Вы находитесь на странице: 1из 37

IPA17-347-SE

PROCEEDINGS, INDONESIAN PETROLEUM ASSOCIATION


Forty-first Annual Convention & Exhibition, May 2017

A SIMPLE-EFFECTIVEEFFICIENT ANALYTICAL MODEL FOR


MULTI-WELL GAS LIFT ALLOCATION OPTIMIZATION

Ricky Aditiya Fandi*


Saesarian Izwardy*
Silvya D. Rahmawati*
Pudjo Sukarno*

ABSTRACT based on this analytical model to conduct a


simulation test.
Gas lift is one of the most common artificial lift
methods used in either single or multiple oil Keywords: Gas Lift, Gas Lift Allocation, Gas Lift
producing wells. Because of high gas prices and Performance Curves, Limited Gas Injection Rate
compressing costs, an optimum allocation for each
well is needed to obtain maximum field oil INTRODUCTION
production, particularly where there is limited gas
supply. Gas lift is operated by injecting compressed gas into
the lower section of tubing through the casing-tubing
A new simple-effective-efficient model is annulus and an orifice (mandrel valve) installed in
constructed here which only needs 3 items of data: the tubing string or using PCTGL (Permanent Coiled
the number of gas lift wells, total gas injection rate Tubing Gas Lift). This compressed gas affects liquid
and gas lift performance curve (GLPC). The production in 2 ways: (a) the energy of expansion
developed model uses a fundamental allocation propels the oil to the surface and (b) the gas aerates
method for 2 gas lift wells. By applying first the oil so that the effective density of fluid inside the
differential to total oil production as a function of tubing is less and, thus, easier to get to the surface.
allocation and total gas injection rate, optimum
allocation as a function of total gas injection rate can In Simplified Method for Gas Lift Well Problem
be obtained. Then, for 3 wells, it will be same as for Identification and Diagnosis SPE 5151 which was
2 wells. 1 well is 3rd well and another is the result of published in 1974, Mayhill analyzed the relationship
the previous calculation of 2 gas lift wells and so it between rate of injected gas and production rate of
goes on. By just inputting data of the number of gas the oil and this relationship is called Gas Lift
lift wells, total gas injection rate availability and Performance Curve (GLPC) as shown in figure. 2.
GLPC of each gas lift well, optimum gas injection He defined that the incremental of revenue
allocation for all wells can be directly and simply production rate can be obtained by the incremental of
obtained without doing any iteration or graphical gas injection rate with decreasing ratio. But an
method. excessive gas injection rate would cause a slippage
effect where the gas phase moves faster than liquid,
Compared with Equal Slope which is a long therefore it would not increase production but would
procedure with lots of graphs and limited to a small decrease production.
number of wells, this model demonstrates more
flexibility and faster performance but yields almost Gomez in his M.S. thesis Optimization of
exactly the same results and can be applied to a large Continuous Flow Gas Lift Systems in 1974
number of wells. Compared with the numerical proposed a procedure to generate GLPC which is
method in commercial software, this model uses less fitted to the quadratic polynomial, the equation can
gas injection rate but produces more oil production. be written as:
This model has been also successfully tested on field
data under both unlimited and limited gas supply. A = 2 + + (1)
macro-excel program is then developed by the author

* Institute of Technology Bandung


for each gas lift well using the least square method. obtained. By inputting data of the number of gas lift
This quadratic polynomial GLPC supports Mayhills wells, total gas injection rate availability and
GLPC that there is a certain optimum gas injection quadratic GLPC of each well, the gas injection
rate for a gas lift well which produces maximum allocation for all wells can be directly and simply
revenue. obtained without doing any iteration or graphical
methods.
The challenge then rises as there is more than one
well in the field which has to be injected. By using Compared with Equal Slope which is a long
Gomez quadratic polynomial GLPC, it would be procedure with lots of graphs and is limited to a
easy to find maximum oil production by determining small number of wells, this model demonstrates that
the optimum gas injection rate of each well using it is more simple-effective-efficient and has faster
first derivative. There will not be any problem when performance but yields almost exactly the same
gas supply in the field is unlimited. But the problem results and can be applied for a large number of
arises as gas supply in the field is limited. wells. Compared with the numerical optimization
method in commercial software, this model uses less
In Economic Approach to Oil Production and Gas gas injection rate but produces more oil production.
Allocation in Continuous Gas Lift which has been This model has been also successfully tested on field
published by Kanu et all in 1982, they established data under both unlimited and limited gas supply. A
the Equal Slope allocation method for when there is macro-excel program is then developed by the author
limited gas supply. They used GLPC slope of each based on this analytical model to conduct a
well and discovered that total optimum gas lift simulation case study
injection would be obtained if the GLPC slope at
certain points of each well are equal. Then they used ANALYTICAL FORMULATION FOR 2 GAS
this fundamental to establish a graphical step by step LIFT INJECTION WELLS
procedure. But the weakness of this equal slope
method is its long procedure and it needs at least 3 Using Eq. (1) if there are 2 gas lift wells, quadratic
graphs including GLPC to obtain the allocation for of each quadratic GLPC is shown :
each well. Moreover, for more than 2 gas lift wells,
equal slope will be such a long method and needs 1 (1 ) = 1 12 + 1 1 + 1 (2)
more consistency to maintain the accuracy related to
the using of graphs. 2 (2 ) = 2 22 + 2 2 + 2 (3)

