I Society of Petroleum Engineers

SPE 28749

Production Performance of a Retrograde Gas Reservoir:

A Case Study of the Arun Field
Deddy Afidick, N.J. Kaczorowski, and Srinivas Bette, Mobil Oil lndonesia Inc.
SPE Members

Copyright 1994. Society of Petroleum Engineers, Inc.

This paper was prepared for presentation at the SPE Asia Pacific Oil & Gas Conference held in Melbourne, Australia. 7-10 November 1994.

This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author@).Contents of the paper.
as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Papers presentedat SPE meetings are subject to publication review by Editorial Committeesof the Society
of Petroleum Engineers. Permissiin to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract should contain conspicuous acknowledgment
of where and by whom the paper is presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 750833836, U.S.A. Telex 163245 SPEUT.

ABSTRACT
apply for two phase conditions of a gas condensate well,
several semi-analytrcal and numerical methods were
The Arun field is one of the world's giant retrograde gas
developed. Here, our intension is not to present an exhaustive
reservoirs. Approximately 10 years after production began, a
literature review. However, the reader is referred to Chopra3
significant loss in well productivity occurred in some of the
for some reference to prior work. In this paper we present the'
wells. The study shows that this productivity loss was due to
application of compositional modelling to pressure transient
near wellbore condensate accumulation, and documents its
response of wells affected by condensate dropout, and to
effects on production and pressure transient response.
predict future well performance.
A radial, single well, compositional model was used to study
The Arun field is one of the world's giant retrograde gas
this effect and confirm that the productivity loss was due to
reservoirs. Well test analyses indicated possible liquid
liquid accumulation. The model was also used to predict the
accumulation effects. This was confirmed with well
future performance of the well. The model matches well
production data and the pressure transient response of productivity plots. A conceptual, single layer compositional
model was used to verify that liquid accumulation would
affected wells.
cause the same type of behavior observed in the field.
This work identifies near wellbore condensate accumulation ~ubsequthtly,a multi layer compositional model was used to
model a specific well.
as an extremely important factor to consider when predicting
future well performance as some of the productiviies are
reduced by 50%. The work also details how production data
and well test analysis can be used to quantify the effects of BACKGROUND
near wellbore condensate accumulation on well productivity.
The Arun field is located on the northern coast of Aceh
Province in North Sumatra, lndonesia (Figure 1). Mobil
INTRODUCTION operates the field, which began production in 1977. The
average reservoir pressure and temperature were 7,100 psia
The engineering aspects of gas condensate well performance and 352°F at a datum elevation of 10,050 ft-ss. The reservoir
have been a subject of research and development for many is a thick limestone formation with a thickness of over 1,000 ft
years. Recognizingthat classical analytical methods (such as in local areas and covers a productive area of over 23,000
Al-Hussainy, et. al.' and Govie?) for dry gas wells do not acres. The initial condensate to gas ratio (CGR) was 65
BblIMMscf at separator conditions of 1,250 psia and 68°F.
The field currently produces 3.4 Bscflday of separator gas
References at end of paper from a total of 78 producers with an average reservoir
 PRODUCTIONPERFORMANCEOF A RETROGRADE GAS RESERVOIR SPE 28749
A CASE STUDY OF THE ARUN FIELD

Figure 2 - Constant composition expansion

hn

Figure I- Location map

pressure of 2,250 psia.

After initial separation, gas is sent via pipeline to PT Arun, an

LNG plant. Unstabilized condensate is also sent to the LNG
plantfor further separation. A side stream of separator gas is
sent to a field NGL plant where extraction of LPG components
is removed and sent to the LNG plant The residue gas
supplies field fuel, domestic sales, and injection.
0 2 4 6 8 10 12 14 16

Gas injection was implemented as soon as field production Time, hwrs

began to accelerate liquid recovery. Currently 25% of the
produced gas is injected. The lean gas is injected on the Fiiure 3 - TypicalArun pressure iransient response
periphery of the resenroir to sweep condensate rich gas
towards the producers. build up period. A typical log-log derivative curve of the build
up period is shown in Figure 4. The cunre exhibits two
This makes the Arun resenroir a compositionally dynamic different stabilization regions which represent zones of
system where retrograde condensation, water vaporization,
and lean gas injection affect resenroir behavior.

A fluid sample was taken prior to production. Experimental

data revealed the dew point pressure to be 4,400 psi.
Retrograde behavior was determined as shown in Figure 2.
Gas began condensing at the dew point and increased with
lower pressure to a maximum liquid dropout filling about 1.I%
of the pore volume. Further reduction in pressure caused
vaporization of a small portion of the liquid.

