (thousands
of wells) approach to modify production/injection profiles as well as to react to water breakthrough
in order to improve waterflood performance.
Incorporation of ICD-AICD performance into the waterflood analysis models will allow accurate forecast
and informed design and control of the wells in order to maximise oil recovery.
We first discuss the Advanced Well Completions (AWCs): their history, variety, impact on well
performance, application envelope, and design/control methods. We then review the latest model being
developed and finally explain how the AWCs' performance is incorporated into the new model.
AWCs have Flow Control Devices (FCDs) installed in the production tubing in front of the production
or injection intervals. FCDs (one or several) can be installed as frequently as on every tubing joint, often
amounting to hundreds per well. Some sort of annular flow isolation (e.g. gravel pack or packers) is also
preferable. An example schematic view of an open-hole production well with packers is shown in Fig. 1,
though AWCs can also be installed in cased-hole wells, injection wells, multi-lateral wells, etc.

Figure 1—Schematic view of a well with AWC

The principle of operation of AWCs is simple: the pressure drop across an FCD is non-linearly dependent
of the flowing fluid rate (see Eq. 1), while the pressure drop across the reservoir is mostly linear. This
allows the higher-producing zones (here, by zones, we mean AWC intervals between adjacent packers) to
experience higher pressure drop across the AWC, and vice versa, resulting in a more uniform inflow/outflow
profile. FCDs can be generally classified as Passive (represent a fixed restriction), Active (the restriction
can be controlled), and Autonomous (cannot be externally controlled but autonomously react to unwanted
fluid). A good overview of the AWC technology evolution, types, and applications can be found in (Eltaher
et al., 2014, Al-Khelaiwi et al., 2008).
AWCs are well suited for wells producing from different quality zones. This can mean heterogeneous
reservoirs, differentially depleted layers, compartmentalized reservoirs, multiple-reservoir developments,
unfavourably saturated (e.g. oil-rim) reservoirs, etc. The wells with AWC are often horizontal, deviated, or
SPE-200746-MS 3

multi-lateral – the longer the well reservoir contact, the bigger the heterogeneity-related problems. Also, the
installation of Passive AWC in homogeneous reservoirs to reduce the heel-to-toe effect is not uncommon.
In fact, Passive Inflow Control Technology actually began with the development of the Inflow Control
Device (ICD) in the early 1990s with the idea to control inflow in order to minimise the heel-to-toe effect
in a long horizontal well producing from the Troll field with a massive, homogeneous, oil-rim reservoir
offshore Norway (Henriksen et al. 2006). The ICD market has expanded in the last decades with multiple
installations worldwide in the North Sea, Middle East, Asia, Latin America, and North America in both
land and deepwater wells. Passive AWC comprises a range of ICD types different in design and in how
tolerant they are to erosion and flowing fluid properties, yet in a nutshell, every ICD is a fixed restriction
device whose performance is described by
(1)
in which the ‘bed’ parameter a relates the extra pressure drop ΔPICD added by an ICD restriction to the flow
rate Q squared.
More information on the ICD types and the exact formulae used to find strength a for each ICD type
can be found in (Al-Khelaiwi, 2013).
Active FCDs, a.k.a. ICVs (Interval Control Valves), vary in design depending on how they are controlled
(e.g. electric, hydraulic control lines), how many restrictive positions an ICV can have (normally around
4 to 10), etc. Either way, the pressure drop across an ICV for a given position is still described by Eq. 1.
Setting the ICVs for different zones to restrictive positions in order to balance the inflow/outflow is a good
control option (proactive control) resembling ICDs, although it is not uncommon to control ICVs by setting
to the fully open position initially, and shutting upon the onset of high water production from the zone in
question (reactive control). This technology may (to some extent) substitute for the workover operation, yet
and so is particularly popular in subsea wells. More information on the ICV types, applications, and control
can be found in (Haghighat Sefat et al., 2016).
The recently introduced Autonomous Inflow Control Devices (AICDs) react to "unwanted" (in oil
production) fluid phases (i.e. free gas and water) restricting their flow in-situ and improving recovery.
The AICD performance is described separately for single-phase oil flow and for single-phase water or gas
flows, since these performances differ significantly, while an empirical or theoretical formula is suggested to
predict the performance at mixed flow concentrations. Several types of Autonomous Inflow Control Devices
(AICDs) are available and have been successfully applied in production wells, though their modelling and
performance understanding is still under research. (In this work, this problem is set aside because we only
assume single-phase flow situations – either oil or water – as will be explained later.) More on the AICDs
can be found in Eltaher et al. (2014). For a given fluid phase, the AICD performance can be acceptably
matched to Eq. 1.
The AWC design process requires specification of the FCD types, restriction levels, installation
frequency, as well as specification of the annular flow isolation to maximise the value added by the AWC
while minimising its cost per extra barrel produced, risk of installation, and operational problems. AWC
design is a complex mathematical problem that so far has been approached in different ways:

• Ideally, the AWC value added to the whole field production should be evaluated, since AWC
action in each well has a long-term impact on the field recovery. This implies a multi-dimensional
optimisation problem using a numerical reservoir simulation tool to forecast the added value for a
given AWC design. This task is challenging and can be approached either directly or sequentially:
e.g., the well placement is optimized, then the annular flow isolation for an arbitrary AWC is
identified, and finally the AWC is optimized. To our knowledge, no examples of such optimisation
are yet published.
4 SPE-200746-MS

• The common approach (a.k.a. the snapshot design) is to design the AWC to at least improve, not
necessarily optimise, the oil recovery, using the wellbore model only. This simplifies the problem
and allows use of a stand-along wellbore simulator that is much faster than the fully coupled
wellbore-reservoir simulators. The reservoir inflow/outflow performance is somehow estimated
at one specific time, often at the start of well production, and feeds into the wellbore model. The
wellbore model is then used to vary the AWC specifications to reduce the flow imbalance between
zones while not imposing too high a restriction. This results in a more uniform waterflood front
and is believed to positively impact the oil recovery, though the rigorous reservoir simulation to
test this is often ignored. A range of such snapshot methods have been developed: analytical and
semi-analytical methods to reduce the heel-to-toe effect in a homogeneous reservoir (Birchenko
et al., 2010) or to reduce the heterogeneous inflow imbalance (Birchenko et al., 2011); the type-
curve method for ICD completion design (Prakasa et al., 2019); numerical wellbore model-based
designs (Al-Khelaiwi, 2013); etc.
Needless to say, such methods lack the capability to predict the impact of the AWC design on the actual
waterflood development: oil recovery efficiency, fractional flow rate profile, etc. This paper offers a solution
to this limitation.

Review of Fractional Flow Analysis for Advanced Well Completion


The following section is explained in more detail in Prakasa, et al. 2019
The Buckley-Leverett (BL) (Buckley and Leverett 1942) method describes two-phase, immiscible
displacement in a linear system. Eq. 2, the fractional flow equation, leads to Eq. 3, which calculates the
location (x) of a given water saturation (S_w), with the following assumptions: capillary pressure effects
can be neglected, no dip, thus gravity forces can be neglected, homogeneous and isotropic porous medium.

(2)

(3)

fw is fractional flow or the ratio of the flow of water at any point, A is cross-section area, k is permeability,
kro and krw are relative permeability of oil and water, respectively, qt is total rate, μo and μw are viscosity of
oil and water, respectively, Pc is capillary pressure, t is time, Φ is porosity, and Sw is water saturation.
Welge (1952) evaluated the average saturation behind the water front by drawing a tangent to the fw
curve originating at the initial water saturation (Swi)(Fig. 2). Point A, the intersection, is Sw at the flood
front and the reciprocal of the slope is the water front's velocity. The average saturations behind this front
are found by extrapolating the tangent fw=1.
SPE-200746-MS 5

Figure 2—An example of fractional flow curves analysis

Welge determined the average water saturation at breakthrough , point B in Fig. 2, by integrating
the water saturation over the distance from the point of injection to the value of the saturation at the flood
front. The calculation continues post-breakthrough (Eq. 4) with the tangent line to fw=1 being drawn to find
the average water saturation. The calculation stops once the layer is fully saturated with water, i.e. when
(Sw=1-Sor).

(4)

Later, the Welge analysis was developed to determine the oil recovery and the water fraction as a function
of cumulative volume of water injected for non-piston-like displacement. The Welge method calculates the
formation's water saturation ( ) behind the flood front assuming the oil is homogeneously displaced with
water. The formation behind this flood front has a "pseudo" relative-permeability, and , based on
the average water saturation, . Ahmed (2001), Craig (1993), and Dake (2001) all proposed predicting the
front's displacement by finding the Kr values by direct correlation from the value of based on the actual
relative-permeability curve. This assumes both piston-like displacement and linear relative permeability
curves.
Fig. 3 illustrates waterflooding a layer's formation volume between the producer and the injector. P1
refers to pressure at the injection well's location (x0) or at the start time of the displacement process. Pf
is the pressure at xf, the tip of the saturation front where water is displacing oil. The difference between
pressure P1 and Pf is ΔPw. Mobile oil is present in the formation between the water front and the producer
(L- xf). ΔPO is the pressure difference between the displacement front and the production well's annulus and
ΔPFCD (or FCC)) is the pressure difference across the FCDs. Fig. 3 indicates that FCDs are only installed
in the producer and an extra pressure drop, ΔPFCDi, must be added to ΔPw if FCDs are also installed in the
injection well.
6 SPE-200746-MS

