Advances in

Vol. 3, No. 4, p. 343-354, 2019

Geo-Energy Research
Original article
Application of intelligent well completion in optimising oil
production from oil rim reservoirs
Eric Broni–Bediako *, Naziru Issaka Fuseini, Richard Nii Ayitey Akoto, Eric Thompson Brantson
Petroleum Engineering Department, University of Mines and Technology, Tarkwa, Ghana
(Received August 31, 2019; revised September 6, 2019; accepted September 12, 2019; available online September 22, 2019 )

Citation: Abstract:
Broni-Bediako, E., Fuseini, N.I., Akoto, Intelligent well application has proven useful in maximising oil production from oil rim
R.N.A., Brantson, E.T. Application of reservoirs. Intelligent wells are equipped with downhole sensors and surface controlled
intelligent well completion in optimising downhole inflow control valves (ICVs) which should be strategically controlled by the
oil production from oil rim reservoirs. operator. Challenges however arise in determining the best reactive control strategy (RCS).
This paper seeks to develop an effective RCS (algorithm) that will maximise oil production
Advances in Geo-Energy Research, 2019, and to ascertain how the proposed RCS will fare when porosity, permeability, oil-water
3(4): 343-354, doi: contact and skin factor change. An anticlinal oil rim reservoir with a horizontal well was
10.26804/ager.2019.04.01. modelled and run using ECLIPSE 100. The well was later made intelligent by installing
Corresponding author: ICVs and a RCS was designed to control the valves. Three RCS were proposed but the
algorithm that produced the maximum cumulative oil was selected to be the optimal. The
*E-mail: ebroni-bediako@umat.edu.gh intelligent well yielded more cumulative oil and gas than the conventional horizontal well.
Keywords: It also delayed water breakthrough and reduced cumulative water production. Sensitivity
Conventional well analysis on porosity, permeability and skin positively affects the developed reactive control
intelligent well completion strategy whereas oil water contact variations yielded poor results. Economic analysis of the
intelligent well for 20 years showed that the application of the intelligent well completion
oil rim reservoir in the oil rim reservoir was profitable.
reactive control strategy
water coning

1. Introduction cusping has always been the problem of a production engineer

whose aim is to maximize oil production, hence the need to
Oil reservoirs are always characterised by production prob-
devise methods of delaying and/or minimising water and gas
lems during the life of the reservoir. Depending on the type of
production (Sarkodie et al., 2014).
problem, appropriate techniques are required to solve them.
The petroleum sector has over the years been pursuing
Some of the problem faced during oil production are, gas
the implementation of remotely controlled and monitored
cusping problems, water control problems, permeability im-
well completion termed, intelligent well completion (IWC).
pairments, equipment failures and several others. An inherent
It enables a producer to effectively monitor and shut valves at
problem that is associated with production of oil from oil
desired locations (Robinson, 2003).
rim reservoirs is early production of water as well as gas
The intelligent well technology is one of the most impor-
cusping. Early water production from an oil rim reservoir
tant technologies needed in reservoir optimisation, increasing
causes corrosion of tubulars, scale/salt deposition, gas hydrate
recoverable reserves, enhancing oil recovery and reducing wa-
formation, disposal problems of the water itself and high cost
ter cut. In 1997, Saga Petroleum installed the first ever intelli-
of lifting the water (Anon, 2016). The production of oil is
gent well technology in the world. This technology has hitherto
usually expected after millions of dollars and several man-
achieved a widespread implementation and development in
hours is spent in developing a reservoir. However, production
almost all parts of the world (Huang et al., 2011). However,
engineers are faced with problems in dealing with unwanted
IWC is faced with issues of longevity, reliability and high
fluid such as gas and water in oil rim reservoirs as they
cost attached to its usage. It is therefore necessary to evaluate
produce the oil (Sarkodie et al., 2014). Water production
the following factors before considering the implementation of
during the life of a well is inevitable but early and massive
IWC. Equipment diameters and available space, fluid velocity,
production of the water can be managed. Having to identify the
pressure drop and erosion. It is also important to evaluate and
cost-effective method of managing water production and gas
establish means of protecting sensors, cables and control lines

https://doi.org/10.26804/ager.2019.04.01.
2207-9963 c The Author(s) 2019. Published with open access at Ausasia Science and Technology Press on behalf of the Division of Porous
Flow, Hubei Province Society of Rock Mechanics and Engineering.
344 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354