In this paper, a new simple-effective-efficient Where:


allocation model is constructed which only needs
data: (1) the number of gas lift wells, (2) total limited = , produced oil rate; = , gas injection
gas injection rate, and (3) quadratic polynomial rate
GLPC of each gas lift well. This model is constructed
using a fundamental allocation method for 2 gas lift And gas rate to be injected to 2 wells is written as:
wells. By applying first partial derivative to total
produced oil as a function of allocation and total gas 2 = 1 + 2 (4)
injection rate, optimum allocation as a function of
total gas injection rate can be obtained. Then, for 3 Then, adding Eq. (2) to Eq. (3) yields total produced
wells, it would be same as 2 wells. 1 well is 3rd well oil of 2 gas lift wells:
and another is the result of the previous calculation
of 2 gas lift wells and so it goes on. 12 = (1 , 2 ) (5)
This model can directly determine how much gas Entering eq. (4) to (5) with 1 substitution, yields
injection allocation for well in wells system as total produced oil of 2 gas lift wells as a function of
a function of total gas injection rate to be injected to optimum allocation for the second well and gas rate
wells. Then, by getting gas rate to be injected to to be injected to 2 wells:
1 wells which is the rest of the gas after injecting
into the well, it can be directly determined how 12 = (2 , 2 ) (6)
much gas injection allocation for ( 1) well as a
function of gas injection rate to be injected to 1 To obtain maximum 12 , taking partial first
wells. The procedure continues to repeat for other
derivative of Eq (6) 12 = 0 in constant 2
wells until gas injection allocations for all wells are 2
condition, yielding optimum gas allocation for 2nd allocation. By applying the same procedure for 3
well as a function of gas rate to be injected to 2 wells: wells as 2 wells, the author obtains the results

1 1 2 1 +1 2
2 (2 ) = (1 2 + 2 (1 2 ) ) (7) 12 = 22 [ 1+2 ] + 2 [ ]+ [1 +
1 +2 1 2 1 +2
1 ( )2
2 4 (1 +2 ) ] (GLPC of previous 2 wells)
Recalling Eq. (4), the author gets allocation for the 1 2

rest of the wells, in this case this is the first well:\ 3 = 3 32 + 3 3 + 3 (GLPC of 3rd well)

1 = 2 2 (8) 2 12

The 1 and 2 above are optimum allocations for the 3 13


first well and the second well in order to obtain 3 3
maximum oil production. After getting optimum
allocations, maximum oil production as a function of = +
{ + [ (
+
gas rate to be injected to 2 wells can be obtained by
substituting Eq. (7) to Eq. (6), yields: ) + ( ) ]}
(Optimum allocation for 3rd well)
2 1 +1 2
12 (2 ) = 22 [ 1+2 ] + 2 [ 1 +2
]+ The third terms at above equation is neglected
1 2
1 ( )2 because it will not be used for the next process.
[1 + 2 4 (1 +2 ) ] (9)
1 2
The equation is then derived for nth number of wells.
With Eq. (9), maximum oil production by injecting Optimum gas lift allocation for well as function
certain gas rate with certain optimum allocation can of total gas rate to be injected to n-wells, , as
be known. Then if we see further, Eq. (9) looks like shown :
another quadratic GLPC as shown in Eq. (1).

Formula Derivation
( ) = [ ( )

=
= (
)
=
From previous derivation, the author can make a
summary of system 2 gas lift injection wells with a
certain total gas injection rate, 2 as shown
+ ( )]

= =
GLPC of 2 wells
(Ricky Model) (10)
1 = 1 12 + 1 1 + 1 (GLPC of 1st well)
2 = 2 22 + 2 2 + 2 (GLPC of 2nd wel) and

1 1 = (11)
2 12
with limitation such as :
2 2
GLPC is quadratic polynomial
= ( + ( ) )(Optimum
+ 2 of GLPC is more than 95 %
allocation for 2nd well) Positive sloped section of GLPC is used
2 1 +1 2
12 = 22 [ 1+2 ] + 2 [ 1 +2
]+ [1 + For the next explanation, the author calls the model
1 2
1 (1 2)2 the Ricky Model.
2 ]
4 (1 +2 )
ALGORITHM OF THE MODEL
Then, using the summary above, the gas allocation
method for 3 wells to be conducted in a similar way Let us say that there are n wells in the field with total
as the following: Simplify the 3 wells problem into a gas injection rate, , using Eq. (10) optimum gas lift
2 wells problem - 1 well is 3rd well and another is the allocation for well, , as function of gas
result from previous calculation of 2 wells gas lift injection rate for n wells, can be obtained. Then
using Eq. (11) obtaining the rest rate of gas supply to production will not be used as it is not efficient
be injected to (n-1) wells, 1 , after some gas has anymore.
been injected to well. Then using Eq. (10) again,
obtaining optimum gas lift allocation for ( 1) The result of the Ricky model is shown in the table
well, 1 , as function of gas supply for (n-1) wells, 2.
1 . Then the procedure continues repeating until
allocations for all wells are obtained. Thus, all From the table 2, with total gas injection rate 4
allocations should satisfy the equation . MMSCFD and certain GLPC of each well as shown
in the figure 4, the author obtained optimum
allocation for each well which would produce
=1 = (12) maximum oil production rate of 10226.9 BOPD.

