SPE 122824

Analysis of Inflow Control Devices

Bernt S. Aadnoy, SPE, University of Stavanger, and Geir Hareland, SPE, University of Calgary

Copyright 2009, Society of Petroleum Engineers

This paper was prepared for presentation at the 2009 SPE Offshore Europe Oil & Gas Conference & Exhibition held in Aberdeen, UK, 8–11 September 2009.

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
Inflow Control Devices (IDCs) were initially developed to avoid water coning problems in long horizontal wells. They have
been used with success the past 15 years. There are, however, issues that needs to be resolved.

A pressure drop model of the ICD is presented herein. The physical model of the ICD consists of pressure drop equations from
the reservoir, through the screen, through the flow conduit, through the ICD nozzle and into the production tubing, and,
pressure drop through the lower completion system. Evaluation of the model shows that for current commercial tools,
turbulent flow through the ICD dominates the pressure drop, leading to a density controlled flow. This is fortunate as density
varies much less than viscosity over the production life of a field.

Due to the inherent non-linear nature of a production system, the pressure drop versus flow rate will vary with degree of
depletion. An ICD may be optimal initially, but not when the reservoir pressure is depleted. This paper also presents a new
designer IDC concept which maintains constant flow regardless of the degree of field depletion. It is based on a hydraulic
feedback principle, and ensures controlled flow throughout the life of the oil field.

The Inflow Control Problem

Figure 1 shows a typical coning problem in a long well in a thin reservoir. Because the production tubing is very long, there is
a considerable pressure drop in the tubing itself. The oil at the far end (the toe) has to overcome this pressure drop. The oil at
the heel does not see this pressure drop. The consequence is that the flow rate is higher at the heel than at the toe. The area
close to the heel will produce more liquid hydrocarbons, resulting in coning of the water-oil-contact or the gas-oil-contact.
When water or gas production starts, the process will accelerate. Over time as the water production increases, it leads to
reduction in oil production and more produced water disposal problems. Most of the oil near the toe will not be produced, and
new wells are required to drain this. The recovery aspect is of utmost importance for these types of wells.

Figure 1: Waterconing in long horizontal well. Figure 2: Even drainage with ICD
2 SPE 122824

To overcome this problem and to control and equilize the horiziontal well drawdown ICDs are installed at every connection in
the production tubing. Figure 2 shows that by applying flow restrictions, coning can be reduced and or controlled. Ideally the
entire water (or gas)-oil-contact should be parallel to the production tubing. If the waterfront enters the tubing over the entire
length at the depleted stage ultimate recovery has taken place.

In the following a flow analysis will be performed, to better understand the functionality of the ICD.

A hydraulic model for the ICD

Figure 3 shows a typical ICD tool. Screens are wound over base pipes. Axial rods provide standoff from the base pipe
surface, and also provide conduit for oil to flow towards the ICD. At one end the basepipe the ICD is installed.

Figure 3: A typical ICD tool

Oil from the reservoir enters the outside of the tool, flows through the screens into a pathway along the base pipe. Going
through the pathway the oil enters a chamber before going through several orifices. After passing through the orifices, the oil
flows through a number of large holes inside the casing. These orifices control the flow and are actually the ICD.

The coupled flow model includes the pressure drop from the reservoir through the ICD and into the base pipe. The flow and
pressure analysis to follow is based on a typical ICD application in the North Sea. The flow path is coupled in a series of
pressure losses and will be broken into the following components:

• The outside screen

• The conduit below the screen
• The chamber
• The orifices
• The holes through the casing

The outside screen

Each slot is a rectangular opening with typical dimensions of 0.01 in x .503 in (0.25 mm width). The screen wire is wrapped
onto a number of axial pins that are welded on the casing so that the effecive flow area is reduced. Analysing the actual
geometry of the screen revealed that 11% of the outside surface is the actual flow area. For one meter length of screen, the
effective inflow area is calculated to to be 12320 mm2.