of the IW system (Mahmood et al., 2018). on the evaluation of investment and the return on investments
Optimisation methods (control techniques) are applied to and as such, sensitivity analysis on some of these were carried
intelligent well technology to balance production along the out. He found out that, reduced oil price resulted in positive
wellbore length, control water breakthrough, and ensure early NPV but skin negatively affected production from IWC.
economic oil production (Masoudi et al., 2013). The control Several other control strategies have been proposed for
strategies include open and closed loop. Closed loops are various forms of reservoirs but important challenges still arise
further classified as either feedback (reactive) or proactive. from developing an effective control strategy that will be
The most commonly used inflow control approach is the applied on intelligent wells in order to optimise the production
‘proactive’ closed-loop strategy (Dilib et al., 2015). Raoufi of oil from an oil rim reservoir. As such, this paper seeks to
et al. (2015) proposed an optimisation algorithm based on provide an effective and efficient RCS to tackle early water
simulated annealing (SA) algorithm to obtain an optimum breakthrough and gas cusping from thin oil reservoirs so as
control strategy and determining an operation that maximises to maximise oil production, by using a horizontal well with
the net present value (NPV). This algorithm ensures that, the ICVs implemented at various segments of the well.
water that is being produced does not exceed the desired limit
for water production. The path followed by the SA algorithm 2. Resources and methods
for finding the optimum results yielded about 13.91% increase
in cumulative oil and a decline of 31.89% in cumulative water 2.1 Reservoir model description
production.
Raoufi in one of his publications titled “optimisation of The reservoir’s geologic model used for this research was
flow control with intelligent well completion in a channelised built using the Cartesian (block centred) grid system of the
oil rim reservoir” in 2011, developed a mathematical model eclipse software, specifically ECLIPSE 100 (black oil model).
based on the trust region method. As part of his work, he Data used in the modelling of the thin oil rim reservoir was
compared three control strategies used in optimisation of obtained from Chang (2014) and presented in Table 1. The oil
production from oil rim reservoirs. These include fixed flow rim reservoir is an anticlinal reservoir with a gas cap above
control devices, ON/OFF control valves, and infinitely variable and an inactive aquifer below it and it contains light oil within
control devices. The comparison of the valve control strategies a column of 35 ft. In modelling the reservoir using ECLIPSE,
indicated that, the ON/OFF control valve algorithm yields the 50 cells were used in both the x and y directions, and 20
best cumulative oil production. In the ON/OFF valve control cells in the z direction for simplicity and optimal run time.
strategy, the inflow control valves (ICV) are installed at posi- The dimensions of all the cells were made equal, having a
tions to trigger oil production and stop water production based length by width by height measurements of 200 × 200 × 5
on pre-set values. Adekunle (2012) proposed an algorithm to respectively, so as to obtain the anticlinal shape. The reservoir
optimise oil production from oil rim reservoirs. His method is a homogenous reservoir with a uniform porosity of 0.2 and
was based on what he called “trial and error method”. He a permeability of 30 mD in x, y and z directions. The reservoir
used data simulation models from oil rim field called chevron model had a total of 5,000 cells and it is located at a depth
X field but he did not disclose the data due to confidentiality. of 3,505 ft. Fig. 1 shows a diagram of the modelled anticlinal
The Schlumberger reservoir simulator was his choice of tool oil rim reservoir.
in modelling the reservoir. He implemented Darcy’s flow rate
equation into the eclipse software “WCONPROD” keyword 2.1.1 Model initialisation
under the “Schedule” section and set the flow rate to the The geologic (static) model of the reservoir was converted
maximum value. Modelling for Open/Shut ICDs with the into dynamic model by the rock saturations and fluid model.
Schlumberger Eclipse 100 simulator was achieved by applying Since the reservoir contains water, oil and gas, both the water/
the CECON keyword in the SCHEDULE section to set a
water cut limit of 90%, at which the errant connections are
automatically SHUT. Simulations were run on the basis of
trial and error by varying the “WCONPROD” keyword for
three different oil production rates, that is 5,000 stb/day,
7,500 stb/day and 10,000 stb/day. Based on the result, the
IWC producing at 10,000 stb/day proved optimal. Under this
production rate, there was 988.38 Mstb (5.41%) increase in
Field Oil Production Total and a 0.6% increase in Field Oil
Efficiency. However, Field water Production Total increased
by 132.63 Mstb (17.81%) for IWC based on the trial and
error method but it remained below the 90% economic limit
set. Economic evaluation was done using net present value
calculation of cash inflow-cash outflow to determine the eco-
nomic value of the project if intelligent wells are used. The
effect of uncertainties in variables such as labour, cost of Fig. 1. Anticlinal oil rim reservoir.
equipment, raw materials and oil price can have major effect
 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354 345

Table 1. Main parameters of reservoir model.

Parameter Value Unit

Porosity 20 %
Horizontal permeability 30 mD
Vertical permeability 30 mD
Thickness 35 ft
Gas oil contact 3,505 ft
Water oil contact 3,540 ft
Oil API 48.8 API
Oil viscosity 1.01 cP
Oil formation volume factor 1.08 bbl/stb
Dimensions of reservoir 50 × 50 × 20 Grid blocks
Dimensions of grid block 200 × 200 × 5 ft3
Rock compressibility 3.14 × 10−6 psi−1
Oil density 49.99 lb/ft3
Water density 63.698 lb/ft3
Gas density 0.050674 lb/ft3

kro vrs sw krw vrs sw bubble point pressure of 4,351.1 psi. In addition to that,
1.2 0.7 the reservoir also has a gas cap containing dry gas and its
properties is described under the keyword PVTO in the model
1 0.6
data. Its phase pressure is given as 200 psi at a viscosity of
0.5
0.8 0.012826 cP and gas formation volume factor of 15.54 ft3 /scf.
0.4 The reference pressure of the rock is 2,949 psi and the rock
Krw
Kro

0.6
0.3 compressibility is given as 3.14 × 10−6 psi−1 . The initial oil
0.4 saturation is 0.77 and its residual oil saturation is 0.24. The
0.2
0.2 0.1
initial water and irreducible water saturations are 0.76 and
0.23, respectively. However, the initial oil saturation from the
0 0
0 0.2 0.4 0.6 0.8 1 gas oil relative permeability is given as 0.475 and its residual
Sw saturation is 0.037. The initial gas saturation is 0.963 and
its residual gas saturation is 0.525. The oil-water and gas-oil
Fig. 2. Oil-Water relative permeability curve. relative permeability curves for the fluids are shown in Figs.
2 and 3, respectively.

kro vrs sg krg vrs sg 2.2 Well modelling and placement

1.2 1.2
A horizontal well named N1, with measured depth (MD)
1 1
of 4,950 ft and a true vertical depth (TVD) of 3,350 ft was
0.8 0.8 modelled using the WELSPEC keyword. A horizontal well
was the preferred type of well for thin oil reservoirs due
Krg
Kro

