You are on page 1of 14

SPE 39967

Le- 4$’
. . .

Society of Petroleum Engineers

A New Pressure Transient Well Testing Method for Gas Condensate Wells
W. Jatmiko, LEMIGAS Jakarta, T. S. Daltaban, Imperial College London.

SPE Members

Copyright 1998. Seaefy of Petroleum Engineers Inc


uhase including, density, viscosity. formation voIume factor
and the solubil;ty are assumed to be function of the pressure
This paper was prepared for presentation at the f 898 SPE Gas Technology Symposium het~ i!!
Calgary, AIberta, Canada, t>ttl March 1998 variable only. In gas condensate systems considered in this
This paper was selected for presentation by an SPE Program Committee following review of
paper, the compositions and properties of the gas phase and
information contahed m an abstract submitted by the author(s) Contents of the paper, as the liquid phase is assumed to vary with the pressure, and is
presented, have not been rwewed by the Society of Petroleum Engineers and are subject to
mrrectfon by tfm author(s) The material, as presented, does not necessarily reflect any considered that it is not a valid approach to treat them as one
poslion of the Scciety of Petroleum Engineers, ik officers, or members Papers presented at
SPE meetings are subject to publication revkw by Editorial Committees of the Society of of a ‘liquid’ dissolved in a ‘gas’ (the inverse black oil model).
Petroleum Engineers flectronm reproduti”on, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
The composition dependence with pressure may need to be
prohibted Permission to reproduce in print is restricted to an abstract of not more than 300 treated explicitly.
worr& illustrations may not be coPmd The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented Write Librarian, SPE, P O In this paper the flowing pressure response in a gas-
Sox 833838. Richardson. TX 75083-3838. U S A, fax 01-972-952-9435
condensate reservoir will be discussed, and the liquid drop out
effects on its performance will be examined. For the sake of
Abstract simplicity, the discussion is confhed to a ‘homogeneous’
One of the most important parameters for analysing reservoir where a uniform absolute permeability is assumed.
wellbore pressure response in gas condensate reservoirs is the The primary objective is to apply the iterative line source
pseudopressure function, This function correlates the reservoir solution method7 to analyse the flowing pressure response of a
pressure and saturation distributions, and hence the non-linear well in a gas-condensate reservoir. Initially any damage in the
gwerning equation can be transformed into a linear equation vicinity of the wellbore is ignored.
and solved using a standard method. When the reservoir pressure drops below its saturation
This paper describes a method of analysing wellbore pressure, there are three important factors dominating the
pressure response of a gas condensate reservoir under transient reservoir performance, namely the pressure distribution, the
conditions. In this study, the pressure distributions are saturation distribution and the total compressibility. If these
calculated using iterative line source approximation and factors can be quantified properly, the flowing pressure
superposition principles. Initially, the steady-state saturation is response can be used to deduce the actual reservoir properties.
assumed. The reservoir saturation distributions are estimated The purpose of this work is to present an analytical
ffom the sandface saturation and the two-phase flow radius expression of reservoir pressure and saturation distributions,
using a qudratic equation. The pseudopressure function is then and addressing the reservoir characteristics during transient
calculated based on these distributions. The pressure conditions. The well is assumed to produce under a constant
distribution is then recalculated based on this saturation molar rate. An appropriate dimensionless pseudopressure
distribution. The material balance method is used to verifi the function will be used to compare the reservoir performance
saturation distributions. The dimensionless wellbore with the liquid solution.
pseudopreisure is calculated from the wellbore pressure data
and compared with the dimensionless pressure (Iiquid Mathematical Formulations
solution). The ditlhsivity equation for the ith component can be
In general the proposed method shows a good agreement stated as follow* :
with the liquid solution during transient conditions and hence
it can be used to estimate the transport properties of gas
ho~o hgPg Yi
condensate resewoirs. ‘Xi +—
Vp .!!25 oP~xi+sgPgYi) (1)
f w. Pg )]
ktl (