Figure 3—The formation volume behind the flood front is divided into multiple blocks

The water saturation value (Sw) of the formation between the injector and the water front gradually
decreases from (1 – Sor) at the injector to Swf at water front (Fig. 3). This can be modelled by a series of
blocks with appropriate water saturation and relative permeability values (Fig. 4) starting from x0 to n1 (=
Δn1) and continuing to block N and ΔnN (or xf). (P1–Pf), the pressure difference between x0 and xf, is equal
to the sum of the pressure drops across all the blocks between x0 and xf (Fig. 4 and Eq. 5).

(5)

Figure 4—Several blocks with different relative permeability values are created behind the water front.

As displayed in Fig. 4, the region between x0 and xf is consistent with unique Krw and Kro for each block.
Our proposed method treated this region as one block with an averaged relative permeability value for each
phase, krwavg and kroavg. Thus, neglecting capillary pressure:

(6)

The average oil and water relative permeability behind the water front are:

(7)
SPE-200746-MS 7

(8)

The mobility value of the displacing phase behind the flood front together with the system mobility ratio
(Eqs. 9 and 10) is now calculated:

(9)

(10)

The BL method determines both the position, xn, and its saturation value, Swn. Dake (2001) explains
that the position of any water saturation is proportional to the derivative of fractional flow over the water
saturation (Eq. 11), allowing the position of each saturation block relative to the front (Eq. 12) to be
determined.

(11)

(12)

This approach, called the Extended Welge (EW) method, was an extension of extended DP and BL
Methods (Muradov et al. 2018, Prakasa et al. 2019) and hence the fractional flow analysis is possible by
adding the FCD's non-linear pressure drop to the system (Eq. 1).
The fluid velocity in each layer is:

(13)

Where coefficients B and C are:


(14)

(15)

Note that one of the main differences with the extended DP method's equation is the value of the mobility
ratio of the displacing phase, λ(avg, diplacing phase).

Proposed Model to Include Gravity and Friction


So far, the presented technique of rapid modelling has only been applied to vertical wells in horizontal,
non-communicating reservoirs. In these cases, the distance between injectors and producers is relatively
long and the Darcy law entirely governs the fluids' flow without considering the effect of gravity. On the
other hand, many of the passive FCC completions are installed in horizontal wells in reservoirs where the
distance to the aquifer or gas cap is relatively short, and the displacing fluids are moving in the vertical
direction, i.e. the role of gravity can no longer be neglected. Furthermore, the previous methods assumed
the friction pressure drop along the well length is negligible, i.e. an infinitely conductive well. This is an
acceptable assumption for a vertical well or a short horizontal well with a small flow rate. However, many
wells today have extensive horizontal sections and operate with a large flowing rate, i.e. the role of friction
can no longer be neglected.
8 SPE-200746-MS

Figure 5—Illustration of vertical displacement around a horizontal well (Vela, 2011)

This section describes the implementation of the previously described, modified BL methods for an
AWC installed in perfectly horizontal well. The reservoir-well-aquifer interaction replicates a fractional
flow displacement model of an aquifer, which acts as the injector well. Thus, the standoff (H) between the
aquifer to the horizontal well is equivalent to the previously modelled inter-well distance (l). The previously
derived equations are extended to include the role of gravity force and is incorporated into the solutions. The
remaining assumptions are no capillary pressure, ΔPo=ΔPw, volumetric replacement, or steady-state flow.
Consider the vertical displacement depicted in Fig. 5:
(16)
Similarly, to the horizontal displacement case, the system of pressure equations can be discretised into
the pressure drop in water-phase, the pressure drop in oil-phase, and the pressure drop in the flow control
completion.
(17)
The first and the second components in the right side of this equation are expanded based on the Darcy
law, and the pressure drop in the completion (third component) is substituted based on Eq. 1.