0.6 0.6
to the fact that they provide a large contact area with the
0.4 0.4
thin oil column as opposed to vertical wells. The well was
0.2 0.2 completed and perforated at seven regular intervals of 200 ft
0 0 each, i.e., eight perforations using the COMPDAT keyword.
0 0.2 0.4 0.6 0.8 1 The wells position was varied and simulations were done until
Sg the optimal position was obtained. The optimal position is the
grid coordinates that yields the maximum oil recovery and also
Fig. 3. Gas-Oil relative permeability curve. lies within the thin column of oil where coning is possible.
Using the well control data keyword “WCONPROD”, an oil
flow rate target of 6,000 bbls/day and a bottom hole production
oil saturation functions (SWOF) and the gas/oil saturation target value of 250 psia was used in the simulation. Fig. 4
functions were needed to initialise the model. The reservoir shows a horizontal well placed inside the reservoir.
contains live oil with dissolved gas-oil ratio of 1.5208 at
 Figure 1 Anticlinal Oil Rim Reservoir
346 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354

Figure 4 Top View of the Reservoir with the Horizontal Well Fig. 5. A schematic of multi-segment well structure (Kumar et al., 2016).
Fig. 4. Top view of the reservoir with the horizontal well.

ECLIPSE using a homogenous model of subcritical flow

through a pipe constriction using the Eqs. (1) to (3) (Al-
2.3 Well segmentation
Ghareeb, 2009; Sampaio et al., 2012).
For valve placement in ECLIPSE, the horizontal well
had to be divided into segments to enable the placement of δ P = δ Pcons + δ Pf ric (1)
the valves (Aitokheuhi, 2004). Well segmentation provides
detailed analysis of fluid flow in horizontal and deviated wells. ρv2c
δ Pcons = Cu (2)
Flow rates of oil, water and gas at segments can be monitored 2C2v
and when the well is segmented. The keyword WELSEGS was
L2 2
used to define the segment structure of the horizontal well N1 δ Pf ric = 2Cu f ρV (3)
in this research. The keyword has two main record fields, the D p
first field describes the top segment (segment nearest to the where
wellhead) and sets some general flags. The second field can δ f ric accounts for additional frictional pressure drop in
consist of one record if only one segment is to be defined, but valve segment (psi)
in this paper, eight segment structures were described so that δcons accounts for the effects of the constriction (psi)
each perforation is allocated a single segment within a 200 ft Cu (unit conversion constant) = 2.159×10−4
length. ρ = density of the fluid mixture, (lb/ft3 )
The segmenting was done in a way that each node of a vc = flow velocity of the mixture through the constriction
segment lies within a connection (perforation). Prior to the (ft/s)
segmenting, the keyword WSEGDIMS was used to set array Cv = dimensionless flow coefficient of valve
dimensions for multi-segment wells. It defined the maximum f = Fanning friction factor
number of multi-segment wells in the model, the maximum l = Additional length of piping in segment (ft)
number of segments and the maximum number of branches per D = Diameter of the pipe (not constriction) (ft)
well. Fig. 5 shows a schematic diagram of multi segments well. v p = flow velocity of mixture through pipe(ft/s)
The segments needed to be completed before installing the Several forms of ICVs exist and based on their operations,
valves. The keyword COMPSEG and its related data records they are either categorised as fixed flow control devices, binary
allowed for the completion of multi-segmented wells (a well (ON/OFF) or infinitely variable ICV (Rauofi and Mashishi,
with more than one segment). It defines the locations of 2011). The type of ICV used for this paper is the binary ICV
completions in a multi-segment well and ECLIPSE software due to the nature of the algorithm that will be developed.
allocates each completion to a well segment (Muhammad, Binary ICV assume only two positions, that is, either fully
2008). open or fully closed. Binary ICVs also provide a firm control
of both gas and water in various segments (Sarkodie et al.,
2014). Hence, it was chosen as the best ICV for the intelligent
2.4 Inflow control valve (ICV) modelling well.
In modelling ICVs using ECLIPSE, eight records are
Eclipse keyword “WSEGVALV” was used under the
needed to complete the model. The first is the well name
SCHEDULE section to model inflow control valves and assign
on which the ICV is to be installed in, and for this work,
them to each segment. This keyword designates specific well
the well name was “N1”. The second record is the segment
segments to represent a sub-critical valve in a multi-segment
number to contain the valve, the third is the dimensionless
well. This imposes an additional pressure drop in the segment
flow coefficient (Cv ) and the fourth is the cross sectional area
due to flow through a constriction with a specified area of cross
of the constriction. The remaining records can be defaulted so
section. The pressure drop across the device is calculated by
that ECLIPSE takes them from the WELSEG data. Eight (8)
 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354 347

Start Production Start Production

Open ICVs at All Produce for a

Timestep
Segments

Produce for one Which ICV has the

Timestep Highest SWCT?

Which Segment Close ICV

has the Highest
WCT? Keep ICV Open
Closed ICV

Close ICV in that No Has SWCT Yes

Segment Fallen Below any
Other SWCT?

Fig. 6. A schematic of RCS 1 (Algorithm 1).