The algorithm of the Ricky model is shown . Sensitivity to Gas Injection Allocation

1 2 To make sure that the allocations shown in the table


= = 1 2 are the most optimum allocations, the author
1 performs sensitivity and makes variation in
( ) 1 (1 ) 2 (2 ) 2 (2 ) allocation data but with the same amount in total gas
injection rate of 4 MMSCFD. The sensitivity results
are shown in the table 3 .
1
= 2 In same amount of total gas injection rate, the
Where
2 sensitivity yields lower oil production than the result
obtained by Ricky model which is lower than
= gas allocation for well
10226.9 BOPD. The result supports the theory that
the allocation result which is obtained by the model
To use this model, the user should input GLPC
is the most optimum gas injection allocation.
values of each well and the number of gas lift wells.
After all data has been inputted, the model will
Injection with Unlimited Gas Supply
process it to obtain optimum gas injection allocation
results for all wells. But the user should do an
When gas supply in the field is unlimited, then to
additional algorithm if one or more gas lift wells are
obtain total maximum oil production, each well
zero allocated, which means that those wells should
should be produced in a maximum oil rate. This will
not be gas lifted. The User is to eliminate those wells
be obtained if the gas injection rate of each well is
and repeat the calculation without including them
optimum which is the highest point in GLPC. The
until all wells which are included have been non-zero
highest point in GLPC has zero gradient in which
allocated. Figure 3 shows the algorithm to use the
gradient is obtained by first derivative.
model.
Quadratic GLPC of the simulation case:
SIMULATION AND ANALYSIS
1 = 804.2312 + 1673.71 + 867.62
Simulation using Published Data
2 = 787.1422 + 1686.32 + 580.24
The simulation of a gas lift field consisting of 6 wells 3 = 716.8632 + 16873 + 939.32
is constructed here. The simulation data is obtained 4 = 412.1642 + 1332.24 + 939.22
from Kanu et all. The gas supply is limited at 4 5 = 724.0752 + 1769.15 + 780.66
MMSCFD. There are 6 six points in each well to 6 = 883.5262 + 2090.36 + 1197.5
construct GLPC. Table 1 shows the data of the gas
lift performances of each well. Taking zero first derivative of each GLPC, the author
gets optimum allocation for each well for unlimited
GLPC of 6 gas lift wells shown in figure 4 is satisfied gas supply as shown in table 4 :
with quadratic polynomial proposed by Gomez with
2 > 0.96 in this case. To make this more satisfying The table 4 shows that to obtain maximum total oil
in quadratic GLPC, it is suggested to plot the data in production, 7.31 MMSCFD gas injection rate is
the curve until maximum oil production point, needed with certain allocation for each well as shown
because the data which is higher than maximum oil above. Then, if the point of view of the case above is
changed, let gas supply in the field be 7.31 make any error because it is always related with
MMSCFD. With the Ricky model, optimum numbers and algebra.
allocations are obtained and shown in the table 5 :
The result is absolutely matched with the previous On the other hand, with this Ricky method, the
results which uses highest point of GLPC as an operation is not only much more simple-effective-
optimum gas injection allocation. efficient and faster, but also very valid. This method
just needs one graph which is GLPC, and input data
Comparison with Equal Slope Allocation Method of the number of gas lift wells and total gas injection
rate and the Ricky method yields very matched gas
Because the equal slope method has commonly been injection allocation compared with equal slope. If
used for many years, it is desirable to compare the there are any alterations in input data, the user can
Ricky model with it. Actually, either the Ricky easily change the data, and the new result will appear
model or equal slope method uses the same immediately. The model is therefore very flexible to
fundamentals that is gradient or first derivative of be applied from one gas lift system to others. In
GLPC. But related with simplicity, operation speed comparison,if there are any alteration in input data in
and efficiency, these models are good to be the equal slope method, it will be significantly harder
compared. work because the user needs to draw the graphs
again.
From GLPC which is given in the simulation case,
the equal slope method establishes first derivate to Simulation using Field Data
get gradient of GLPC as a function of gas injection
rate. Figure 5 shows the relationship between The author uses an oil and gas field which is located
gradient vs gas injection rate for each well. Then, in East Kalimantan., named field X. The field is a
establish the master plot, shown in figure 6, which is mature field with hundreds of wells where almost all
gradient of GLPC vs total gas injection rate. Total of them use either conventional gas lift or PCTGL
gas rate itself is the sum of gas injection rates of all (Permanent Coiled Tubing Gas Lift) as their artificial
wells associated with the same gradient. lift. Table 8 and Table 9 in the Appendix shows
wellbore and reservoir characteristics of the field.
By the equal slope method, for total gas injection gas
rate of 4 MMSCFD as discussed in the simulation Production History Matching
case, gradient 750 is obtained. For sensitivity, total
gas injection rate of 5 MMSCFD yields gradient 525. In this simulation, the author uses PROSPER and
Then by inputting these numbers back to figure 5, GAP software from Petroleum Expert. The algorithm
allocation for all wells can be obtained. Table 6 starts with constructing the wellbore model using
shows the results of the Ricky model and equal slope PROSPER, then continued by matching process.
method as a comparison. After constructing all wells, the author constructs a
pipeline network model using GAP, then continued
by matching process. Running simulation is done
As shown in the table 6, the result between the Ricky after both wellbore model and pipeline network
model and equal slope are very matched. This result model have been constructed. Figure 11 shows the
shows that the Ricky model is very valid although the algorithm.
equal slope method has also been valid for many
years. However, the Ricky model offers a simple- Well test data which is used in the simulation is the
effective-efficient way to directly and simply obtain latest data taken by the company. Table 10 in the
optimum gas injection allocation to produce Appendix shows well test data.
maximum total oil production without doing any
graphical method that the equal slope uses. Constructing Wellbore Model
Specifically, the equal slope method needs to
establish at least 3 graphs including GLPC to get Almost all wells in Field X have dual-monobor
optimum allocation for each well. More wells makes completion as the reservoir is braded type. So lots of
equal slope more difficult to be done because lots of bottom-up perforation should be done alongside
graphs have to be constructed. Moreover, in the production. This dual-monobore completion will
graphical method, the error is always bigger than the make it easy for engineers to do bottom-up
analytical method for sure. The graphical error often perforation without using any packer. Figure 12
occurs when the process is not computerized while shows a well sketch of a dual-monobore completion
the analytical method, computerized or not, will not system
The algorithm to construct the wellbore model is As can been seen in Table 11 above, the biggest
shown in figure 13 deviation from matching process is 1%. So wellbore
Well system data consists of: models which have been constructed can be
accepted.
1. completion
2. production method Figure 15 and Table 12 shows GLPC of all wells
after matching process is finished with their
PVT data consists of: quadratic polynomial. 2 value is more than 96%.