As deliirabilii became more critical to meet LNG contracts,

dehrability estimates became more important. To improve
these estimates an intensive pressure transient well testing
program began in 1989, at which time the reservoir pressure
had fallen below the initial dew paint. By 1993 all wells were
tested at least once.

A typical Arun well test response is shown in Fiiure 3. The

test consisted of three onshour flow periods followed by a Fiiure 4 - Typical/og-/ogderivafiw j M
SPE 28749 D. AFIDICK, N. KACZOROWSKI, S. BETTE 3

diierent effective gas permeabilii-thickness products (bh). 1.I% affects the gas relative permeability very little. Even
In this case, &h in the inner zone is lower than that of the when the maximum liquid drop out is reached the gas relative
outer zone. permeability is 0.99. The flow of fluids in the reservoir is
affected very little with condensate dropout. However, the
The most common explanations for this type of behavior are liquid accumulation around the well severely restricts the flow
spherical flow or multi layer effects. Wth the reservoir of gas in the near well region.
pressure below the dew point, another possibility was the
effects of liquid accumulation around the well bore.

Liquid accumulation occurs because producing a well creates ,

a relatively large pressure drop in the vicinity of the well. Gas 8
-

migrating to the well originates away from the well where the #

pressure is higher. This gas is in vapor liquid equilibrium at Gas 8

\
the higher pressure. As the gas migrates to the well, pressure
decreases and a small fraction of the gas condenses close to
the well. This condensate is below the critical liquid saturation
(Sd and does not flow. As more gas is produced, the small
amount of gas which condenses begins to accumulate until
the critical liquid saturation is reached. Condensate then 0 0.2 0.4 0.6 0.8 1
flows into the well as a liquid phase. Total Liquid Saturation

The bank of condensate which accumulates around the well

bore can be envisioned conceptually as shown in Fiiure 5. Fiiure 6 - Experimental relative permeability data
Initially a small bank forms and is entirely below the critical
saturation. Later, the area immediately around the well In order to prove liquid accumulation was affecting the well
reachesthe clitical saturation followed by a transition zone of tests, Productivity Index (PI) plots were generated. PI is
decreasing liquid saturation. Eventually, when the reservoir defined as the total well stream (TWS) rate d i e d by the
pressure reaches the point of maximum liquid dropout, the drawdown pressure. Psuedo-pressures (m(P)) are used in
transition zone terminates at the maximum liquid saturation calculating PI. Flowing bottom hole pressures P )(, were
(S, ,,,Ain the reservoir. estimated from measured flowing well head pressures with
compositionaltubing hydraulics. Interpolation of static bottom
hole pressures was used to estimate the reservoir pressure
(PJ. The PI plot for an Arun well is shown in Figure 7.

..
-
'4

i
P
'ti
*k
k b
.
\\
\ Maximum Wd
Maximum Liquid Diupout
Dropout

Distance from Well

-
Fiiure 5 Typical wrxknsate accumulation as a function of
5300 5100 4900 4700 4500 000
Reservoir Pressure, psia
4100 3900 3700

time

In May 1990 some experimental work was performed to -

Figure 7 P1 of a typical Arun well as a fundion of reservoir
estimate the critical liquid saturation and the gas relative pressure
permeability (kJ. Figure 6 shows the results obtained for a
core sample from the Arun field. The critical liquid saturation
was 51% while the gas relative permeability at the critical A significant drop in well productivity occurred when the
liquid saturation was 0.18. Notice that for the Arun fluid flowing bottom hole pressure went below the dew point. Thii
system the small amount of liquid dropout in the resenroir of was considered strong evidence that the well tests were
 PRODUCTION PERFORMANCE OF A RETROGRADE GAS RESERVOIR SPE 28749
A CASE STUDY OF THE ARUN FIELD

affected by liquid accumulation. Spherical flow and multi

layer effects are not affected by the dew point, so no change
I
in productivity should occur. Only liquid accumulation can 0.9
account for both the well test effects and the significant loss
of productivity below the dew point.