The inflow velocity per meter length is given by:

v( m / s) =
q
=
(
Q m 3 / sek) = 81.17Q
A 12320x10 −6 m 2

The pressure drop is modelled as a laminar flow between two plates as defined by Bourgoyne et.al (1986) which derives the
equation for pressure drop between two plates. The final result is:
SPE 122824 3

wh 3 ΔP
Q=
12μ L

Defining wh as the effective flow area the pressure drop becomes:

ΔP(bar ) =
12μQL
=
( )
12μ Ns / m 2 Q(m 3 / s)1(m)
= 15.58x10 4 μQ (1a)
Ah 2
12320 x10 −6
(m ){0.25x10 } (m)
2 −3 2 2

Assuming an oil viscosity of 0.5 Poise, the above equation becomes:

ΔP(bar ) = 779Q(m 3 / sec) (1b)

The conduit below the screen

The axial flow through the conduit towards the nozzles is a little more complex. At any given point the flow is the cumulative
flow from the screen openings upstream. This will increase from one end of the conduit to the other.

Another complexity is that the shape of the conduit is a rectangle. We will define an equivalent hydraulic radius (Bourgoyne
et.al., 1986) and use the flow equations for a circular hole. The size of each conduit is 0.503 in x 0.202 in. Subtracting the
area of the rods the wire is wrapped on, the effective conduit area is 0.381 in x 0.202 in or 9.7 mm x 5.1 mm. The hydraulic
radius is the given by:

Area 9.7 x5.1

R H (mm ) = = = 1.67 mm
Wettedperimeter 2(9.7 + 5.1)

The hydraulic diameter is four times the hydraulic radius or 6.69 mm. The laminar pressure drop for a circular pipe is given
by:

32 μv
ΔP = L
d2
In consistent units and using the hydraulic diameter, the pressure drop becomes:

32μ( Ns / m 2 ) Q( m 3 / s )
ΔP(bar ) = x1(m) = 203000μQ (2a)
π
4
( ) 2 2
0.00669 2 m 2 0.00669 m ( )
Again assuming a viscosity of 0.5 cP, the equation becomes:

ΔP(bar ) = 11.5Q(m3 / s) (2b)

The chamber

All conduits enter into a chamber before flowing through the nozzles. Because this chamber is relatively large, the velocity is
small. This pressure drop can be neglected.

The nozzles

Initially assuming fully turbulent flow through the nozzles and using the pressure drop across a nozzle from Bourgoyne et.al.
1986 is given by:

ΔP( Pa) =
(
1 2 ρQ 2 ρ kg / m 3 Q 2 m 6 / s 2
ρv = =
) ( )
2 2 A2 2π 2 r 4 m 4 ( )
4 SPE 122824

The nozzles have a minimum diameter of 1/8 in. which is equivalent to a radius of 1.59 mm. The density of the oil is assumed
to be of 0.75 specific gravity or 750 kg/m3. The pressure drop is then given by:

ΔP(bar) = 5.98x10 7 Q 2 (3)

The pressure drop upstream from the nozzle is considered negligible.

The total pressure drop

The total pressure drop of the system is the sum of the individual pressure drops derived above.

Adding the screen, conduit and nozzle pressure losses gives the total pressure drop. If the length of the screen is defined as
L(m), the total flow rate across this length as Q(m3/s) and n is the number of nozzles, the total pressure drop becomes when
assuming a density of 750 kg/ m3 and a viscosity of 0.5 cP:

Q2
+ 11.5Q + 5.28 x10 7
Q
ΔP (bar ) = 779 (4)
L n

From a typical North Sea application values would be:

Flow rate: Q = 5 gal/min = 18.9 l/min=3.2 x 10-4 m3/s

Screen length: L = 10 m

Inserting into the pressure drop equation above results in:

6.12
ΔP (bar ) = 0.0098 + 3.69 x10 − 3 + (5)
n

The worst case is that all ten nozzles are open. In that case the total pressure drop is 6.12 bar. The pressure drops distribute
as follows:

Screen: 0.16 %
Conduit: 0.06 %
Nozzles: 99.76 %

The relative area between the flow paths are as follows:

Area of 5 m of screen: 5 x 12320 = 61600 mm2

Area of conduit: 44 x 9.7 x 5.1 = 2177 mm2
Area of 10 nozzles: 10 x π x 1.62 /4 =20 mm2

The screen has a very large open area as compared to the conduit which has 3.5 % of this area. The nozzles have only 0.03 %
of the screen area, or 0.9 % of the conduit area. This shows that the pressure drop is largely controlled by the nozzles. The
flow therefore varies with density due to turbulent flow, but is not sensitive to variations in viscosity as discussed in the
following.