Fig. 7. A schematic of RCS 2 (Algorithm 2).

valves were installed at segments two (2) to nine (9) and they
assumed a fully opened position of area 0.022 ft (3.168 square
of the reservoir and the problems the algorithm intended to
inch) with a Cv value of 0.66.
solve, which is water and gas coning and the most effective one
was selected based on its ability to maximise oil production by
2.5 Development of a reactive control strategy controlling coning. Unlike produced water, gas is a desirable
component which contributes revenue and so the algorithms
Developing an effective RCS to be applied on an in-
targeted produce water reduction and also finding a way of
telligent well in oil rim reservoirs requires careful analysis
delaying gas production. By delaying the production of gas
and understanding of the reservoir characteristics and as such
in the reservoir, the gas provides a gas cap drive mechanism
sequential steps were applied in developing the algorithm. The
to aid oil production which is of much importance. The three
following steps detail how the algorithm was developed using
reactive control strategies are detailed as follows.
the “ACTIONX” keyword in ECLIPSE. Other forms of control
keywords exist in ECLIPSE and these include: ACTION, AC-
Algorithm 1 development process
TIONS, ACTIONG, ACTIONW and ACTIONR but the choice
of control keyword for this work is the ACTIONX keyword The first algorithm was designed to specifically target water
since it provides more flexibility and allows comparison of the production in the segments of the well. The algorithm basically
well’s parameter against another. Unlike ACTION (G, R, S, instructs all ICVs to open at the start of production and then to
W) which allows comparison of a well quantity against only shut ICV in the segment that has the highest water cut amongst
a constant value. The ACTIONX keyword plays an integral them all. That ICV is kept closed and production is carried
part of developing an effective RCS since it specifies a set out for a 30-day period after which all the ICVs are opened
of keywords to be processed if a set of conditions are met again and a query of which segment has the highest water cut
(Amangaliyev, 2017). In this paper, since the main objective (WCT) is made again. This was achieved by looping the set
is to delay or if possible minimise water and gas production, of instructions till the 20 years of production ends. Fig. 6 is a
the algorithm focused on mitigating water and gas production schematic of the second RCS that was coded into the model.
at various segments of the well.
The ACTIONX keyword is just like an “IF” condition used Algorithm 2 development process
in various programming languages to write programmes. The
Algorithm 2 was designed to minimise the water cut in the
keyword was placed under the schedule section of the data
segment that has the highest water cut. The algorithm opens
file and it was terminated with the keyword “ENDACTIO”.
all ICVs at the start of production and after 30 days, a query
The ACTIONX facility of the simulator allows for nested
is carried out by the algorithm to determine which segment
conditions, hence complex “IF” conditions can be looped. To
has the highest water cut. The ICV in that particular segment
achieve an efficient RCS, several complex conditions were set
is instructed to close and production carried out for a month
and simulation runs to determine which algorithm will yield
after which another query as to whether the segment water cut
the maximum cumulative oil production and will yield the
(SWCT) in that particular segment has fallen below any other
maximum NPV.
SWCT. If yes, the closed ICV is now opened for a month
of production but if no, the ICV is kept closed and produced
2.5.1 Algorithm development results
for a month and repeated till production ends. A schematic of
Three algorithms were developed based on critical analysis algorithm 2 is represented in Fig. 7.
348 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354

Start Production
independent variable affects a particular dependent variable
Open All ICVs under a set of assumptions (Adusu, 2018). Dynamic and
static reservoir parameters were varied at two extreme cases
Produce for One Time
(best and worst cases). The two static reservoir parameters of
Step importance in this work are absolute porosity and absolute
permeability and the dynamic reservoir parameters are skin
and oil water contact.
Which Which
Segment Segment
has the has the 2.6.1 Static reservoir parameters
Highest Highest
SGOR? SWCT? Porosity and permeability for two extreme cases (best and
worst) were put into the model so as to determine how sensi-
Close ICV in Close ICV in
tive the algorithm is in relation to these variables. These
Segment with Segment with two extreme case values were obtained by multiplying the
Highest Highest base case porosity and permeability (porosity and permeability
SGOR SWCT
used in the main reservoir model) values by a factor of 2
and 0.5 to obtain the best case and worst case values. From
Darcy’s law, a direct variation exists between flow rate and
Has SGOR Yes Yes Has SWCT permeability, hence, there will be an increase in flow rate
Fallen Below Open Fallen Below when the permeability of a reservoir increases (Tarek, 2010).
any Other ICV any Other
SGOR? SWCT?
However, there is no direct relationship between flow rate and
porosity since a reservoir can have high porosity but cannot
transmit flow due to low permeability. That notwithstanding,
No No
Keep Closed there is a generalisation that high porosity will usually result
in high flow rate. Table 2 shows the porosity and permeability
values used in all three cases.

Fig. 8. A schematic of RCS 3 (Algorithm 3). 2.6.2 Dynamic reservoir parameters

The robustness of the selected algorithm in the presence
Table 2. Permeability and porosity vlues for the three cases.
of formation damage (skin) was determined by varying the
skin around the wellbore to extreme cases. In the presence
Absolute permeability (mD)
of skin, the wellbore generates an additional pressure drop
Best case Base case Worst case
around it leading to low productivity index (Tarek, 2010). The
60 30 15 productivity index is a mathematical means of expressing the
Absolute porosity
ability of fluids to be delivered into the wellbore (Aulisa et al.,
0.4 0.2 0.1
2011). The best case value which represents a stimulation was
-2 and the worst case value which represents formation damage
was +5. Simulation was carried out for 20 years at 5 years
interval, and the well’s productivity index (PI) was monitored
Algorithm 3 development process
to see how the algorithm functioned under formation damage
Algorithm 3 was designed to open all ICVs and produce and when stimulation is carried out. Formation damage is
for a time step after which a query is made to find out which related to pressure drawdown and PI by Eqs. (4) to (7).
segment has the highest SWCT and SGOR. The ICV in that
qp
particular segment is then instructed to shut and another query ∆Pskin = s (4)
is made to find out if the SWCT and SGOR in the identified 2πkh
segment have fallen below any SWCT and SGOR respectively qµBo  re 
in the other segments. If yes, the algorithm opens the ICV Pe − Pw f = 141.2 In + s (5)
kh rw
in the identified segment but if no, the algorithm keeps the
ICV closed and production is carried out for one-time step. A Pe − pw f = drawdown (6)
schematic of algorithm 3 is represented in Fig. 8.
q
2.6 Sensitivity analysis PI = (7)
Pe − Pw f
In order to satisfy uncertainties in the reservoir caused by where
changes in the reservoir fluid and rock properties, sensitivity q = oil flow rate (stb/day)
analysis was carried out to determine the robustness of the k = permeability (mD)
selected algorithm. Sensitivity analysis is the study of how an h = thickness (ft)
s = skin factor
 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354 349