Introduction
forl<i<~
For the black oil system commonly described in the
literature, the physical properties of oil phase and the gas

97
2 W. JATMIKO AND T. S. DALTABAN SPE 39967

This equation is subject to the following inner boundary New Approach


condition. Equation(4) is similar to the steady-state solution of a
‘homogeneous’ reservoir above the saturation pressure (liquid
solution). In order to obtain solutions with time dependent
,*=2+[*+*)9J (2) parameter, this research proposes a transient pressure equation
using the line source solution which can be stated as follows.
rw

where qt is the total molar rate instead of total volumetric rate


as used by the black oil model.
Initial conditions are : where :
1.1
mD(r, t)=~el —
4tD () (5)

PC kropo krgpg
p(r,t) = pl. for any time at the outer reservoir radius, re. mD(r, t)=& —+— dp (6)
J(
p(r,t) = pi. for any radius during an initial time, (t=O). p(r,t) P. P% )

and
Equation(1) shows that although gas and condensate
phases exist simultaneously in the reservoir, since the fluid kt
composition keeps changing, it is impossible to separate the tD=~ (7)
equation explicitly as in the black oil system. Since the fluid +Ct r
properties depend on composition, the Boltzman
transformation cannot validly represent the reservoir Fetkovich [1973] and Raghavan [1976] used the same
parameters as one single variable, The equation cannot be technique to derive a transient solution for dissolved gas
solved analytically using this transformation method because reservoirs from the steady-state equation.
the resewoir saturation and pressure cannot be expressed into Although Equation (5) is derived by an arudogy method,
a single value of (r2/t). there is a theoretical background which can be used to justi~
To overcome this difilculty, first the equation is solved this assumtion. Muskat and Meres4 derived an equation to
using steady-state assumptions. For steady-state assumptions, predict the condensate accumulation in a homogeneous
the right hand side of Equation(1) is zero, and hence the left reservoir using a steady-state assumption. This equation can be
hand side can be integrated directly. Secondly, after solving used to estimate the condensate accumulation in the wellbore
Equation(1) under this assumption, the transient solution can for a given PVT data. In order to improve the prediction, they
be obtained by analogy. The detail of this scheme is presented modified the equation by including the flowing pressure
in the following section. variable which can be obtained from welltest data. Equation
(5) will be used to predict the behaviour of gas condensate
Steady-State Solution reservoirs under transient conditions. The total
Although the steady-state solution is normally only valid compressibility, ~ is evaluated for the average value using a
for a constant pressure outer boundary condition (such as with saturation weighted scheme.
water injection), it can setve as an approximation solution in a
depletion drive reservoir, For a radial reservoir. the ditlisivity Pressure-Saturation Relationship
equation in equation(1) for component i is written as : To evaluate the integral in Equation (6), a pressure-
saturation relationship must be available. This research
1d r ‘roPo ~ +krgPg proposes a steady-state pressure-saturation relationship to
———
‘Yi ~ = ~~ ‘oPoxi ‘sgPgYi) (3) predict the reservoir pressure distribution. This solution is only
r% [[ P. ‘ Pg )] ( used as an approximation method when the in-situ saturation
distribution is not available. The ratio of the mobility fraction
During steady-state conditions, the RHS of Equation(3) is zero. exist in liquid state plus that of in the gaseous state of
The solution of diffilsivity equation for a multi-component component i with the total mobility can be expressed as :
system under steady-state conditions is given in the following
formulas. kro P. krgpg
‘-- Xi +— —yi
P. Pg
mr)(r, t)=h(rr)) (4) = Z? ==-<8)
‘rOF’0 ‘rg Pg
—+—
w. Pg