Flow regime evaluation

By investigating if the flow regimes are either laminar or turbulent flow the value of the Reynolds number is evaluated. The
Reynolds number defines the transition between these two flow regimes. The following equation defines this transition
(Lundberg, 1967). For lower values than the Reynolds number the flow is fully laminar, that is when the pressure drop
depends on the viscosity of the fluid. For higher Reynolds numbers the flow goes into turbulent flow, where the pressure drop
is dependent on the fluid density. The Reynolds number can be expressed as:

vd
Re = ≤ 2320
ν
SPE 122824 5

where: v = average flow velocity

d = pipe diameter
ν= kinematic viscosity

The kinematic viscosity is:

μ
ν =
ρ
where: μ = fluid viscosity
ρ = fluid density

The transitional velocity between laminar and turbulent flow can be expressed as:

2320ν (m 2 / s ) 2320v(m 2 / s ) μ ( Pas)

v(m / s) = = (6)
d ( m) ρ (kg / m 3 )d (m)

It is convenient to use the kinematic viscosity as it implicitly contains both viscosity and density. However, the value varies
considerably with fluid type and temperature as seen in the following Table 1.

Table 1: Transitional flow rates for various parts of ICD.

Path: Fluid type: Diameter(m): Critical vel.(m/s) Critical Flowrate(l/hr)
Screen - - -
Conduit Water 0.00669 0.17 26.4
Light Oil 0.00669 1.74 2.64
Nozzles Water 0.0032 0.36 0.3
Light Oil 0.0032 3.63 0.03

The reservoir temperature of the produced oil is often less than 100°C. If we assume an average value between 50 and 100
degrees, water will have a dynamic viscosity of 5x10-4 Pas and for light oil 0.5 x10-4 Pas. Inserting these values into the
equation above we obtain:

1.16 x10 −3
For water: v(m / s ) =
d ( m)
−4
1.16 x10
For oil: v(m / s) =
d ( m)

Produced oil often has a viscosity one order of magnitude lower than lubricating oil.

Using the Reynolds number analysis above the flow regimes in the various parts of the ICD tool is evaluated next by
determining the transitional velocity and multiplying that value with the total flow area to obtain the critical flow rates.
The result is shown in Table 2 below. For a typical North Sea value of 18.9 l/min, only the nozzles obtain turbulent flow.
The initial assumptions of turbulence through the nozzles but laminar flow through the rest of the system appears correct.

Comparison with field data

There are many multilateral wells in the North Sea area where ICDs are applied. Typical production values are:

Field A: Oil production : 20 000 bbls/day

A good producer: 2500 Sm3/day
Screen length: 7 000 – 15 000 m per well (2–4 laterals)
430 000 m total
No. of wells: 110 (not all on production)

Field B: Oil production : 25 000 bbls/day

Screen length: About half of Field A
6 SPE 122824

100 wells produce on Field A with an average screen length of 4300 m. This gives an average flow per meter screen of 3.1
liter/m screen per hr.

The maximum flow rate of a good producer is:

3 3
2500 m / day1000liters / m
= 6.9liters / m, screen, hour
24 hrs / day15000 mscreen

Each screen is about 10 m long with nozzles at each end. The flow ICD is therefore equivalent to approximately 35
litres/hour.

Comparing the flow rates above with the transitional flow rates from Table 2, the following is seen:

Table 2: Flow regimes in the ICD

Path: Fluid type: Transitional flow Case A Flow regime
rate(l/hr) flow
Screen - - Laminar
Conduit Water 26.4 35 Laminar
Oil 2.64 35 Turbulent
Nozzles Water 0.3 35 Turbulent
Oil 0.03 35 Turbulent*
*Always turbulence through nozzles, ref. earlier discussion.
.

It can be observed that if water is produced through the ICD we get turbulent flow in the nozzles that control the pressure drop.
For oil production turbulence may arise in the conduits as well. Having a closer look at the viscosity aspects of the tool using a
table with typical values from Norwegian offshore fields is shown in Table 3.