SINGLE WELL MODEL

To confirm that a well undergoing liquid accumulatian would IL: 0.1
behave in the same manner observed in Arun wells, a single 0
0 0.2 0.4 0.6 0.8 1
well, compositional, 2dimensional (r-z coordinate system)
simulation model was used. The effects of liquid water Total Liquid Saturation
vapotizing into the vapor phase because of the high reservoir
temperare is included in the model (Bette' and
Heinemann4).
-
Figure 9 MoclreI gas relafive permeabile c u m

The model conrr;isted of a single 7654 layer of homogeneous The applicabili of this model for generating well test data
properties with 11 radii1cells of varying widths. The inner cell was confirmed by generating a pressure transient test while
radius was 10 R with subsequent cells getting larger. Figure the reservoir pressure in the model was at 5,250 psi. Care
8 illustrates the cell dimensions along with the reservoir was taken to ensure that the flowing bottom hole pressure
properties. The well was completed over the entire interval to remained above the dew point pressure throughout the test.
eliminate partisl penetration effects. The pressure response was analyzed using a well test
analysis software. Excellent agreement was obtained
betweenthe parameters from the analysis versus those used
in the simulation model. A comparison of the results is shown
Diedd
i- in Fgure 10. This confirmed the applicabilityof the simulator
to model pressure transient behavior. Notice that the
derivathre curve in Figure 10 does not exhibit the hump during
*

I
RadialCells the early time as shown by the field test (Figure 4). The hump
a
Number Radius. R * represents wellbore storage effect, which was not simulated
1 I0 in the model.
2 25 2 Well
3 50 i
4 90
5 150 Homogeneousformatiin
6 250 4 Thiiness765 R
7 400
8 650 4 Fbro~ity15 %
s 1,050 Rsrmeability 9.2 md
10 1,700
11 2,800 d

-
Fiiure 8 Single layer model

Initial work with this model indicated that the experimentally

deinred relative permeability cunres were not representative of
the reseivdr and the fluid phases. Therefore, to account for
the interfadat properties of the accumulated liquid near the
wellbore, the gas relative permeability curve was modified
slightly as shown in Figure 9. The experimentally defined
critical liquid saturation was honored and the gas relative
permeabilii at the critical liquid saturation was increasedto
aocumulation
-
Fgure 10 Test h m single layer modrel, prior fo conclensate
0.435. Thii forced the gas curve to be a straight line.
SPE 28749 D. AFIDICK, N. KACZOROWSKI. S. BElTE 5

Usingthe model, a second well test was generated after a 3- Both stabilized regions of the derivative curve matched with
month shut in period at a reservoir pressure of 3,660 psi, well the transition period between these regions matched fairly
below the dew point pressure. The generated pressure profile well. However, in some of the field tests, the transition zone
was analyzed analytically. The results are shown in Figure did not match very well. Thii is a result of the simplified
11. The derivative curve exhibits the same character as that assumptions used in the radial composite model.
observed from our field tests (Figure 4). Stabilized regions
developed depicting two regions of different b h . Figure 12 illustrates gas relative permeability as a function of
distance from the well from the simulation model and that
assumed in the radial composite model. The analytical
model, which consists of only two regions, does not account
for the transition from the inner zone with S, to the outer zone
with connate water saturation (SJ or S, Thus, the
analyb'cally determined derivative curve reaches the second
stabilized region sooner than the simulation model.

Figure 11 - Test fkom single layer model, atter condensate 0 20 40 60 80 1W

accmula~ D'iance from Well, R

An analytical radial composite model was used to interpret Figure 12 - k, as a function of distance fkom the well
this test. The interpretation results are very close to the
values used in the simulation model. The ratio 0of the
inner k to the outer k is .445 which is very close to the relative Satisfiedthat liquid accumulation can cause the characteristic
gas permeability at the critical liquid saturation used in the behavior seen on Arun well tests, well productivity was
model. generated as a function of reservoir pressure. Fgure 13
illustrates the results of the simulation model. When the well
It is important to note that gas relative permeability at critical pressure passed through the dew point productivity was
liquid (k,at SJ saturation can be determined from the two quickly and severely affected by liquid accumulation. Pi
stabilized regions of the derivative curve for the Awn fluid
system. Thii is the most important factor in determining welt
productivity loss. Thii is so for the Arun system where k,
away from the well is essentially unaffected by liquid dropout.
However, other fluid systems which have higher maximum
liquid dropout can impact k, away from the welt. In these -
g 0 . a :..%***+.
=
, *?*.*-*.*
............. j...........L
.. .
;
..-..
d.2.
:
,
:
: :
j i!
.............
systems, the ratio of inner k and outer k represents the ratio n .. ..
i: i i
i
(M) of k, at S, and k,at S,. , If M is available from core ............. j............;.:..i.i i :
............;............i............i............ 4 ............
: : :
i "
data, the effect of condensate accumulation can be estimated ! I i
....... . .... ....
.....
;
...... /..i~iiiiiiiiiiiii
. ............ ...i..
..... ...i
..... .
i
.....
.....
.........a
from the inflow equation for the radial composite model :
..
............ . ... ............ ............
2 0.16
i
i
ij...**
i ".+-
I