1. oil API Constructing Pipeline Network Model


2. SG gas
3. Water salinity The algorithm to construct the pipeline network
4. GOR. model is shown in Figure 16
Downhole survey consists of: We started by creating the pipeline network as shown
in figure 17. In the field, the pipeline network is a
1. Geothermal survey complicated fishbone-like which has lots of cross
2. Well depth section, elbows, fitting and chokes. Otherwise in the
3. Bottomhole eqipment simulation model, the author created simple
fishbone-like network and all restrictions are
Gas lift data consists of: explained in matching process sub-chapter. There are
2 major fishbone-like pipeline networks, the upper
1. SG gas lift one is connected to 25 psig Separator-1, the lower
2. Injection depth one is connected to 25 psig Separator-2.
3. Gas lift device specification
Table 13 in the Appendix shows well distance to
IPR data consists of: major pipe in the field.
1. Reservoir pressure Matching Pipeline Network Model
2. PI (input before matching is 2)
3. Skin (input before matching is 0) There are several reasons why pipeline network
4. Watercut should be matched, such as:
There are several assumptions which are used in the 1. Fishbone-like model in the simulation is a simple
simulation, such as: form of actual pipeline network.
1. Using PI input method for IPR construction, PI
2. There are lots of fittings, elbows, cross section
= 2 for first input
and other restrictions which are not recorded in
2. Damage skin < 3, damage skin = 0 for first input
the database.
3. GOR = 300 if data is unavailable
4. Water Salinity = 1000 ppm
3. In the field, there are actually more than 14
5. No impurities data
producing wells, but only 14 wells which are gas
6. Ambient temperature = 80 F
lifted and only those 14 gas lift wells are
7. Gaslift gas gravity = 0.7
included in the model.
8. Casing injection pressure = 1300 psig
Because of these reasons, additional chokes are
Matching process was done by adjusting the value of
created and placed close to each wellhead to
the most uncertain variable because there is not any
overcome the problem above. ID choke is then
direct-measurement to that variable. In the case,
variable that author adjusted in order that the pipeline
those variables are PI and skin. Matching criteria in
network model is well-matched with well test data.
this process is that deviation of oil rate between
Matching criteria in this process is that deviation of
simulation and well test data in certain FTHP
oil rate and FTHP between simulation and well test
condition is less than 2%. Figure 14 shows the
data is less than 10%. Figure 18 shows the algorithm
algorithm of the matching process.
of pipeline network model matching
Then matching result is shown in the table 11 .
As can be seen from table 14 above, the matching GLPC of each gas lift well without doing any
result can be well-accepted as maximum deviation is iteration or graphical method.
10%
2. This model is proven to be as valid as the equal
Simulation Results slope method and GAP with deviation of 0.27%
Simulation is divided into 2 major parts, the first part 1.77%, but it is much more simple-effective-
is to compare Ricky Model and GAP with actual data efficient and faster.
from the field, and the second part is to compare
Ricky Model with GAP when certain amounts of gas 3. Any error in the new analytical model is greatly
are available. The first part is named Case-1, and the minimized compared to using the graphical
second part consists of Case-2 (gas available 4.5 method. This model is also an explicit algorithm
MMscfd), Case-3 (gas available 3 MMscfd) and which is a one way calculation, so it will be much
Case-4 (gas available 2 MMscfd) faster than the numerical method and there will
not be any worry of divergence.
From case-1, it is known that Ricky Model and GAP
shows relatively the same result of both gas 4. If there are any alterations in input data, the user
consumption and oil production. And compared with can easily change the data, and the new result
actual data, to produce relatively the same oil will appear immediately. So the model is very
production, the Ricky Model can save about 2.5 flexible to be applied from one gas lift system to
MMscfd (42%) gas consumption than actual data. others.
is the results from all cases: case 1, case 2, case 3
and case 4. While the graph for case 2, case 3 and 5. To produce relatively the same oil production in
case 4 are shown in the Appendix. field X, the Ricky Model can save about 2.5
MMscfd (42%) gas consumption when
compared to actual data.

SUGGESTIONS

1. In particular, the objective of the model can be


changed from maximum oil production to
maximum economic profitability.
From case-2, case-3 and case-4, they show the same
trend which is that the Ricky Model and GAPs 2. Good result for wells which have distance of
results are relatively the same with maximum less than 3000 ft to the separator
deviation of 1.77%. It is a very good result as using
the Ricky Model is much simpler than using GAP in 3. Minimum gas injection rate can be changed
daily operation. With the Ricky model it doesnt from 0 MMscfd to minimum kick off injection
need to do any iteration method like numerical
method does in GAP. Because of that, it doesnt need 4. Maximum gas injection rate can be changed
to worry about any divergence as the Ricky model is from the highest point of GLP to the maximum
an analytical model. So it is always convergence, injection capacity by considering the
after input of the data then the result will always economical point of view
appear. Beside that, and from a computerization
point of view, an analytical model like the Ricky 5. Splitting a GLPC into several quadratic
model will not consume much memory. So the polynomial to achieve more than 95% 2
solving process will be light even for a field with lots
of wells and complex configuration. NOMENCLATURE