Table 3: Typical oil data from Norwegian oil fields. From Skjaveland and Kleppe(1992).
3
Field: Oil density(kg/m ) Oil viscosity(cP) Dynamic viscosity
(Pas)
-5
Balder 914 3.0 300 x 10
-5
Draugen 824 0.68 68 x 10
-5
Gullfaks 838 0.40 40 x 10
-5
Gyda 822 0.28 28 x 10
-5
Heidrun 922 2.29 229 x 10
-5
882 0.75 75x 10
-5
Oseberg 850 0.43 43 x 10
-5
Smørbukk 832 0.14 14 x 10
-5
Snorre 690 0.42 42 x 10
-5
Frigg 835 4.83 483 x 10
-5
Troll 900 1.60 160 x 10
-5
Ekofisk 838 0.13 13 x 10
-5
Eldfisk 842 0.10 10 x 10

From Table 3 it can be observed a significant variation in oil properties. The kinematic viscosity varies between 0.012x10-5
and 0.578 x 10-5 m2/s for the fields shown, giving a ratio of 48:1. The density, on the other hand varies between 822 and 914
kg/m3, a ratio of 1.11:1.

For high flow rates, the flow through the conduit may become turbulent. However, because the relative area between conduit
and nozzle is large, and therefore a significant velocity contrast, the nozzle will always dominate the pressure drop.

Further investigation of laminar flow

It is clear from the previous analysis that a system containing a restriction is controlled by the restriction. This means that
most of the pressure drop occur across the restriction, and, because is usually is turbulent flow, it is controlled by the fluid
density.

From Table 3 it is obvious that the density is relatively constant in oil reservoir, whereas the viscosity shows significant
variations. For that reason it is desirable to let the pressure drop be controlled by density, that is turbulent flow.
SPE 122824 7

Although not desirable, we would like to investigate if it is possible to design a tool where the pressure drop is controlled by
viscosity. If a very long pipe is used, the frictional pressure drop is so large that the flow rate becomes very low, actually
laminar.

Undertaking a hypothetical case and assuming that the pipe is so long that the laminar pressure drop dominates, the pressure
drop in a circular pipe with laminar flow is defined by:

32μv
ΔP = L
d2

The maximum flow velocity to have laminar flow is given by the Reynolds number:

vd ρvd
Re = = ≤ 2320
ν μ

Solving for the pipe length:

Pd 2
L ( m) =
32μν

The critical transitional speed between laminar and turbulent flow is:

μ
v = 2320
ρd

Using the North Sea oil data from Table 3, we will compute the transitional speed and the pipe length for a 1/8 in. pipe. The
results are shown in Table 4.

Table 4: Flow velocities and pipe lengths for laminar flow

3
Field: Oil density(kg/m ) Oil viscosity(cP) Dynamic viscosity Critical velocity(m/s) Pipe length(m)
(Pas)
-5
Balder 914 3.0 300 x 10 2.38 28
-5
Draugen 824 0.68 68 x 10 0.60 480
-5
Gullfaks 838 0.40 40 x 10 0.35 1400
-5
Gyda 822 0.28 28 x 10 0.25 2800
-5
Heidrun 922 2.29 229 x 10 1.80 48
-5
882 0.75 75x 10 0.62 422
-5
Oseberg 850 0.43 43 x 10 0.37 1232
-5
Smørbukk 832 0.14 14 x 10 0.12 11667
-5
Snorre 690 0.42 42 x 10 0.44 1061
-5
Frigg 835 4.83 483 x 10 4.19 10
-5
Troll 900 1.60 160 x 10 1.29 95
-5
Ekofisk 838 0.13 13 x 10 0.11 13706
-5
Eldfisk 842 0.10 10 x 10 0.09 21778

The results of Table 4 show that if 1/8 in tubes are wrapped around the base pipe, they must be very long. Also, it can be
observe that if 10 nozzles should be replaced by long tubes, 10 of these tubes must be wrapped in parallel around the base
pipe. From this discussion we may conclude that the presently used ICDs are controlled by turbulent flow, and that it is not
likely that they are sensitive to laminar flow. In other words, the current ICDs are not sensitive to variations in fluid viscosity.
This is actually good, as there is much less variations in density than viscosity.

Limitations with current ICDs

The flow through the ICD depends on the pressure drop. The pressure drop is proportional to density and the squared flow
rate. Figure 6 shows this nonlinear relationship.
8 SPE 122824

Figure 4: Flow characteristics for an ICD.