I
ti300 5100 4900 4700 4xXl 4300 4100 3900 3700
Reservoir Pressure, psia

The analytical solution to the radial composite model is

superimposed on the results of the simulation in Figure 11. -
Figure 13 PIof the single layer model
 PRODUCTION PERFORMANCE OF A RETROGRADE GAS RESERVOIR SPE 28749
A CASE STUDY OF THE ARUN FIELD

dropped from 0.341 to 0.186, a reduction of about 451, as immediately surrounding the well reaches vapor liquid
soon as condensate accumulated in just the first cell. The equilibriumwittr the condensate at the higher shut in pressure
first cell was filled up to S, within a short time due to its but does not change significantly with time. Shutting in the
relatively small volume compared to the gas throughput. well does not improve the well p r o d u w (Fussel').
Productivity continues to decline, reaching a 50% reduction,
as liquid accumulates but at a much lower rate. To investigate if liquid will revaporize at lower reservoir
pressure, thii model was depleted to a reservoir pressure of
To further investigate this rapid decline in PI, the single layer 500 psia. Figure 15 shows liquid saturation in the first three
modelwas run with the first cell refinedto five 1-ft cells. The inner d k as a function of reservoir pressure and the effect it
resutt is shown in Fgure 14. PI drastically drops when the first has on PI. The reduction in oil saturation due to
1-ft cell was filled to S ., At the time, condensate had not revaporization occurred in cell #3 long before cells #2 and #1.
started accumulating beyond the 1-ft radius. The declining Pi was not significantly improved until the oil saturation in cell
liquidsaturation prior to the rapid accumulation of condensate # I was reduced. Again, this confirms that condensate
is due to the water vaporizing into the vapor phase. accumulation immediately around the wellbore significantly
affects well productivity.

5050 50M) 4950 4900 4850 4800 4750 4700 4650 4600
Resewoir Pressure, M a
2500 m 1500 1000 500
Reservoir Pressure, psia
-
Figure 14 E W o f candensafe ~ccumuIati0n1 4around the
wellbore -
Figure 15 EffM of liquid revaporizetion on PI

This example illustrates a very small zone with liquid

saturation at S, significantly affects well PI. For a high rate,
high CGR well, this small zone is filled to S, as soon as the MULTI LAYER MODEL
flowing bottom hole pressure falls below the dew point.
To investigate the effect of multiple layers on both liquid
The model with the original radial cells was also used to accumulation and well test response, a six layer model was
invesligate if the accumulatedcondensate revaporizes when constructed. The radial dimensions are identical with the
the well is shut mfor a long period. The model simulated a 2- single layer model. Figure 16 illustrates the thickness of each
year shut in period after the reservoir pressurewas well below layer along with the reservoir properties assumed. These
the dew point with the following results: propertieswere obtained from a detailed geologic description
in the region of an indiiuaf Arun well. In general, porosity
JJauid Saturation and permeability decrease from top to bottom. The same
Year Cell #I Cell Re6 Cell #11 relative permeability curves as in the single layer model was
0 .512 .15 .142 used for all layers.
1 .512 .15 .I42
2 .512 .I5 .A42 The modelshowed that the rate of condensate accumulation
differed from layer to layer. The amount of liquid
The results indicate that liquid saturation in each cell remains accumulation is influenced by the gas throughput.
constant during the 2-year shut in period. There are several Consequently, layers with higher permeability accumulated
explanations for this but the primary reason is that there is condensate and developed the inner zone of reduced k,
very little gas migration at the shut in well . The gas faster than the low permeability layers. Figure 17 shows thii
SPE 28749 D. AFIDICK, N. KACZOROWSKI, S. BETTE 7

Layer h, ft k, md Porosity : .
1 30.5 49.0 .214 ,
2 55 28.0 .22
3 95 11.8 .209
4 52 17.1 .219
5 162 2.6 .I27 :
6 370 1.5 .12 :
10-4 10-3 10-2 10-1 100
drn(P)(psi2/cp) vs dt (hr)
-
Figure 16 Six-layer model
-
Figure 18 Actual vs 6-layer model tests
clearly. The proportion of condensate accumulation in each
layer is almost identicalto the proportion of kh. The higher kh
layers accumulated condensate slightly lower than their kh
proportion as these layers were more severely affected by
condensate accumulation. Consequently, the proportion of
gas throughput in these higher kh layers was curtailed. Thus,
liquid accumulation has a normalizing affect on layered
systems affecting high kh layers more than low kh layers.