(18)

Note that this equation is similar to that explained by the previous section, see Fig. 3. The main difference
is that now there is a correction for the gravity term, "γ".
Finally, such a displacement system can expressed as a quadratic equation (Eq. 19):

(19)

Where coefficients B, C, and D are defined as:


(20)

(21)

(22)

Where "γ"o and "γ" mix are the specific gravity for oil and mixed fluids behind the front, and
SPE-200746-MS 9

Figure 6—1D column vertical layer displacement

For more a detailed explanation of incorporating gravity into AWC modelling, refer to Prakasa, 2018.
To include friction pressure-drop, we need to overcome the challenge that the frictional pressure drop is
sensitive to the change of the fluids properties dynamics along the well. Various models have been developed
to accurately predict this phenomenon; however, they are mostly assuming constant phase of oil, i.e. oil
is always in the liquid phase. In our methods, we leverage a series of nodal analyses followed up by data
analytics. Although this is pragmatic, it is relatively more practical while preserving the accuracy of the
prediction.
It is started by building the well completion schematic in a nodal-analysis platform similar to a standard
nodal-analysis routine. In this process, however, there are multiple reservoir/zones. Different parameters
are then set for each producing reservoir zone. Thereafter, traverse curves are composed and the data points
for each zones are documented with the associated commingled zones. This way, for each physical state of
the zones, the associated pressure drops (including friction) are available, thus maintaining the accuracy of
the model. The friction pressure drop is later predicted through a series of interpolation. Our proprietary
calculators automated this process and a workflow can be exercised in a fraction of a second.
10 SPE-200746-MS

Figure 7—Nodal analysis and data analytics to predict friction pressure drop

Workflow
Note that this model assumed steady-state conditions, thus the injection rate is equivalent to total production
rate. In this section, injection rate is used to maintain consistency against the fractional flow's terminology.
This injection practically equals the total production.

Model Initialisation
Fractional flow analysis is carried out using relative permeability values to calculate, and draw the fw and
curves, find the value for the flood front saturation, and the average water saturation (Fig. 2, Eqs. 2
and 3). The saturation front (Swf and average water saturation ( ) behind the front are constant prior to
breakthrough. Kroavg and Krwavg of the displacing front will also be constant during this period.
Fractional flow analysis is applied to all zones/layers. For each layer, we calculate an average front's
velocity, and the layers are sorted based on the required time for water to arrive at the other end, i.e. time
before breakthrough. The layer with the earliest time before BT will be the first layer and reference (Layer
R).

Time period between the start of injection and water breakthrough in Layer R
The evaluation is for every incremental x starting from the x0 (the location of injector) to L (the location
of the producer). For example, an injection well and a production well are 1000 m apart and evaluated
at 10 m intervals, thus x0 represents the injection well's location, Δx=10 m, L =1000 m, and hence x100 is
the producer's location. Note that this Δx grid is different from the N blocks in Fig. 4 that represented the
multiphase fluid properties. Strictly speaking, there are N blocks each time the front advances by Δx with
the incremental cumulative water injection (ΔWIk being evaluated using Eq. 23 (i.e. fractional flow analysis
is executed for every xn.

(23)

Averaging the injection rate between Ix-1and Ix gives the incremental time (Eq. 24):

(24)

Eq. 10 calculates the mixed fluids' mobility ratio; the injection rate is given by Eqs. 19 – 22, and the
total injection time by Eq. 25:
SPE-200746-MS 11

(25)
The above step is repeated for each layer of the well's completion. Layer R, the reference layer, is the first
layer to experience breakthrough and the remaining layers are ordered by increasing breakthrough time. The
time required for the water to advance by Δx in Layer R (Eq. 26) is also used to calculate the distance(Δxj,x)
the water advances in the other layers (Layer J). This distance is approximated by replacing the right side
of Eq. 26 with Eqs. 23 and 24.
(26)

(27)

The first two parameters on the right-hand side of Eq. 27 are constant before breakthrough

at Swfj, hence only Δx needs be calculated. The injection rate (I) is also a function of Δx, and a simple non-
linear solver (e.g. Excel, Matlab) can determine this in a fraction of a second.
Repeat this step until Layer R experiences breakthrough. Flow-in and flow-out of the system are equal
(i.e. Qo=I) during this time period. No water is produced and Qw=0.

Time period after initial water breakthrough until all completion layers are producing water
The second layer to experience breakthrough is now designated as Layer R and the previous steps are
repeated until water breaks through into all remaining layers. The post-water breakthrough into those layers
that have experienced water breakthrough (Layer K) is calculated by evaluating every incremental Sw starting
from the initial flood front saturation, Swfk, until oil production stops, i.e. Sw=(1-SOR. For example, Layer
K experiences an initial water front saturation of 0.45 and has a residual oil saturation of 0.2. The oil and
water production rates are calculated by Eqs. 28, 29 and IK=(Qo+QwK). This is repeated in increments of
ΔSw=0.05 from Swk,1=0.45 to Swk,8=0.80, after which Layer K produces 100% water.

(28)

(29)
Eq. 30 provides the time required to increase ΔSw by 0.05 in Layer R. This time is also used to calculate
the increase in water saturation in the other layers (Layer K).
(30)
Layer K's increase in the water saturation (ΔSk,s) is approximated by replacing the right-hand side of Eq.
30 with Eqs. 23 and 24.