pe = external boundary reservoir pressure (psi)

pw f = flowing bottom hole pressure (psi) Table 3. Skin and oil-water contact values for the three cases.
µ = oil viscosity (cP)
Bo = oil formation volume factor (bbl/stb) Skin
re = drainage radius (ft) Best case Base case Worst case
rw = wellbore drainage radius (ft) -2 0 +5
PI = productivity Index (STB/day/psi) Oil-Water contact (ft)
This implies that the additional pressure drop generated 3,595 3,540 3,518
due to skin affects the productivity of the well and therefore, Table 4. Economic model parameters.
low PI is not desirable since it reduces the field of efficiency
(FOE).
Parameters Value Unit
The oil water contact (OWC) of the reservoir was also
Drill and complete intelligent well cost 120,000,000 $/well
varied so that the column of oil within the reservoir attains
OPEX, %Revenue 5%
a value of 13 ft and 90 ft to depict the worst case and
Commodity desk
best case scenario. This dynamic reservoir property varies
Water treatment cost 4 $/bbl
as production from the reservoir is carried out. The selected
minimum and maximum values of the oil column depict Price of oil 71.09 $/bbl
theoretical values of the oil column that can exist in an oil Price of gas 2.712 $/Mscf
rim reservoir. So, applying the IWC coupled with the selected Discount rate 10 %
algorithm (algorithm 3) on these oil columns was carried
out to determine how the algorithm will fare as the size of
the water column changes. To obtain the best case value, an
OPRICE = average oil price over the year ($/stb)
optimistic OWC of 3,595 ft was used in the simulation and
GP = cumulative yearly gas production (MSCF)
this value represents an oil column thickness of 90 ft, since
GPRICE = average gas price over the year ($/MSCF)
the gas oil contact (GOC) is at 3,505 ft. For the worst case
CAPEX = capital expenditure for drilling and completing
scenario, that is 13 ft oil column, an OWC at 3,518 ft was
IWC ($)
used in the model and simulation was run over 20 years. The
OPEX = field operating expenditure ($)
cumulative water produced otherwise known as field water
WP = cumulative yearly water production (bbls)
production total (FWPT) and field oil production total (FOPT)
WCOST = average water treatment cost ($/bbls)
were analysed for the two extreme cases. Table 3 presents
NCF = net cash flow ($)
values of skin and oil-water contact under the circumstances.
PV = present value ($)
i = discount rate (effective) (%)
2.7 Profitability of intelligent well completion n = number of interest compounding periods
NPV = net present value ($)
To determine the profitability of applying intelligent wells, Average prices of commodities as of 10th April 2019 were
the standard petroleum engineering profitability measure, net obtained from oilpricewidget.com (Anon, 2019) and used in
present value (NPV) was calculated. An economic model was the economic model. The cost of drilling and completing the
built to aid in the evaluation of the NPV of the intelligent well. intelligent well took into account the cost of installing ICVs
The model took into account the costs of drilling and com- as well.
pleting an intelligent well, the cost of treating produced water
and field operating expenditure. Since the oil rim reservoir
contained gas and light oil, the income from producing oil and
3. Results and discussion
gas was calculated on yearly basis and then discounted over 20 3.1 Algorithms comparison and analysis
years. The mathematical Eqs. (8) to (10) were formulated to
encompass all the parameters stated. However, the economic The three developed algorithms were put into the reservoir
model assumes no fiscal regimes and Table 4 shows the model and simulation was carried out for 20 years. The
parameters used in calculating the NPV. cumulative oil produced (FOPT) from the three algorithms
is shown in Table 5. The algorithm that yielded the highest
NCF = (N p × OPRICE + GP × GPRICE )
(8) cumulative oil was considered the best because it will lead to
− (CAPEX + OPEX +WP ×WCOST ) the highest NPV. Algorithm 3 yielded the maximum cumula-
NCF tive oil amongst the three proposed algorithms for the given
PV(Y EARLY ) = (9) well parameter in this research. This can serve as a basis for
(1 + i)n
predicting the optimisation of oil production from other wells.
20
NCF Hence it was chosen to be the optimal and most effective
NPV = ∑ (1 + i)n (10) amongst the three proposed algorithms. The selected algorithm
n=0
was later on, compared with a conventional production model
where to determine its optimality.
N p = cumulative yearly oil production (stb)
350 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354