The entire flow system may be defined as follows:

• Flow from the reservoir into the completion system. This is usually laminar flow.
• Flow through the ICD. This is turbulent flow.
• Cumulative flow from the toe to the heel of a horizontal well. This is laminar at the toe, but often turbulent at the
heel. The degree of turbulence will decrease with reservoir depletion.

From this brief discussion it is evident that the ICD cannot maintain constant flow through the depletion of the field. This is
because of the non-linar nature of the flow system, and because the turbulence in the horizontal section decreases with
depletion. Although an improved recovery is seen, it is difficult to obtain maximum recovery with current ICDs.

The Rygervalve

Observing the shortcomings with current ICDs because they are pressure controlled, Rygervalve (2009) was invented, which
is a flow controlled ICD. This tool will maintain constant flow regardsless of variations in pressure. It is based on a hydraulic
feedback principle, and although precise, it is simple and reliable. It is at an early stage of development, but lab tests have
confirmed the functionality.

Installing the new ICD controller in a horizontal well would result in a water-gas contact or an oil-water contact surface that
moved parallel to the well. Knowing the initial distance between the fluid contacts and the well trajectory, a maximal recovery
would result. The characterisics of the new ICD is shown in Fig. 7.

Figure 5: Characterisitcs of the Designer ICD.

The basic version of the new ICD is shown in the middle horizontal line. The flow is constant regardless of differential
pressure across the valve. This is the basic version.

For complex reservoirs one could consider other flow control properties. The reservoir engineer may require more flow in one
section and less flow in another section. The Rygervalve can also be set up for both increasing and decreasing flow with
change in pressure. This is also shown in Fig. 5. Performing drainage simulations over the lifetime of the reservoir, the
SPE 122824 9

reservoir/completion engineer can determine optimal flow throughout the wellbore. Calibrated Rygervalves are then installed
during the running of the completion string.

Conclusions

A hydraulic model for a typical ICD tool is developed. It consists of flow through the outer screen, flow through the conduit
and flow through the nozzles. This model is used to compute pressure drop throughout the system based on production data
from the North Sea and fluid data from the Norwegian oil fields. The following conclusions are reached.

The outer screen is always in laminar flow, whereas the nozzles always are in turbulent flow. More than 99% of the pressure
drop occurs through the nozzles. The pressure drop through the ICD tool is therefore controlled by the density of the produced
fluid. The fluid viscosity has only negligible influence.

A hypothetical analysis investigates a tool of very long pipes that are dominated by laminar flow. For this case the pipes must
be much longer than the screen at which ICD is installed. It is not considered realistic to build a viscosity controlled tool with
the data and dimensions used in the North Sea today.

The relative pressures from the formation to the base pipe changes during reservoir depletion. The ICD is a pressure
controlled device and may not provide optimal drainage during depletion. A new invention, a flow rate controller called the
Rygervalve provides constant or predetermined flow with depletion. With this tool the reservoir engineer can control the flow
throughut depletion, resulting in optimal recovery.

Acknowledgements

The authors would like to acknowledge Gisle Vold and Weatherford Norge for technical review and for providing figures for
the paper.

Nomenclature

A area
L length
P pressure
Q flowrate
RH hydraulic radius
d diameter
n number of nozzles in ICD
h height
w width
v velocity
μ viscisity
ν kinematic viscosity
ρ fluid density

References

Bourgoyne A.T., Millheim, K.K., Chenevert, M.E. and Young, F.S.(1986). Applied Drilling Engineering. Society of Petroleum Engineers
Textbook Series, Vol. 2, (1986), ISBN 1-55563-001-4.

Lundberg, T(1067). Hydromaskinlara. Akademiførlaget Gumperts, Gøteborg, 2nd edition (1967).

Rygervalve (2009). Norwegian Patent NO 20082109: Anordning ved stromningsregulator ( A device for control of flow). International
patents pending.

Skjaveland, S.M. and Kleppe, J.(1992). SPOR Monograph: Recent Advances in Improved Oil Recovery Methods for North Sea Sandstone
Reservoirs. Editors S. M. Skjæveland and J. Kleppe. Norwegian Petroleum Directorate. ISBN 82-7257-340-7.