CONCLUSIONS , = oil production rate,


, = gas injection rate,
1. A Simple-Effective-Efficient Analytical Model = total gas injection rate,
to determine optimum multi gas lift well , , = quadratic GLPC coefficient
allocation is established which only requires the = the number of wells
following data: (1) the number of gas lift wells, FTHP= Flowing tubing head pressure
(2) total limited gas volume, and (3) quadratic RM = Ricky Model
Dev = deviation optimum allocations for all those wells also change?
There may be also be a question about which GLPC
REFERENCES should be inputted to Ricky Model as different
pressure different GLPC.
Fitra, Ulfia R., Rahmawati, Silvya D., Sukarno, P.,
Soewono, E. 2015. Optimization of Gas Lift In this case, the author has input some dummy data
Allocation in Multi-well System, A Simple in one well which has several FTHP data. FTHP
Numerical Approach. Proceedings of 39th Annual range is 50 to 210 psig. As shown in figure 35, when
Convention and Exhibition Indonesian Petroleum FTHP is increased GLPC will go down.
Association (IPA), May 2015.
GLPC above is then matched with quadratic
Gomez, V. 1974. Optimization of Continuous Flow polynomial with 2 more than 97%, as shown in the
Gas Lift Systems. M.S. Thesis, U. of Tulsa. table 15.
Guo, B., Lyons, W.C., and Ghalambor, A. 2007.
Petroleum Production Engineering A Computer Refers to Eq 1, then Author made curve between
Assisted Approach. Gulf Professional Pub
, , and 2 versus FTHP, as the result shown in
Kanu, E. P., Mach, J., and Brown K. E. 1982. figure 36.
Economic Approach to Oil Production and Gas
Allocation in Continuous Gas Lift. J.Pet. Tech, It can be seen from figure 36 above that a, b, and c

February 1982, pp. 1886-1892 change when FTHP change, but 2 does not
change significantly. Even on the first look, author
Mayhill, T. D. 1974. Simplified Method for Gas Lift
can say that curve 2 is straight horizontal. What
Well Problem Identification and Diagnosis. SPE
5151. SPE 49th Annual Fall Meeting, Houston, TX, does it mean?
Oct 6-9, 1974.
Referring to derivation of Ricky Model in sub-
Suhendar, Ade Diar. 2015. Optimisasi Distribusi chapter 2, the most fundamental thing in determining
Jaringan Gas Lift dengan Integrated Asset optimum allocations is first partial derivative. While

Modelling. Thesis for Master Degree Program. variable is first partial derivative of quadratic
2
Petroleum Engineering Department, Institut polynomial. Refers to figure 36 that shows value of
Teknologi Bandung
2 is relatively constant to FTHP, so the author
APPENDIX made an initial hypothetic that FTHP will not
influence optimum allocations for a multi-well gas
Additional SimulationSensitivity FTHP to Gas lift system.
Allocation
The author then did the following simulation to prove
As almost all gas lift operations consists of lots of initial hypothetic using some section of field X as
wells, when there is any condition change in even shown in figure 37
one well, the other wells must be influenced as long
as they are in one network. Usually it is called the This simulation consists of 4 gas lifted oil producing
back pressure effect. When the gas injection rate for wells in connection with 1 separator. FTHP of each
a certain well is changed, increased or decreased, it well will be changed by adjusting pipe length from
must make FTHP (flowing tubing head pressure) well X-196 and well X-919 to main pipe with detail
change. In condition for oil production can be still as shown in Table 16. Then with 2 sensitivities of
increased by injection more gas, the more gas total gas injection rate, 1 MMscfd and 1.3 MMscfd,
injection rate the less liquid density inside tubing, so the result is shown in figure 38 41
FTHP will be also decreased and vice versa.
From both simulation above, it can be seen that
On the other side when FTHP change, GLPC also adjusting pipe length in well X-196 and X-919 will
changes as shown in figure 35, so it must be no change FTHP of all wells (4.2% - 4.7%), while
question about it. However, the big question is optimum allocations are relatively constant
whether optimum allocation for the system also (maximum significant changing is 1.1%). This good
changes. Moreover in multi-well gas lift operations, result has proved that changing in FTHP of each
when FTHP of several wells may change, do wells in certain system will not change optimum
allocations. It also shows that GLPC which is condition, the result of optimum allocations will be
inputted to the Ricky Model is GLPC in any FTHP same.
TABLE 1

GLPC DATA OF 6 WELLS (KANU ET ALL, 1982)

Well 1 Well 2
Qg Qo Qg Qo
MMSCFD BOPD MMSCFD BOPD
2.84 1420 2.62 1310
1.66 1660 1.42 1420
1.34 1680 1.14 1430
1.02 1700 0.85 1415
0.66 1660 0.53 1330
0.34 1330 0.16 800
Well 3 Well 4
Qg Qo Qg Qo
MMSCFD BOPD MMSCFD BOPD
3.32 1660 3.9 1755
1.85 1850 1.99 1990
1.5 1875 1.57 1960
1.13 1880 1.14 1900
0.74 1840 0.7 1750
0.27 1330 0.24 1200
Well 5 Well 6
Qg Qo Qg Qo
MMSCFD BOPD MMSCFD BOPD
3.51 1755 3.7 1850
1.83 1830 2.25 2250
1.47 1835 1.85 2310
1.07 1790 1.43 2380
0.68 1700 0.96 2390
0.22 1120 0.37 1850

TABLE 2

ALLOCATION SIMULATION RESULTS BY RICKY MODEL

Qg Qo
MMSCFD BOPD
Well 1 0.579 1567.4
Well 2 0.592 1302.7
Well 3 0.655 1736.7
Well 4 0.709 1676.4
Well 5 0.705 1668.1
Well 6 0.760 2275.5
Total 4.000 10226.9
TABLE 3