(31)

Water has already broken through in Layer K. Hence, Eq. 31 is now necessary to determine the first two
parameters on the right-hand-side of Eq. 31, , both of which are function of Sw. A non-

linear solver is required to solve for the value for Swk,s and also provides the change in the mixed fluid's
mobility ratio and the injection rate (Eqs. 19 – 22).
12 SPE-200746-MS

Time period after all layers have experienced water breakthrough


All layers are designated Layer k apart from Layer R, the last layer to experience breakthrough. The previous
workflow continues and is repeated until the well's oil production ceases and Sw=(1-SOR in all layers (Eq. 32).

(32)

Summarising each layer's performance: cumulative oil produced, Np (Eq. 33), cumulative water produced,
Wp (Eq. 34), cumulative water injected, Winj (Eq. 35) at any time.

(33)

(34)

(35)

Extending the Workflow for Constant Well Production


The workflow for constant-rate well production requires approximating the layer rates prior to the
breakthrough at an arbitrary pressure drop by calculating the flow rates at every x (Eqs. 19 – 22). Their
sum provides the total well rate. The pressure drop required for the chosen well rate is found as previously
with an optimisation routine. This process is repeated for each Δx prior to breakthrough, after which it is
repeated for each ΔSw increment.

Verification of the Horizontal Well Model in a Heterogeneous, Box-shaped


Reservoir Model
The horizontal well is completed in a stratified heterogeneous reservoir, with communicating layers. A
large aquifer underlying the reservoir supports the reservoir pressure and preserves steady-state flow
during production, and the oil recovery is driven by bottom-up water displacement. However, reservoir
heterogeneity results in inefficient recovery as the water flows faster in the more permeable layers, and
consequently, the layers with lower permeability will be unswept. This model assumed friction is negligible
along the well.
The horizontal well completion contains 50 reservoir layers with log-normal permeability distribution
(Table 1). The size of the layers is 50m × 2m, and the horizontal well is segmented across each of these
layers, i.e. segment length = 50 m, (the small width of 2m is chosen to ensure 100% areal sweep efficiency).
The stand-off (h, the distance between the well and the aquifer) is 70 m. The same Brook-Corey medium
oil relative permeability correlations are used.

Table 1—Reservoir properties along the horizontal well completion

Parameters K Width Length Area Φ μw μo Swi Sor kwe koe no nw

Layers mD m m m2 cP cP

12
1 - 50 (min) 80 2 50 100 0.25 1 4 0.2 0.1 0.5 1 2 3
(max)

Each reservoir layer resembles the event in the system described by Fig. 6 and is treated as an individual
displacement. The previously explained workflow is applied to this scenario.
SPE-200746-MS 13

The well is produced with BHFP 200 bar, and the aquifer pressure is 243 bars, making a 43-bar
drawdown. The flow variation along a screen completion is clearly seen as illustrated by the height of water
encroachment in Fig. 8. The water has invaded the more permeable layers, leaving the lower permeability
layers unswept. An advanced completion, such as ICD or AICD is required to improve the sweep efficiency
of this well. Completions with Table 2 strengths were evaluated.

Figure 8—Illustration of horizontal well swept by vertical water displacement from the aquifer. Top
figure is the permeability distribution and bottom figure is the water saturation distribution. Each layer
is treated as an individual system and analysed with the modified BL method for an AWC completion.

Table 2—Completion properties for the box-shaped model, for non-pistonlike displacement

maximum (single-phase)

Case name bar/(rcm/d)2 bar/(rcm/d)2

Screen 0 (open-hole completion) 0 (open-hole completion)

ICD 0.008 (medium strength FCC) 0.008 (medium strength FCC)

AICD/AFCD 0.008 (medium strength FCC) 0.016 (high strength FCC)

Figs. 9 and 10 show the oil and water production rates, respectively, for these completions. The analytical
results showed an excellent agreement with the numerical results from the reservoir simulation. For this
example, Fig. 9 shows that installing flow control devices provides similar cumulative oil production after
500 days. As observed in Fig. 10, the AFCD's cumulative water production was reduced compared to the
screen (Stand Alone Screens) completion.
14 SPE-200746-MS

Figure 9—Comparison of horizontal well's cumulative oil production for the screen, ICD, and AFCD completion.

Figure 10—Comparison of horizontal well's cumulative water production for screen, ICD, and AFCD completion.