98
SPE 39967 A New Transient Well Testing Method for Gas Condensate Wells 3

‘r. P.
steady-state pressure and saturation relationship can be
established. The steady-state pressure-saturation relationship
Vo
(9) may be obtained by dividing Equation (9) and Equation (10).
‘(r) = kropo krgpg
—- +-——
Po Pg k ,0 _ PgUr)Po
—_
k rg poV(r)Pg
‘rg Pg

Pg This equation provides a saturation-pressure relationship


V(r) (lo) for steady-state conditions for a multi-component system. It is
‘ro Po ‘rg Pg
–--— + – – analogue of the steady-state gas-oil ratio equation. Except for a
#o Pg binary system, the matrix representation of the linear equation
system is not square (the number of unknown variables is Iess
Equation (S) can be written as than the number of linear equations), therefore the solution is
not always unique. The matrix representation does not create
L(r) xl + V(r) YI= ~1 (11) an orthogonal system, hence the matrix composition needs to
be rearranged. To tackle this problems, this study uses the
It is assumed in this research that in the two-phase flowing single value decomposition method to rearrange the matrix
region, there is equilibrium. Then, from the vapour liquid coetlicient distribution so that it equals to the square matrix
equiIl%rium theory6 , the following constraints must be representation, and hence becomes solvable.
f&lled.
Reservoir Pressure Distribution
1. The summation of liquid and gas fraction of component i in Itis evident that the pseudopressure fimction determined
any time at any radius must be unity. If Equation (9) is from the reservoir integral method should represent the actual
combined with Equation (1O), the summation will be equal reservoir pressure and saturation distributions. To calculate
to unity. this transformation integral, first the reservoir pressure
distribution must be estimated. This research proposes a
L(r) + V(r) = 1. (12) procedure to estimate a reservoir pressure distribution in
advance, so that if the saturation distribution is ready, the
2. The total mole fraction in the liquid and gas phases must be pseudopressure function can be obtained directly. For the first
equaI I. estimation of the saturation distribution, the steady-state
solutions for a multi-component system are utilised. Actually,
this procedure is similar to that in the black oil system
~xi =1 (13) presented previously, except that in the gas condensate system,
j=l
instead of using the volumetric rate, it uses the total molar
rate. Expanding Equation (5)
$Yi =1 (14)
i=( Pe
5&’+*]@= _!!&ei(*l (17)
J(
Obviously, this equation states that the sum of mole fractions p(r,t) Po

of the individual components for a given phase equals to


one< After estimating the reservoir pseudopressure function from
the steady-state approximation, the transient pressure
3. The liquid and gas fugacity of a component i must be qual. distribution will be predicted using Equation (17) with t D is
the dimensionless time evaluated using the average total
fiv = flL (15)
compressibility, ~. For a given radius, this equation is solved
Equation (15) ensures the thermodynamic equilibrium by using an optimisation procedure. The fluid properties are
specifying that the fugacity of component i in the liquid phase, calculated from an equation of state, and the values are
f~, equals to the fugacity of component i in the vapour phase, tabulated as a function of pressure. The details of PVT
fiv This requirement enabies one to compute the vapour evaluation is presented in the following section. Since the
pressure from an equation of state, pressure distribution is not a saturation sensitive parameter,
The equilibrium theory is required to calculate the liquid and is more dependent on radius and total molar flowrate, it is
and gas fractions of component i, If this is satisfied, then the sufficient to determine the transient pseudopressure function as