5300 5100 4900 4700 4500 4300 4100 3900 3700

Reservoir Pressure, psia

-
Figure 19 PI of the 6-layer model vs. the actual Arun well

the model to that generated from field data is shown if Figure

19. An excellent match was obtained.
Layer

-
Figure 17 Condensate accumulation in different layers FUTURE WORK

With these results, we are confident that the model used in

A well test was generated at a reservoir pressure of 3,660 psi. this study represents the actual Arun well. Some future work
The calculated derivative curve is shown in Figure 18 to be done based on his study include :
superimposed on the derivative curve generated from field
data. An excellent match to the two stabilized regions was - Effect of lean gas injection breakthrough on condensate
obtained. Even though ,a layerswere used in this model, the revaporization
generated transition between stabilized regions and the radius
of the zone with S, did not match the actual data. This is due -Methods of removing the zone of condensate accumulation
to the limitation of using a finite number of cells. immediately around the wellbore to improve productivity by
injecting miscible fluids
A comparison of the well productivity profiles generated with
 PRODUCTION PERFORMANCE OF A RETROGRADE GAS RESERVOIR SPE 28749
A CASE STUDY OF THE ARUN FIELD

- Effect of condensate accumulation on a much leaner gas pF = average reservoir pressure, psia
reservoirs P, = flowing bottom hole pressure, psia
Q = flow rate, Mscflday
re = drainage radius, ft
CONCLUSIONS r~ = radius of inner zone, ft
rw = wellbore radius, ft
Liquid accumulation has occurred in the Arun reservoir. This stc = critical liquid saturation
was identified through well test interpretation and PI plots. ,S, = maximum liquid dropout
This conclusion was verified with compositional simulation. so = Skin factor at Q=O
Other conclusions from this study are : sw = connate water saturation
Z = compressibility factor
- Even with a fairly lean gas, liquid accumulation reduced P = viscosity, cp
individual well productivities by about 50%

- The most dominant factor which determines productivity loss ACKNOWLEDGMENTS

is k, at S,
The authorswould like to thank Pertamina-BPPKA and Mobil
- For the Arun fluid system, k, at S , can be determined from Oil Indonesia for their permission and support to publish this
well test analysis. S , cannot be determined. paper. Also, Mobil Exploration and Producing Technical
Center (MEPTEC) for their review and valuable comments.
-The most critical region affecting productivity is immediately
around the wellbore.
REFERENCES
-The amount of accumulation is controlled predominantly by
gas throughput. Consequently, zones of higher kh contain 1. Al-Hussainy, R., Ramey, H.J., Jr., and Crawford, P.B. : "The
the most liquid accumulation. Flow of Real Gases through Porous Media," JPT(May
1966) 624-636 ; Al-Hussainy, R. and Ramey, J.H., Jr. :
- The accumulated liquid does not re-vaporize if the well is "Application of Real Gas Flow Theory to Well Testing and
shut in for an extended period. Deliverability Forecasting," JPT (May 1966) 637-642

- A radial composite model can be used to analyze well tests. 2. Govier, G.W. : Theory and Practice of the Testing of Gas
k,of the inner and outer regions can be determined but the Wells, Energy Resources Conservation Board, Alberta
transition region cannot be modelled. (1975)
- Condensate revaporization begins in zones away from the 3. Chopra, A.K. :"Transient and Steady-State Aspects of Gas
well. Productivity does not significantly improve until Condensate Well Performance," paper No. 88-39-42
revaporization begins immediately around the wellbore. presented at the 39th Annual Technical Meeting of the
Petroleum Society of CIM, Calgary, June 12-16, 1988

4. Bette', S., Heinemann, R.F. : "Compositional Modeiling of

High Temperature Gagondensate Reservoirs with Water
CGR = Condensate to Gas Ratio Vaporization," paper SPE 18422 presented at SPE
D = non-Darcy coefficient, dayhllscf Symposium on Reservoir Simulation, Houston, February
h = formation thickness 6-8.1989
k = pemeability
+ effective gas permeability
b 5. Fussell, D.D. : "Single-Well Performance Predictions for
4 =k,atS, Gas Condensate Reservoirs," JPT (July 1973) 860-870
= relative gas permeability
h G = Liquified Natural Gas
LPG = tiquified Petroleum Gas
M = kinner z o n J L t e r zone
m(P) = gas pseudo pressure
NGL = Natural Gas Liquid
PI = Productivity Index