SENSITIVITY RESULTS TO ALLOCATION BY RICKY MODEL

Case 1 Case 2
Qg Qo Qg Qo
MMSCFD BOPD MMSCFD BOPD
Well
0.71 1650.5 0.58 1567.8
1
Well
0.59 1302.7 0.71 1380.7
2
Well
0.65 1736.7 0.65 1736.7
3
Well
0.58 1573.2 0.59 1581.7
4
Well
0.71 1668.1 0.71 1668.1
5
Well
0.76 2275.5 0.76 2275.5
6
Total 4.00 10206.8 4.00 10210.6
Case 3 Case 4
Qg Qo Qg Qo
MMSCFD BOPD MMSCFD BOPD
0.58 1567.8 0.77 1679.5
0.59 1301.2 0.71 1380.7
0.76 1807.4 0.76 1807.4
0.59 1581.7 0.59 1581.7
0.71 1668.1 0.59 1572.4
0.77 2283.2 0.58 2112.7
4.00 10209.4 4.00 10134.4

TABLE 4

ALLOCATION RESULTS IN UNLIMITED GAS SUPPLY CONDITION USING HIGHEST


POINT IN GLPC

Gas Injection Rate (MMSCFD)


Well Well Well Well Well Well
Total
1 2 3 4 5 6
1.04 1.07 1.18 1.62 1.22 1.18 7.31
TABLE 5

ALLOCATION RESULTS IN UNLIMITED GAS SUPPLY CONDITION USING RICKY MODEL

Qg Qo
MMSCFD BOPD
Well 1 1.04 1738.4
Well 2 1.07 1483.4
Well 3 1.18 1931.8
Well 4 1.62 2015.7
Well 5 1.22 1861.3
Well 6 1.18 2433.8
Total 7.31 11464.4

TABLE 6

COMPARISON OF ALLOCATION RESULTS USING RICKY MODEL AND EQUAL SLOPE


METHOD

Qg, 4 MMSCFD Qg, 5 MMSCFD


Equal Equal
Ricky Ricky
Slope Slope
Model Model
Method Method
Well 1 0.579 0.575 0.720 0.718
Well 2 0.592 0.595 0.736 0.740
Well 3 0.655 0.655 0.813 0.815
Well 4 0.709 0.705 0.983 0.988
Well 5 0.705 0.705 0.861 0.860
Well 6 0.760 0.760 0.888 0.890
Total 4.000 3.995 5.000 5.011
TABLE 8

WELLBORE CHARACTERISTIC DATA

Depth, Inj Depth, Pres,


No Well String
ft ft psig
1 X-107 LS 5247 3000 1750
2 X-147 LS 5226 4200 1700
3 X-148 LS 1500 1400 350
4 X-157 LS 5753 2400 695
5 X-170 SS 2600 2450 700
6 X-172 LS 1682 1400 400
7 X-175 LS LS 2394 1300 900
8 X-175 SS SS 3888 1100 630
9 X-189 SS 3342 2400 1135
10 X-905 SS 3182 2800 700
11 X-916 SS 2750 750 877
12 X-919 LS 5600 3100 900
13 X-938 LS 4754 4600 992
14 X-940 LS 1592 1400 527

TABLE 9

RESERVOIR CHARACTERISTIC DATA

GOR, scf/STB 161 - 1000


WC, % 0 - 92
Oil API 30.2 - 51.1
SG gas 0.659 - 0.784
Water Salinity, ppm 1000
Depth, ft 1500 - 5753
Inj Depth, ft 750 - 4600
Tres, F 1300 - 201.7
Pres, psig 350 - 1750
TABLE 10

WELL TEST DATA

Last test
Well String FTHP, Qg, Qo,
Ref Date
psig MMscfd BOPD
X-107 LS 2/17/2015 140 0.5 101
X-147 LS 2/19/2016 90 0.5 187
X-148 LS 4/27/2016 155 0.341 50
X-157 LS 11/22/2015 100 0.5 604
X-170 SS 3/16/2016 150 0.207 201
X-172 LS 7/13/2015 85 0.35 50
X-175 LS 9/10/2015 110 0.28 86
X-175 SS 2/2/2016 125 0.25 528
X-189 SS 4/16/2016 220 0.402 367
X-905 SS 3/27/2016 140 0.6 245
X-916 SS 2/8/2016 140 0.3 432
X-919 LS 4/14/2016 60 0.528 230
X-938 LS 2/1/2016 110 0.5 346
X-940 LS 4/8/2016 200 0.89 432

TABLE 11

WELLBORE MODEL MATCHING RESULT


TABLE 12

QUADRATIC POLYNOMIAL GLPC OF EACH WELL

Well GLPC : + +
X-107 -75.133 x2 + 150.82 x+ 40.122
X-147 -1209.3 x2 + 610.79 x+ 106.53
X-148 -23.86 x2 + 37.138 x+ 42.265
X-157 -96.188 x2 + 185.64 x+ 461.74
X-170 -6.8712 x2 + 10.092 x+ 18.208
X-172 -22.039 x2 + 39.388 x+ 47.717
X-175 LS -148.7 x2 + 188.97 x+ 719.56
X-175 SS -685 x2 + 409.2 x+ 1244.1
X-189 -197.96 x2 + 334.9 x+ 352.06
X-905 -13.549 x2 + 30.012 x+ 104.11
X-916 -301.89 x2 + 338.67 x+ 1535.6
X-919 -158.51 x2 + 256.19 x+ 105
X-938 -90.977 x2 + 168.3 x+ 258.84
X-940 -127.83 x2 + 213.16 x+ 439.69