99
4 W.JATMIKO AND T. S. DALTABAN SPE 39967

the first approximation, To improve the calculated pressure, where


this procedure is repeated when the transient saturation a31 = log(rm)2 (27)
distribution is obtained later. a32 = Iog(rm) (28)
So3 = Sm (29)
Saturation Distribution
The second parameter which is required to estimate the Using matrix notation,
pseudopressure function is the saturation distribution. A
transient saturation distribution cannot be obtained all a21 a31 s 01
a
analytically.
This study proposes a simple quadratic equation to predict a21 a22 a32 ● b = S(,2 (30)
a reservoir saturation distribution. The procedures for a31 a32 a33 c s 03
[1 1[1
calculating constants in the quadratic equation can be
explained as follows. The wellbore saturation and the radius at
This linear set of equations can then be solved by a matrix
which the reservoir saturation equals (1-SWJ are chosen as the
inversion method. The reservoir saturation distribution is then
reference pressure,
The wellbore saturation can be estimated from the well computed from Equation (18) with the calculated coefficients.
The saturation at a radius beyond the two-phase envelope
testing data. In this study, the wellbore saturation is obtained
radius equals to the initial saturation (1 - Swc).
from the steady-state mode14.The radius at which the reservoir
For a gas-condensate system, the wellbore saturation must
saturation equaIs to (1-Swc) has been predicted from the
previous pressure distribution. be approximated through the use of steady-state assumptions.
This is due to the fact that in this system the flow equation
The general quadratic equation is as follow :
cannot be transformed into separate oil and gas equations as it
were in the black oil system. As a result, there is no an explicit
So= a log(r)2 + b log(r) + c (18)
saturation equation similar to that of the black oil system.
Therefore it is assumed that the wellbore saturation can be
for rw < r < r2P~,
obtained during the test. In this case the wellbore saturation
and
can be estimated from the steady state solution, Equation( 16).
so= 1- Swc. for r greater than r2Ph. (19)
Since this method is only an approximate solution, there must
be a procedure to check the accuracy of this approach. A
where :
material balance method can be used as a reference because it
log(r) = the logarithmic of the reservoir radius
can be monitored by measuring the total output rate during a
so = the corresponding saturation.
welltesting operation.
‘2ph = the maximum two phase flow radius. estimated
from pressure distributions.
Material Balance Error
Material balance error is defined as the difference between
Constants a.b and c are determined using three linear
the accumulated produced fluid and the weighted reservoir
equations. The first two equations are :
saturation changes for a particular period, In a conventional
a11a+a12b+c=So1 (20) material balance method, the reservoir pressure is assumed
a21a+a22b+c=So2 (21)
constant, and equals to the average reservoir pressure. This
assumption generally is only valid when the reservoir is under
where :
pseudo-steady state conditions.
al 1 = 1og(rw)2 (22) For a transient period, however, reservoir pressure and
a12 = log(rw) (23)
saturation distributions are not constant, but vary with
a21 = 1og(r2ph)~ (24) distance. To take into account the radial saturation
a22 = I%(rzph) (25)
distribution, in this study the reservoir model is divided into a
and
number of grid block, nb. Each grid block has constant
Sol = welTbore saturation,
pressure and saturation values. The total molar accumulation,
S.2 = initial saturation. (1 - Slvc)
qt can be calculated as the weighted saturation difference
between time steps. The gas-condensate reservoir material
the third equation is selected arbitrarily from the reservoir.
balance error can be calculated as follow :
Suppose rm is any radius between rw and rzph. and Sm is an
arbitrary saturation which will be estimated from the quadratic
equation. The third equation can be written as :

a31. a+a32b+c=So3 (26)