A scenario from a real-case field


We also tested our model in a modified real-field scenario with a long horizontal well completed in an
onshore carbonate reservoir, with a 1500m-long reservoir section, compartmentalized into 10 zones with
uniform 150m interval lengths. To improve its initial low productivity, the well would be stimulated
with multi-stage hydraulic fracture (MSF) treatment. Such operation would create a series of uncertain
fracture networks that exacerbate the heterogeneity of the payzone. From previous knowledge, a log-normal
distributed PI is expected, ranging from (P10) 1 to (P90) 350 - m3/day/bar, shown in Fig 11. Through these
prominent fractures or faults, unwanted water production would freely flow from the aquifer, bypassing the
oil influx. Hence, high WC% is anticipated, which may cause issues for surface facilities and result in choke
back or shut-in wells. The standoff distance between the well and Water Oil Contact (WOC) is expected
to be about 70m, uniform porosity of ∼10%, and initial reservoir pressure of 360 bar. Brooks and Corey's
SPE-200746-MS 15

(1966) relative permeability curves are assumed to allow explicit calculation of the fw and fw' values (see
Prakasa, et al. 2019 for a detailed explanation) with fluids' properties displayed in Fig. 12.

Figure 11—PI distribution

Figure 12—Fluids Properties

Autonomous Inflow Control Devices are installed to improve the flow distribution and to choke back the
unwanted fluids in a well's later life. Fluidic Diode-AICD is a suitable device for this kind of operation since
it would allow injection operations (such as MSF stimulation) without adding any additional equipment or
features into the device.
To evaluate its performance, we model the completion using routine snapshot modelling. With a target
of 1200 m3/d liquid production, the influx will be highly heterogeneous following the permeability of
the reservoir as displayed in Fig. 13. Those prominent zones are the source of the early breakthrough
encroachment.
16 SPE-200746-MS

Figure 13—Distribution Profile of PI and Permeability

These PI distribution are innate and can not be predicted in advance, i.e. the position where the high or low
conductive PI after the MSF operation is uncertain. Hence, installing uniform instead of varying strength or
settings of flow control devices per zone (i.e. the same number of AICD modules per zone for all 15 zones) is
more appropriate to mitigate these uncertainties while still creating compartments. We performed a strength
sensitivity analysis from more to less restrictive to optimize the well performance with the inclusion of the
AICD completion. For this scenario, we used fluidic diode Range 3 AICD (Least, et al. 2012).
For the initial time step where the reservoir is purely oil, the improvement of well performance by
installing AICD completion can be measured by having more uniformity (i.e. in this phase AICD is operating
as ICD). As displayed in Fig. 14 (bottom picture), installing a stronger restriction will improve distribution.
The previously high influx observed in the prominent zone 6 screen completion is gradually restricted by
having an increased choke (fewer number of AICD modules per zone) until it balances with the previously
low contributing zones (e.g. zone 4). Such uniformity could be, however, at the expense of lower PI. The
additional pressure drop from having a stronger restriction will lower the overall flowing bottom hole
pressure. The optimum completion will improve the flow distribution while maintaining the minimum
tubing head pressure per operator objectives and well/network capabilities and also operates within the
ratings of the AICD modules.

Figure 14—Distribution of PI and Permeability

Conventionally, the subsequent analysis will be to evaluate the performance of the AICD completion in
the presence of unwanted fluids in the later life of the well. The fluid saturation of each zone will be modified
SPE-200746-MS 17

with different number of water cut or gas oil ratio percentage values, often treated as either proportional
with PI or the influx ratio of each zone.
In this scenario, since the zones are saturated with some water, the AICD will react with more flow
restriction compared to the initial restriction during oil production. Note that the restriction would not
behave homogenously since this additional restriction is also a function of the fluid properties and the phase-
composition of the flow. The limitation of doing such an exercise is that the non-linear pressure drop in the
device will alter the system pressure, and fluids would move in a different encroachment behavior. This
phenomenon would render fluids phase-composition distribution in the later time of the well as uncertain,
and hence the associated AICD's (or any other flow control device) dynamic evaluation should be treated
as a low-confidence model in the presence of so many uncertainties. We applied our proposed model in
this scenario with a similar setup. We ran sensitivity of AICD restriction (i.e. AICD modules per zone) and
evaluated the performance accordingly. This model, however, allowed evaluation of the performance of the
completion in long-term parameters. The data points curve-fitted with coefficients for Eq. 1 are 0.188 and
0.687, respectively, for oil and water, as shown in Fig. 15.

Figure 15—Curve fitting of AICD performance

The same process was repeated for 3, 6, 9, 12, 15 AICD modules with coefficients tabulated in Table 3.