From Fig. 9, it was observed that the developed algorithm

Table 5. Field oil production total comparison of algorithms. 3 minimised oil production at the early years of production
and increased it at the latter years, surpassing that of the
Algorithm number FOPT (bbls) conventional well by 135,557.8 bbls. This was because, at the
1 2,528,524.0 early years of production, there was massive water and gas
2 2,534,880.8 measurement at several segments and so the algorithm 3 had
3 2,579,034.3 manage it by closing the ICVs to prevent its influx into the
wellbore. The Intelligent Well (IW) that was producing based
Table 6. Intelligent well versus conventional well. on algorithm 3, produced 2,579,034.3 bbls of oil whereas the
conventional well produced 2,443,476.5 bbls of oil (Table 6).
Well type FOPT FGPT FWPT This indicates an increase in oil production for the IW due to
(bbls) (Mscf) (bbls) RCS.
Conventional well 2,443,476.5 35,059,240.0 3,361,211.8
Fig. 10 shows FGPT comparison of IWC and CW. From
Intelligent well 2,579,034.3 35,120,676.0 3,020,343.8 Fig. 10, with regards to gas production, the intelligent well
produced a cumulative gas of 35,120,676 Mscf whereas
the conventional well gave a cumulative gas production of
3.2 Intelligent well completion results and analysis 35,059,240 Mscf (Table 6). Despite the slightly higher gas
production in the intelligent well, the intelligent well delayed
In order to determine the effectiveness of the selected its gas production at a lower rate than the conventional well
algorithm, the algorithm coupled with IWC was compared for about 12 years, so as to use its gas cap drive mechanism
with conventional well. Field parameters such as FOPT also to provide energy for the oil recovery. From the 14th year,
known as the cumulative oil production, field gas production the cumulative gas production of the intelligent well peaked
total (FGPT) and field water production total (FWPT) for to match with the CW, and later surpassing it to yield a
the intelligent well coupled with the developed algorithm was slightly higher cumulative gas production. In order to ascertain
compared with the conventional well and the results are shown the impact of reinjection of the produced gas, gas injection
Table 6. simulation was carried out on the IWC and the results obtained
From the Table 6, it was observed that, the conventional indicated a large increase in the cumulative oil production
well (horizontal well) without ICVs and a proper reactive (4,329,942.5 bbls), cumulative gas production (65,772,400
strategy produced more water (3,361,211.8 bbls) than the MScf) and cumulative water production (7,619,186.5 bbls)
intelligent well coupled with the developed RCS (algorithm 3) (Fig. 10).
which produced 3,020,343.8 bbls of water. Again, there was an Fig. 11 shows a graph of cumulative water of the CW and
optimisation in the cumulative oil produced when an intelligent IWC. From Fig. 11, the intelligent well also decreased the
well was used in the oil rim reservoir. Fig. 9 shows a FOPT Field Water Production Total (FWPT). The cumulative water
comparison of the intelligent well coupled with algorithm 3 produced from the conventional well was 3,361,211.8 bbls and
and a conventional well. The graph was obtained from the 3,020,343.8 bbls from the IWC as shown in Table 6, which is
result viewer of the ECLIPSE simulator software. about 10.14% reduction in produced water for the IWC.
From the Fig. 11, it is observed that, the IWC increased

Fig. 9. Comparison of FOPT from IWC and CW. Fig. 10. Comparison of FGPT from IWC and CW.
 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354 351

Table 7. FOPTs and FOEs after permeability variations.

Field Best case scenario Worst case scenario

parameters (high permeability case) (low permeability case)
CW IWC CW IWC
FOE 0.095 0.101 0.087 0.090
FOPT 2,532 2,668 2,321 2,413
(bbls) 615.3 623.5 028 758

Table 8. FOPTs and FOEs after porosity variations.

Field Best case scenario Worst case scenario

parameters (high porosity case) (low porosity case)
CW IWC CW IWC
FOE 0.088 0.089 0.092 0.100
FOPT 4,681,421 4,764,500 1,230,266 1,333,927
(bbls) - - - -

BEST CASE CW BEST CASE IWC

WORST CASE CW WORST CASE IWC

Cumulative Oil Production (x106 bbl) 3

2.5

1.5

0.5

0
0 5 10 15 20 25
Time (Years)
11 FWPT Comparison of IWC and a CW
Fig. 11. FWPT comparison of IWC and CW.
Fig. 12. FOPT comparison of IWC and CW under permeability variations.

the water breakthrough time to about 2 years (730 days) unlike

the convention well whose breakthrough time occurred within which has been previously modelled) by a factor of 2 and
a month. The delay in water breakthrough time was due to 0.5 respectively is presented in Table 7. The FOPT and FOE
the shutting of the ICVs during the period that massive water of the two extremes are illustrated in Table 7. The intelligent
production was eminent. well coupled with algorithm 3 yielded higher FOE and FOPT
than the conventional well in both the best case and worst
case scenarios. This indicates how robust the algorithm 3 is
3.3 Sensitivity pattern results and analysis even under varying permeability of the oil rim reservoir. This
ultimately means that, the developed RCS will function better
The results obtained from varying porosity, permeability,
than a conventional well even when the oil rim reservoir has
oil water contact and skin so as to determine how robust
an absolute permeability as low as 15 mD and as high as 60
algorithm 3 is, in relation to these variations are tabulated
mD. Fig. 9 shows the graphical representation of Table 7.
and discussed under this section.
From Fig. 12, it was observed that, the developed algorithm
functioned more effectively under the best case, thus high
3.3.1 Results of sensitivity of absolute permeability
permeability. This is attributed to the fact that an increase
The results after obtaining the best and worst case by in permeability increases flow rate and consequently leads
multiplying the base case porosity (the porosity of the reservoir to higher cumulative production. The IW produced 136,008
352 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354

Table 9. Productivity index variations of IWC over 20-year period.