100
SPE 39967 A New Transient Well Testing Method for Gas Condensate Wells 5

Compressibility Factors
Isothermal compressibility for any material is defined by : Pseudopressure
It% For a multiphase flow system reservoir, it is possible to
~=–—— (32)
relate saturation in terms of pressure using the pressure-
~e
saturation relationship evaluated at the wellbore. It enables one
The negative sign indicates as the pressure increase, the to calculate the pseudopressure during the transient conditions.
This pseudopressure is referred to as sandface integral
compressibility factor decreases. For density,
pseudopressure, since the saturation is calculated using the
#@ (33) flowing bottomhole pressure, ~.
Pa If the reservoir is homogeneous without any reservoir
For a variable composition system with n number component, damage, this pseudopressure is stilcient to describe the
the apparent fluid density is calculated according to the fluid saturation changes in the reservoir. If, however, the
compositions. permeability around the wellbore is different from the actual
reservoir permeability, the saturation changes at this area wiIl
The fluid compressibility can be calculated from the above be different, and hence the pseudopressure fimction does not
equation using the calculated apparent fluid density. As the represent the actual pressure and saturation distributions.
reservoir pressure drops beIow the saturation pressure, the Alternatively, a pseudopressure which includes the
weighted compressibility factors for each phase should be pressure and saturation changes in the reservoir is empIoyed.
added together to represent the total reservoir compressibility, These distribution are taken from the current pressure and
saturation distribution. This pseudopressure is referred to as
~= Co So+ CgSg+c\\, S\~+cr (34) the reservoir integral function because it takes into account
lateral pressure and saturation variations. The reservoir
where Cr is the rock compressibility, ~ is the water integral function is valid for a homogeneous reservoir with or
compressibility, Cg is the gas compressibility and co is the oil without any reservoir damaged. The above procedures to
(condensate) compressibility. In this study. since the water calculate pressure and saturation distributions can be employed
phase is immobile and the water compressibility is small to evaluate the pseudopressure.
compared with other fluid compressibilities, it is considered
constant. As in the black oil system, the total compressibility Semilog Plot Analysis
factor is evaluated using a saturation weighted average In welltest analysis semi log plot procedures provide a
scheme, so that it represents the actual reservoir rapid method to calculate the reservoir permeability. In a
compressibility for each particular time. The averaged conventional single phase drawdown test, this is performed by
compressibility is estimated as follow : plotting the flowing pressure versus the flowing time on
logarithmic coordinates. In a homogeneous reservoir, when the
— Cij HCPVJ weIlbore storage has ceased, this plot should yield a straight
Ct = (35) line with the slope inversely proportional with the
~ HCPV permeability.
j= 1 In multi-phase welltesting, as it was presented in
Equation(4), if the pseudopressure function of a particular test
where is plotted versus the flowing time on semilog coordinates, it
reveals a straight line with the slope is inversely proportional
~ HCPV = $rch~ ‘(SOPO +Sofo) (r: -rf.~ (36) to the absolute permeability.
j=l j=] [{ dt } )]
This study proposes a method to analyse a semilog plot of
a multi-phase drawdown test using pseudopressure fimctions
HCPVj = hydrocarbon pore volume at grid block j calculated from the transient pressure and saturation
Ctj = total compressibility at grid block j distributions outlined previously. This is to overcome the
difficulties encountered when the permeability near the
It is obvious that the average total compressibility increases wellbore different from the actual reservoir permeability
considerably as the reservoir pressure decreases, The use of caused by reservoir damage or stimulation. This makes the
initial compressibility to calculate the dimensionless pressure wellbore pseudopressure function (snndface integral method) is
response in a gas-condensate reservoir yields erroneous results. diHerent from the reservoir pseudopressure function (reservoir
In this study, the Peng & Robinson equation of state is used to integral function). Therefore if the earlier procedure is applied
evaluate the PVT of the flow rate. to analyse a multi-phase drawdown test in a such condition,
the results could be misinterpreted.