Table 3—Coefficient of different modules

aoil awater

3 Modules 0.0209 0.0764

6 Modules 0.0052 0.0191

9 Modules 0.0023 0.0085

12 Modules 0.0013 0.0048

15 Modules 0.0008 0.0031

The results for these simulations are displayed in Figs. 16 to 18. The significant improvement is
clearly seen when comparing flow control completions and screen completions. The improvement of RF at
breakthrough is the consequence of having delayed the breakthrough while maintaining the oil production
rate before BT. Increasing the restriction could further improve the RF; the selection will, however, need
further iterations and the inclusion of economic analysis. Since rate results as function of time are now
available, project economic rankings are now possible.
18 SPE-200746-MS

Figure 16—Fractional recovery comparison at breakthrough

Figure 17—Comparison of different breakthrough time for different completion


SPE-200746-MS 19

Figure 18—Comparison of different RF at 1400 fractional time days

Discussion and Conclusions


The presented workflow both models and predicts the long-term value added by FCDs. It is the bridge
between a "snapshot" inflow evaluation and comprehensive numerical reservoir simulation. The main
objective is to provide an insight into underlying physics in the reservoir in an easily understood form, by
enabling quick-look screening of completion designs with a tool that can be easily realised by most engineers
employing available spreadsheet resources. We based our work on the model that has been thoroughly
explained and verified (Prakasa, 2019), and incorporate the additional effect of gravity and friction force.
This workflow can also be used to compare the numerical simulation as a means of alternative tools for
decision-making and to complement snapshot or full-field reservoir modelling.
Incorporating a description of the AWC's performance into the waterflood analysis model allows
forecasting the downhole flow control configuration's production profile, the oil recovery, and the
economical benefits. The method proposed is particularly useful for rapidly designing and optimising a
completion that maximizes chosen criteria, e.g. oil recovery, unwanted fluid % decrease, economic value
added, etc.
Our method is particularly useful for rapidly designing and optimising a completion that maximizes a
chosen criterion, e.g. oil recovery, economic value added, etc. The complexity of existing modelling tools
ensures that this task is rarely examined in detail. The method's transparency and ease of implementation
of its algorithms can make it a useful tool for well and reservoir engineers.

Acknowledgments
The authors thank Halliburton for the support and encouragement to publish this paper.

Nomenclature
Values are in SI units and at reservoir conditions, unless otherwise stated
A - Effective area perpendicular to flow
a - Flow control completion strength defined by equation 1
b - Formation volume factor
fw - Fraction flow rate of water (watercut at downhole conditions)
20 SPE-200746-MS

h - Layer height
k - Horizontal permeability
L - Distance between wells
M - Mobility ratio
n - Exponent for modified Brooks-Corey functions
ΔP - Pressure difference
P - Pressure
Pc - Capillary pressure
S - Saturation
ΔS - Movable saturation (1-Sor-Swi)
t - Time
u - Fluid flow velocity
V - Cumulative liquid produced
x - The distance index for front advancement evaluation between injector and producer
λ - Fluid mobility (i.e. rel.perm./viscosity)
- Fractional flow derivative

Subscripts
abt - After breakthrough
avg - Average
ann - Annulus
bt - Breakthrough
bbt - Before breakthrough
e - External
f - Denote the saturation or the location of flood front
Fcc - Flow control completion
Injector - Pressure drop occur at injector well
j, k, R - Refers to Layer j, k, or R respectively
Layer - Pressure drop occur at layer
N - Number of blocks with different water saturations behind the flood front
n - The block's index (behind the front) for average relative permeability calculation
o - Oil
oh - Openhole
or - Residual oil (saturation)
Producer - Pressure drop occur at producer well
r - Relative (permeability)
s - The saturation index for postwater breakthrough evaluation
w - Water
wi - Irreducible water (saturation)
wf - Waterfront

Abbreviations
AICD Autonomous Inflow Control Device (a class of FCDs)
AFCD Autonomous Flow Control Device (a class of FCDs) = AICD
AWC Advanced Well Completion
BL Buckley-Leverett
SPE-200746-MS 21

DP Dykstra-Parsons
DCF Discounted Cash Flow
EW Extended Welge
FCC Flow Control Completion (equivalent with FCD)
FCD Flow Control Device
FOE Field Oil Recovery
FOPT Field Oil Production Total (Cumulative oil production)
FWPT Field Water Production Total (Cumulative water production)
MSF Multi Stage Fracturing
PV (reservoir) pore volume
Rcm (at) Reservoir conditions - cubic meters (units)
RE Oil Recovery Efficiency (recovery factor)
RF Recovery Factor
WI Volume of Water Injected
WOC Water Oil Contact