TABLE 13

LIST OF WELL DISTANCE TO MAJOR PIPELINE

No Well Station Distance (m)

1 X-107 Sep-1 107


2 X-147 Sep-2 276
3 X-148 Sep-1 58
4 X-157 Sep-2 173
5 X-170 Sep-2 400
6 X-172 Sep-1 455
7 X-175 LS Sep-1 139
8 X-175 SS Sep-1 139
9 X-189 Sep-1 6444
10 X-905 Sep-2 456
11 X-916 Sep-1 208
12 X-919 Sep-2 95
13 X-938 Sep-2 143
14 X-940 Sep-2 283
TABLE 14

RESULT OF PIPELINE NETWORK MODEL MATCHING

Qg FTHP Oil Rate


Actual Actual Calc Dev Actual Calc Dev
X-107 0.5 140 140.73 1% 101 92.6 -8%
X-147 0.5 90 90.31 0% 187 206 10%
X-148 0.341 155 153.3 -1% 50 51.3 3%
X-157 0.5 100 97.86 -2% 604 608.5 1%
X-170 0.207 150 152.56 2% 201 201.8 0%
X-172 0.35 85 89.41 5% 50 48.6 -3%
X-175
0.28 110 123.59 12% 86 81.2 -6%
LS
X-175
0.25 125 123.59 -1% 528 532.2 1%
SS
X-189 0.402 220 222.8 1% 367 360.8 -2%
X-905 0.6 140 136.75 -2% 245 249.1 2%
X-916 0.3 140 150.27 7% 432 414.1 -4%
X-919 0.528 60 59.5 -1% 230 230.8 0%
X-938 0.5 110 105 -5% 346 340.6 -2%
X-940 0.89 200 199.21 0% 432 434.5 1%

TABLE 15

QUADRATIC POLYNOMIAL OF GLPC IN FIGURE 35

FTHP GLPC R2
50 y = -925.3x2 + 1629.9x + 1264.8 0.983
70 y = -936.57x2 + 1678.7x + 1177.3 0.983
90 y = -969.37x2 + 1738.1x + 1090.8 0.983
110 y = -977.51x2 + 1761.5x + 1011.9 0.981
130 y = -987.08x2 + 1780.6x + 937.2 0.98
150 y = -1002.2x2 + 1804x + 863.34 0.98
170 y = -1012.8x2 + 1822.5x + 790.25 0.98
190 y = -1018x2 + 1829.8x + 723.13 0.978
210 y = -1020.2x2 + 1833.9x + 656.21 0.977
TABLE 16

PIPE LENGTH ADJUSTING

Case 1 Case 2

Pipeline X-196 81 ft 200 ft

Pipeline X-919 1273 ft 700 ft


Figure 1 - Schematic of Gas Lift System in an Oil Field[4]

Figure 2 - Gas Lift Performance Curve (Mayhill, 1974)


Figure 3 - Algorithm to use the Ricky Model
Welll 1 : y = -804.23x2 + 1673.7x + 867.62
3000
R = 0.9647
Well 2 : y = -787.14x2 + 1686.3x + 580.24
2500 R = 0.9616

2000
Qo (BOPD)

1500

Well 3 : y = -716.86x2 + 1687x + 939.32


1000 R = 0.9767
Well 4 : y = -412.16x2 + 1332.2x + 939.22
R = 0.9714
Well 5 : y = -724.07x2 + 1769.1x + 780.66
500
R = 0.9814
Well 6 : y = -883.52x2 + 2090.3x + 1197.5
R = 1
0
0 0.5 1 1.5 2 2.5
Qg (MMSFD)
Well 1 Well 2 Well 3 Well 4 Well 5 Well 6

Figure 4 - Quadratic GLPC of 6 Gas Lift Wells in Simulation Case


1750

1500

Gradient (BOPD/MMSCFD)
1250

1000

750

500

250

0
0.0 0.5 1.0 1.5 2.0
Qg (MMSCFD)

Well 1 Well 2 Well 3


Well 4 Well 5 Well 6

Figure 5 - Gradient vs Gas Injection Rate


1100
1000
Gradient (BOPD/MMSCFD)

900
800
700
600
500
400
300
200
100
0
2.0 3.0 4.0 5.0 6.0 7.0 8.0
Total Qg (MMSCFD)

Figure 6 - Equal Slope Master Plot

Figure 11 - Algorithm of Analyzing


Figure 12 - Well sketch of dual-monobore completion system (Ade Diar Suhendar, 2015)
Figure 13 - Algorithm to Construct Wellbore Model

Figure 14 - Algorithm of Wellbore Model Matching


1800

X-107
1600
X-147

X-148
1400
X-157

1200 X-170

X-172
Qo, BOPD

1000
X-175 EE

800 X-175 FF

X-189
600 X-905

X-916
400
X-919

200 X-938

X-940
0
0 0.5 1 1.5 2
Qg, MMscfd

Figure 15 - Gas lift performance curve of each well

Construct Import IPR


pipeline & and TPR Generate
Run
surface model IPR and
simulation
facilities from TPR in GAP
network Prosper

Figure 16 - Gas lift performance curve of each well


Figure 17 - Pipeline network model

Figure 18 - Algorithm of pipeline network model matching

Case-1 : Compare with actual data from the field


Gas Allocation

0.9

0.8
Gas Allocation, MMscfd

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
X-107 X-147 X-148 X-157 X-170 X-172 X-175 X-175 X-189 X-905 X-916 X-919 X-938 X-940
LS SS