101
6 W. JATMIKO AND T. S. DALTABAN SPE 39967

Since it gives a rapid solution of the reservoir permeability dashed lines show the saturation distribution from Equation
calculation, a semilog plot analysis also provides a guidance (18) using the steady-state wellbore saturation.
for a typecurve matching evaluation. Theoretically, both Figure(1) shows that the condensate saturation
methods should give similar results. distribution predicted by the analytical method increases
steadily indicating the condensate accumulation in the
Application resetvoir. This figure shows that the calculated saturation
The new welltest analysis procedure described previously distributions do not agree with the numerical results. There are
will be implemented to analyse a gas condensate reservoir. The two important factors contributed to these discrepancies. First,
reservoir characteristics were presented in the appendix. The the inaccuracy in predicting the wellbore saturation from
main objective of this paper is to examine the flowing pressure welltest data. In this study, the wellbore saturation values are
performance of a well in a gas-condensate reservoir. estimated using the steady-state assumption because the
The fluid properties and composition are given in the governing equation of the multi-component system cannot be
appendix. These properties are simulated using an EOS. To separated as liquid and gas equations. Second, the condensate
select the most appropriate EOS, the EOS saturation pressure saturation obtained from the numerical results of this study
is matched with the observed data. After matching the resulted in non-zero values although the reservoir pressure was
saturation pressure, the fluid properties for the compositional still above the dew point pressure. In this study, we observed
simulation are generated with the selected EOS. The PVT data that the higher the difference between the calculated and the
analysis is performed using a commercial PVT package. A observed saturations, the higher the material balance error will
number of EOS were tested for the saturation pressure be. Therefore, in order to compare the calculated and the
agreement. The heavier component, C7+ is splitted into three observed saturation distributions, these factors should be
groups. minimised.
The corrected Peng Robinson EOS with the third However, as far as the two-phase flow boundary prediction
parameter modification is selected because it yields the closest is concerned, the proposed method agrees with the numerical
saturation pressure agreement with the actual data. After results. Figure(1) shows that the predicted two-phase flow
splitting the heavier component, the calculated saturation regions essentially agree with the numerical results. In general,
pressure is found 10be 4379 psi [297.9 atm]. Having calculated as the times increase, the condensate flow movement is
the saturation pressure with the actual data, a flash calculation obviously presented. The flatdashed line in F@me(2) shows
procedure is performed until the abandonment pressure of 100 that the condensate saturation has reached its saturation value
psi [6.8 atm], The fluid viscosity is calculated using Lohrenz- (the wellbore saturation) and the additional condensate
Bray method8. saturation is represented by the saturation movement.
Compared with the steady-state solutions, the proposed method
Results Simulated drawdown tests are acquired from a can be used to estimate the condensate accumulation more
numerical reservoir simulator using a well known commercial accurately.
reservoir simulator. The reservoir initial pressure is set equal Figure(1) shows that for t ~ = 3000, the simulator gives
to the saturation pressure of 4379 psi [297,9 atm]. The initial
the dimensionless two-phase flow radius eqwds to 190,
water saturation is set equal to the 0.3 and the water phase is
considered immobile, The relative permeability table follows whereas the the proposed method gives 240. for tD = 16000,
the Corey’s relative permeability function using an end point the agreement gets better. The simulator predicts this equal to
permeability equal to 0.7 and the exponent number is two. The 490 and the new method gives 500. For the total rate equals to
main objective of these drawdown tests is to examine the rate 10,000 lbm/day [4535 kgm/day] and t ~ = 25,000 the
and permeability effects on the flowing well pressure response. dimensionless two-phase flow radius predicted by the
The totaI molar rate, qt is set from 2000 Ibm/day [907 simulator is 520, and from the new method it equals to 530. A
kgrn/day] to 50000 Ibtn/day [22675 kgndday], whereas the similar agreement is also shown by Figure(3).
absolute reservoir ermeability ranges from 10 md to 100 md To verify the saturaion distributions, Equation (18) is used
[10-11 m2 to 10-1t *n2]
to check the material balance errors versus the average
dimensionless time, ~. Figure(4) shows that the analytical
Saturation Distribution The new method of calculating a
radial saturation distribution outlined previously is saturation distribution yields much lower material balance
implemented for a gas-condensate reservoir study. the wellbore error compared with the numerical one, especially after tD =
saturation-pressure relationship is obtained from the steady- 105. Therefore, if material balance errors are considered, the
state solution. Figure (1) through (3) demonstrate the analytical solution can produce better saturation distribution
condensate saturation distributions calculated using Equation than the numerical solution,
(18) versus the dimensionless radius, r~ on semilog
coordinates. The solid lines are the simulator’s results. The