Skin factor Parameter 5 years 10 years 15 years 20 years

0 Base case PI 8.06 5.71 6.08 8.50
5 Worst case PI 4.64 4.82 5.69 8.86
-2 Best case PI 84.77 10.18 9.87 12.24

BEST CASE (CW) BEST CASE (IWC)

(5.37%) bbls of oil more than the conventional well in high
permeability case and 93,730 (4.04%) bbls more than the WORST CASE (CW) WORST CASE (IWC)

CW for low permeability case. This establishes that, the IWC

Cumulative Oil Production (x106 bbl)

6
coupled with the proposed algorithm 3 will be more effective
in high permeability oil rim reservoirs. For high permeability 5
oil rim reservoirs, fluids are transmitted easily to the ICVs.
The algorithm 3 is designed in a way that the more fluids, 4
specifically, water and gas get to the ICVs the more the ICV
generates pressure drop to equalize the high velocity of the 3
fluids, hence functioning more actively. The algorithm has
2
the ability to curtail the massive water and gas breakthrough
associated with high permeability rim reservoirs at early stage.
1

3.3.2 Results of sensitivity of absolute porosity 0

0 5 10 15 20 25
The robustness of the developed algorithm 3 after varying
the absolute porosity of the reservoir for the two extremes (best Time (Years)
and worst case) is detailed under this section. The FOPT and
Fig. 13. FOPT comparison of IWC and CW under porosity variations.
FOE of the best case and worst case for the IW and the CW
are compared and presented in Table 8. The best case porosity
value was obtained by multiplying the base case porosity by
worst case skin factor of +5 were analysed in terms of
a factor of 2 and the worst case by a factor of 0.5.
Productivity Index (PI) over a five-year period. This is because,
From Table 8, the algorithm 3 fared better than the con-
this dynamic property changes with time and so there is a
ventional well in both cases. It yielded higher FOE and FOPT
need to monitor its effect on PI which has a direct relation
than the conventional well. However, one striking observation
with skin. The best case and worst case values of -2 and
was that, unlike the permeability where the best cases resulted
+5, respectively, were chosen because they represented the
in higher FOEs for both conventional and intelligent well, best
allowable values that will not cause convergence problems to
case porosity case gave the lower values of FOEs in the oil rim
the software. The problem results when the effective wellbore
reservoir. That is, FOE for conventional and intelligent wells
radius increases and approaches the pressure equivalent grid
are 0.088 and 0.089, respectively for high absolute porosity
radius. The algorithm 3 yielded the following values of PI of
case but increases to 0.092 and 0.100 for CW and IWC in
the well for each 5-year interval over 20 years and it is shown
low absolute porosity case.
in Table 9.
Algorithm 3 is robust in the case of both high and low
From Table 9, the algorithm 3 functioned best over the first
absolute porosities since the cumulative oil produced in both
5 years and poorly in the next 15 years for the best case. The
cases were more than that from a conventional well. The
extremely high PI value of 84.77 could be attributed to the
algorithm was able to withstand the high and low porosity
fact that, stimulation increases the PI value of a well, more
cases because this static parameter generally describes the
especially during the early stages of production after the stimu-
voidage within the formations which in turn tells how much
lation had just been carried out. The best (highest) PI recorded
oil is in place. When in either high or small accumulations,
for the worst case scenario (skin factor of 5) is obtained in 15
the algorithm is able to shut the ICVs to prevent more water
years, and it has a value of 5.69 and the minimum PI recorded
influx and open when oil accumulates in higher proportions in
during the first 5 years. Therefore, algorithm 3 yields higher
the segments unlike the case of convention wells that would
PI values under any form of stimulation that will reduce skin,
produce more water and less oil due to lack of control. Fig.
than when not stimulated or when there is wellbore damage.
13 shows the graph pertaining to the porosity variation for the
CV and IWC.
3.3.4 Variations of oil-water contact
3.3.3 Results of sensitivity of skin variations The oil water contact depth was varied so as to determine
how the size of the aquifer will affect the cumulative oil
The results obtained from varying the skin values from the
production and cumulative water production. The results of
base case skin factor of 0 to a best case skin value of -2 and
 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354 353

Table 10. FOPTs and FWPTs with variations in oil water contact.

Best case scenario (OWC-3595 ft) Worst case scenario (OWC-3,518 ft)
Field parameter
CW IWC CW IWC
FWPT (BBLS) 2,494.06 2,308.21 4,162,732 3,644,840
FOPT 1,254,225 1,196,975 343,822 296,851
(BBLS) 0 0 9 8