References
Ahmed, T. 2001, Reservoir Engineering Handbook (Second addition), Gulf Professional Pub. ISBN: 0884157709,
9780884157700.
Al-Khelaiwi, 2013, A Comprehensive Approach to the Design of Advanced Well Completions. PhD Thesis. Heriot-Watt
University.
Birchenko, V.M., Muradov, K.M., Davies, D.R., 2010a. ‘Reduction of the horizontal well's heel-toe effect with inflow
control devices’. Journal of Petroleum Science and Engineering. Elsevier B.V. 75 (1–2), 244–250.
Birchenko, V. M., A. Iu Bejan, A. V. Usnich et al 2011. Application of inflow control devices to heterogeneous reservoirs
(in Journal of Petroleum Science and Engineering 78 (2): 534–541. http://dx.doi.org/10.1016/j.petrol.2011.06.022.
Brooks, R.H. and Corey, A.T., 1966. Properties of porous media affecting fluid flow. Journal of the Irrigation and
Drainage Division, 92(2), pp.61–90.
Buckley, S. E., & Leverett, M. C. (1942, December 1). Mechanism of Fluid Displacement in Sands. Society of Petroleum
Engineers. doi:10.2118/942107-G
Craig, F. F., Jr. 1993, The Reservoir Engineering Aspects of Waterflooding, Monograph Series, Richardson, Texas, SPE,
(1993), 3.
Dake, L. 2001. The Practice of Reservoir Engineering (Revised Edition), Volume 36, ISBN: 9780080574431
Dykstra, H. and Parsons, R.L., 1950. The prediction of oil recovery by water flood. Secondary recovery of oil in the United
States, 2, pp.160–174.
Eltaher, Eltazy Mohammed Khalid, Morteza Haghighat Sefat, Khafiz Muradov et al 2014. Performance of Autonomous
Inflow Control Completion in Heavy Oil Reservoirs. Presented at the International Petroleum Technology Conference,
10-12 December, Kuala Lumpur, Malaysia. IPTC-17977-ms. https://doi.org/10.2523/IPTC-17977-MS
Haghighat Sefat, Morteza, Khafiz M. Muradov, Ahmed H. Elsheikh et al 2016. Proactive Optimization of Intelligent-
Well Production Using Stochastic Gradient-Based Algorithms. SPE Res Eval & Eng 19(02). SPE-178918-pa. http://
dx.doi.org/10.2118/178918-pa.
Henriksen, Knut Herman, Eli Iren Gule, Jody R. Augustine. 2006. Case Study: The Application of Inflow Control Devices
in the Troll Field. Presented at the SPE Europec/EAGE Annual Conference and Exhibition, 12-15 June, Vienna,
Austria. SPE-100308-MS. http://dx.doi.org/10.2118/100308-ms.
Least, B., Greci, S., Burkey, R. C., Ufford, A., & Wilemon, A. (2012, January 1). Autonomous ICD Single Phase Testing.
Society of Petroleum Engineers. doi:10.2118/160165-MS
Muradov, K. M., Prakasa, B., & Davies, D. (2018, August 1). Extension of Dykstra-Parsons Model of Stratified-Reservoir
Waterflood To Include Advanced Well Completions. Society of Petroleum Engineers. doi:10.2118/189977-PA
Muradov, K, Eltaher, E & Davies, DR 2018, ‘Reservoir simulator-friendly model of fluid-selective, downhole flow control
completion performance’ Journal of Petroleum Science and Engineering, vol. 164, pp. 140–154. DOI: 10.1016/
j.petrol.2018.01.039
Prakasa, B. (2018) ‘Novel Methods for Modelling, Design and Control of Advanced Well Completion Performance’, PhD
Thesis, Heriot-Watt University
22 SPE-200746-MS

Prakasa, B, Muradov, K & Davies, DR 2019, ‘Principles of rapid design of an Inflow Control Device Completion in
Homogeneous and Heterogeneous Reservoirs Using Type Curves’ Journal of Petroleum Science and Engineering,
vol. 176, pp. 862–879. DOI: 10.1016/j.petrol.2019.01.104
Prakasa, B, Muradov, K & Davies, DR 2019, Linear and radial flow modelling of a waterflooded, stratified,
non-communicating reservoir developed with downhole, flow control completions, vol. 182, 106340. https://
doi.org/10.1016/j.petrol.2019.106340
Vela, I., Viloria-gomez, L. A., Caicedo, R., & Porturas, F. (2011) ‘Well Production Enhancement Results with Inflow
Control Device (ICD) Completions in Horizontal Wells in Ecuador’, SPE EUROPEC/EAGE Annual Conference and
Exhibition. Vienna, Austria: Society of Petroleum Engineers.
Welge, H., "A Simplified Method for Computing Oil Recovery by Gas or Water Drive," Trans. AIME, 1952, pp. 91–98.

You might also like