102
SPE 39967 A New Transient Well Testing Method for Gas Condensate Wells 7

Pressure Distribution Figure(5) shows reservoir pressure representing the dimensionless pseudopressure obtained from
distributions versus the dimensionless reservoir radius on a the numerical solutions.
semilog graph. The analytical solution is calculated using a This figure shows that the dimensionless wellbore
line source solution, Equation. The reservoir effective pseudopressure response calculated using the analytical
permeability is estimated from an in-situ saturation procedures is exactly matched with the liquid solution. The
distribution. simulator solution on the other hand at the beginning falls
The solid line is the pressure response from analytical below the liquid solution considerably. This is due to the fact
solutions, The markers are the one from the simulator. This that the flowing well pressure response is calculated using
figure shows that the analytical solutions do not agree with the steady-state assumptions and is higher than that predicted by
numerical results, The results show that the reservoir pressure equation( 17).
in gas condensate reservoirs cannot be estimated from single For ~ = 20000, the simulator pressure response drops
analytical solution because they involve highly non-linear below the analytical solution, therefore, the dimensionless
equations which are normally solved using numerical methods prseudopressure response increases sharply exceeds the liquid
Using the same model for the black oil system, we can show reference.
that. the number of components in a gas-condensate system Referring to the material balance errors, Figure(4) shows
strongly affects the reservoir pressure distribution. The that after rr) = 20000, the simulator’s material balance errors
solutions of Equation (1~. may be compared with the increase considerably compare with that from the new method.
numerical results using a simplified condensate reservoir
This figure shows that for t~ = 1000, the simulator’s material
model. In this model the number of the hydrocarbon
components is reduced. Using this technique, it is valid to balance error computed using Equation (31) equals to 0.064
represent the reservoir pressure calculated from the numerical compare with the new method which equals to 2.2 10-6.
methods may agree with that of the analytical solutions as it Whereas for ~= 20000 the simulator’s errors ‘becomes 1.0
was presented in this paper compare with 2.710-5 from the new method. It is reveaIed that
However. as far as the predictions of the two-phase flow since the simulator is unable to maintain the minimum
regions are concerned, the analytical solutions agree with material balance errors, the pressure and saturation
numerical results, The proposed method offers better distributions do not represent the actual transient reservoir
agreement with the numericaI results compared with those of distributions. This study observes that the material balance
the steady-state solutions, In the steady-state, pressures vary checks are important in analysing well performance in a gas
linearly with the logarithmic of reservoir radius, whereas in condensate reservoir.
Equation [17). pressures vary with flowing time and reservoir Figure(9) shows the dimensionless wellbore
radius, pseudopressure performance for qT equals to 10000 lbrn/day
For all times considered, the reservoir pressure [4535 kgrn/day] and k = 100 md [10-10 m2]. This figure
performance from the analytical method characterises a shows that as the total rate increases, the deviation between the
‘homogeneous’ reservoir during transient conditions. This is numerical method and the liquid solution becomes apparent
consistent with the solution for the black oil system both from due to material balance error. On the other hand. since
the analytical method and the simulator results. material balance error is minimised, the new approach yields
For gas-condensate reservoir. during the transient stage, consistent results. Therefore, the new approach can be used to
the simulator pressure responses are different from those estimate the reservoir absolute permeability instead of
predicted using Equation (17). Figure(5) shows that the permeability compressibility ratio, W% in a gas condensate
pressure gradient from the simulator is very small near the reservoir.
wellbore. However. after dimensionless reservoir radius equals
20, they rise sharply, This is perhaps because the number of Conclusions
hydrocarbon components is too high, and hence it afTects the The objective of this paper has been to analyse the flowing
numerical dispersions in the simulator. pressure response of a gas condensate reservoir. Within the
context of the case studies considered here, the following
Dimensionless Pressure Distribution Figures(8) through (10) observations may be made.
present semilog plots of the dimensionless wellbore
pseudopressure, mlt,D versus the average dimensionless time, 1. This study has described a procedure to estimate
~ The m}k,Dis calculated using Equation (6). Figure (8) simultaneously reservoir pressure and saturation distribution
shows the dimensionless pressure performance from a using a line source approximation. This enables one to
drawdown test using qT = 2000 lbm/day [907 kgm/day] and transform the flowing pressure response of a welltesting in a
absolute permeability = 10 md [10-11 m2]. The solid line gas condensate reservoir into a dimensionless pseudopressure
represents the dimensionless pressure from the liquid solution. response, This makes a well test analysis procedure
The square is the one from the analytical solution. The circle is