Actual GAP RM

Figure 19 - Optimum gas allocation case 1

Oil Production

700

600
Oil Production, STBd

500

400

300

200

100

Actual GAP RM

Figure 20 - Oil production of each well case 1


Qg Total, MMscfd

0
Actual GAP RM
Qg Total 6.148 3.676 3.674

Figure 21 - Comparison of total gas injection case 1

Qo Total, STBD

4500

4000

3500

3000

2500
Dev = -0.71 %
2000

1500

1000

500

0
Actual GAP RM
Qo Total 3859 3980.4 3952.3

Figure 22 - Comparison of total oil production case 1


Case 2 : Using gas available 4.5 MMscfd
Gas Allocation

1
0.9
0.8
Gas Allocation, MMscfd

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
X-107 X-147 X-148 X-157 X-170 X-172 X-175 X-175 X-189 X-905 X-916 X-919 X-938 X-940
LS SS

GAP RM

Figure 23 - Optimum gas allocation case 2

Oil Production

700

600
Oil Production, STBd

500

400

300

200

100

0
X-107X-147X-148X-157X-170X-172X-175 X-175 X-189X-905X-916X-919X-938X-940
LS SS

GAP RM

Figure 24 - Oil production of each well case 2


Qg Total, MMscfd

4.5

3.5

2.5

1.5

0.5

0
GAP RM
Series1 4.434 4.434

Figure 25 - Comparison of total gas injection case 2

Qo Total, STBD

4500

4000

3500

3000

2500 Dev = -0.27 %

2000

1500

1000

500

0
GAP RM
Series2 3924.4 3914
Figure 26 - Comparison of total oil production case 2

Case 3 : Using gas available 3 MMscfd


Gas Allocation

0.6

0.5
Gas Allocation, MMscfd

0.4

0.3

0.2

0.1

0
X-107 X-147 X-148 X-157 X-170 X-172 X-175 X-175 X-189 X-905 X-916 X-919 X-938 X-940
LS SS

GAP RM

Figure 27 - Optimum gas allocation case 3


Oil Production

700

600

500
Oil Production, STBd

400

300

200

100

0
X-107 X-147 X-148 X-157 X-170 X-172 X-175 X-175 X-189 X-905 X-916 X-919 X-938 X-940
LS SS

GAP RM

Figure 28 - Oil production of each well case 3


Qg Total, MMscfd
3.5

2.5

1.5

0.5

0
GAP RM
Qg Total 2.999 3

Figure 29 - Comparison of total gas injection case 3

Qo Total, STBD
4500

4000

3500

3000

2500 Dev =
-1.77 %
2000

1500

1000

500

0
GAP RM
Qo Total 3994.8 3924.2

Figure 30 - Comparison of total oil production case 3


Case 4 : Gas available 2 MMscfd
Gas Allocation

0.4

0.35

0.3
Gas Allocation, MMscfd

0.25

0.2

0.15

0.1

0.05

0
X-107 X-147 X-148 X-157 X-170 X-172 X-175 X-175 X-189 X-905 X-916 X-919 X-938 X-940
LS SS

GAP RM

Figure 31 - Optimum gas allocation case 4

Oil Production

700

600
Oil Production, STBd

500

400

300

200

100

0
X-107 X-147 X-148 X-157 X-170 X-172 X-175 X-175 X-189 X-905 X-916 X-919 X-938 X-940
LS SS

GAP RM

Figure 32 - Oil production of each well case 4


Qg Total, MMscfd
2.5

1.5

0.5

0
GAP RM
Qg Total 2.002 2

Figure 33 - Comparison of total gas injection case 4


Qo Total, STBD
4500

4000

3500

3000

2500 Dev =
-1.15 %
2000

1500

1000

500

0
GAP RM
Qo Total 3896.7 3851.9

Figure 34 - Comparison of total oil production case 4


Figure 35 - Sensitivity GLPC to FTHP

2500

2000

1500

1000

500

-500

-1000

-1500
0 50 100 150 200 250
FTHP, psig

a b c -b/(2a)


Figure 36 - , , 2 vs FTHP
Figure 37 - Some section of field X

Simulation 1 : Total 1.3 MMscfd


FTHP
75
70
65
60
FTHP, psig

55
50
45
40
35
30
25
Well - X-114 Well - X-147 Well - X-196 Well - X-919
FTHP Case 1 59.47 28.77 67.65 69.7
FTHP Case 2 60.14 28.77 70 69.17

Figure 38 - FTHP in simulation 1


Optimum Gas Injection Rate
0.7

0.6

0.5
Qg, MMscfd

0.4

0.3

0.2

0.1

0
Well - X-114 Well - X-147 Well - X-196 Well - X-919
Qg Case 1 0.388 0.165 0.619 0.13
Qg Case 2 0.386 0.165 0.619 0.132

Figure 39 - Optimum Gas Injection Rate in Simulation 1

Simulation 2 : Total 1 MMscfd

FTHP
70

65

60

55
FTHP, psig

50

45

40

35

30

25
Well - X-114 Well - X-147 Well - X-196 Well - X-919
Series1 54.13 28.61 61.74 63.52
Series2 54.77 28.61 64.04 63.07

Figure 40 - FTHP in simulation 2


Optimum Gas Injection Rate
0.6

0.5

0.4
Qg, MMscfd

0.3

0.2

0.1

0
Well - X-114 Well - X-147 Well - X-196 Well - X-919
Series1 0.264 0.132 0.539 0.028
Series2 0.261 0.132 0.541 0.029

Figure 41 - Optimum Gas Injection Rate in Simulation 2

Ricky Model Gas Lift Allocation Optimizer Software

Figure 42 - Ricky Model Gas Lift Allocation Optimizer Software

Вам также может понравиться