water contact since the CW yielded more cumulative oil than

Table 11. Profitability results. the IWC under this particular case.
It was established that, intelligent wells function better
Project type Net present value ($) Present value ratio in terms of reducing water production in edge water drive
IWC 145,015,195.07 12.08 reservoirs with thin pay zones (oil rims) than those with thick
pay zones where coning occurs much later during the life of
the field. This is evident from the results obtained from OWC
the variations are shown in Table 10. at 3,518 ft. At this OWC, the oil column was very thin (13
The intelligent well under varying OWC for both cases ft) and this resulted in the intelligent well performing much
yielded lower cumulative water but could not yield maximum function of reducing water production to more than half a
cumulative oil. This was due to the fact that, the aquifer was million barrels as compared to OWC at 3,595 ft where the
inactive, hence could not yield maximum cumulative oil. For intelligent well reduces water production by just 186 bbls.
oil water contact at 3,595 ft, water control by the ICV was It is recommended that further work should be done on
very effective because it led to little water production from the technical feasibility of IW in oil rim reservoirs using
the reservoir. IWC produced 2,308.21 bbls of water and CW optimisation algorithms such as Genetic Algorithm, PSO etc.
produced 2,494.06 bbls as shown in Table 10. The difference In addition, local grid refinement should be considered when
of about 186 bbls of water, as compared to the difference of modelling near wellbore regions in future studies so as to
517,892 bbls of water in the case of oil water contact at 3,518 clearly visualise the coning phenomenon.
ft. In the worst case scenario, the OWC is closer to the well
resulting in an early water breakthrough which occurred within Acknowledgments
35 days, hence higher water production rate. The authors would like to thank University of Mines and
Technology, Tarkwa, Ghana for supporting this research and
3.4 Intelligent well profitability also thank the Editor-in-Chief and all the reviewers for giving
valuable comments to improve upon this paper.
The viability of the project after the NPV evaluation using
the economic model parameters in Table 4. The computations Open Access This article is distributed under the terms and conditions of
were done using Excel Spreadsheet. In order to account for the the Creative Commons Attribution (CC BY-NC-ND) license, which permits
unrestricted use, distribution, and reproduction in any medium, provided the
uncertainties in oil price, which can rise and fall with time, an original work is properly cited.
underlying assumption was made that, the oil price remained
constant. The intelligent well project after the calculations References
yielded an NPV of $145,015,195.07 which is greater than
zero. Therefore, the application of the intelligent well in oil Adekunle, O.A. Intelligent well applications in production
rim reservoirs for the study is economical (Broni-Bediako, wells. Aberdeen, University of Aberdeen, 2012.
2018). In order to further validate the profitability, present Adusu, P.T. Optimising candidate well selection for matrix
value ratio (PVR) calculations was also done and it resulted stimulation-IPR approach. Paper SPE-198707-MS Pre-
in a PVR value of 12.08 which indicates that the project was sented at SPE Nigeria Annual International Conference
economical. and Exhibition, Lagos, Nigeria, 5-7 August, 2019.
Aitokhuehi, L. Real-time optimization of smart wells.
California, Stanford University, 2004.
4. Conclusions and recommendations Al-Ghareeb, Z.M. Monitoring and control of smart wells.
The results of the research indicated that, a well can be California, Stanford University, 2009.
made intelligent when installed with ICVs and with a proper Amangaliyev, B. Advance workover modelling tool using 3D
RCS. This is evident from the results obtained from the intel- models. SIS Software Bulleting in Caspian Region, 2017.
ligent well coupled with “algorithm 3” where the intelligent Anon. Controlling excess water production.
well yielded higher cumulative oil production as against the Anon. Oil and gas live overview.
conventional well. It can also be concluded that, the developed Aulisa, E., Bloshanskaya, L., Ibragimov, A. Long term
RCS is robust under varying reservoir conditions of porosity, dynamics for well productivity index for nonlinear flows
permeability and skin but unrobust for the case of varying oil- in porous media. J. Math. Phys. 2011, 52(2): 1-26.
354 Broni-Bediako, E., et al. Advances in Geo-Energy Research 2019, 3(4): 343-354

Broni-Bediako, E. Oil and gas project evaluation. Tarkwa, Iskandar, Universiti Teknologi PETRONAS, 2008.
University of Mines and Technology, 2018. Pinto, M.A.S., Barreto, C.E., Schiozer, D.J. Optimization
Chang, Y.L. Simulation study on improved oil recovery for of proactive control valves of producer and injector
thin oil rims. Bandar Seri Iskandar, Universiti Teknologi intelligent wells under economic uncertainty. Paper SPE-
PETRONAS, 2014. 154511-MS Presented at SPE Europec/EAGE Annual
Dilib, F.A., Jackson, M.D., Mojaddam Zadeh, A., et al. Conference, Copenhagen, Denmark, 4-7 June, 2012.
Closed-loop feedback control in intelligent wells: Raoufi, M.H., Farasat, A., Mohammadifard, M. Application
Application to a heterogeneous, thin oil-rim reservoir of simulated annealing optimization algorithm to optimal
in the North Sea. SPE Reserv. Eval. Eng. 2015, 18(1): operation of intelligent well completions in an offshore
69-83. Oil reservoir. J. Pet. Exp. Prod. Technol. 2015, 5(3): 327-
Huang, Z., Li, Y., Peng, Y., et al. Study of intelligent wells 338.
for liaohe field. Proc. Eng. 2011, 15: 739-743. Raoufi, M.H., Mashishi, M. Optimization of flow control
Kumar, M., Sharma, P., Gupta, D.K. Completion design op- with intelligent well completions in a channelized thin
timization of multilateral well to maximize hydrocarbon oil rim. 73rd European Association of Geoscientists and
production in bottom water drive reservoirs. Int. J. Eng. Engineers Conference and Exhibition Incorporating SPE
Dev. Res. 2016, 4(2): 897-906. EUROPEC, Vienna, Austria, 23-26 May, 2011.
Mahmood, N., Sultan, Z., Yousof, N. A review on smart Robinson, M. Intelligent wells completions. J. Pet. Technol.
well completion system: route to the smartest recovery. 2003, 55(8): 57-59.
International Conference on Petroleum Engineering, Sarkodie, K., Afari, S.A., Aggrey, W.N. Intelligent well
Dhaka, Bangladesh, 2016. technology-dealing with gas coning problems in pro-
Masoudi, R., Karkooti, H., Othman, M.B. How to get most out duction wells. Int. J. Appl. 2014, 4(5): 121-135.
of your oil rim. 6th International Petroleum Technology Tarek, A. Reservoir Engineering Handbook 4th edition. UK,
Conference, Beijing, China, 26-28 March, 2013. Elsevier Inc, 2010.
Muhammad, B.R. Pressure as an indicator for water Break-
through for horizontal well completion. Bandar Seri