103
8 W. JATMIKO AND T. S. DALTABAN SPE 39987

comparable with the liquid solution, Therefore the reservoir References


properties can be deduced using a trial-and-error method. 1, Fussel, D,D. : “Single-well Performance Prediction for gas
2. Since there is no analytical procedure available to estimate condensate Resemoirs”, Journal of Petroleum Technology,
the saturation distribution, the material balance procedure is June 1973, pp 860-870
invoked to ver@ the reservoir saturation changes. This study 2. Raghavan.R. : “Well-Test Analysis for Well Producing by
has shown that this procedure is significant and if the material Solution Gas Drive Resemoirs “, Sot. Pet. Eng. J. (Aug
baIance errors are beyond the prescribed tolerance, the 1976) pp 196-208
dimensionless pseudopressure becomes inconsistent with the 3. Fetkovich, M.J. : “The Isochronal Testing of Oil Wells”,
liquid solutions, SPE Paper 4529, Presented at the 48th SPE Annual
3. The two-phase flow radius predicted using this method Technical Conference and Exhibition, Las Vegas, Nevada,
agree with the simulator’s calculations, The result is an 1973
improvement compare with the previous calculation using 4. Muskat. M, Meres, M. W. : “The Flow of Heterogeneous
steady-state assumptions. Fluids Through Porous Media”, Physics, Vol 6(l), 1936, pp
346-363.
Nomenclature 5. Jones.J.K Raghavan, R. : Interpretation of Flowing Well
c = compressibility, (I/psi), [l/atm] Response in Gas-Condensate Wells”, SPE Formation
Qexp(–s) Evaluation (Sept. 1988) pp 578-586
ei = exponential integral timction, ei(s)= j —————— 6. Daltaban, T. S.: Numerical Modelling of Recovery Process
s s
jom Gas Condensate Reservoir, PhD Thesis, Imperial
f = fugacity College, London 1985.
h = reservoir thickness (it), [m] 7. Jatmiko. W., Dahaban, T. S., Archer, J.S. “Multi-Phase FIow
HCPV = hydrocarbon pore volume (ft3), [m31 Well Test Analysis in Multi-Layer Reservoirs”. SPE Paper
k = absolute permeability (red), [m*] 36557, Presented at the 71st SPE Annual Technical
k, = relative permeability Conference and Exhibition, Denver, Colorado.
mb, = total material balance error, (lbm/day), [ kgrn/day] 8. Lohrenz, J., Bray, B.G, : “Calculating Viscosities of
mD = Dimensionless pseudopressure Reservoir Fluids from their Composition”, Journal of
nb = number of grid block Petroleum Technology (October 1964), pp 1171-1176.
m = number of hydrocarbon components
P = pressure. (psi), [atm]
qt = total molar rate (Ibn-dday), [ kgndday] Appendix
r = radius (ft), [m] Reservoir Properties
r. = external radius (feet), [m] ● Reservoir external radius, re :1000 ft [304.9 m]
rW = wellbore radius (feet), [m] . Wellbore radius, r~ :0.328 fl [0.1 m]
s = Saturation ● Absolute Permeability :10-100 md
t = time ● Porosity :0.2

tD —
– saturation-averaged dimensionless time ● Reservoir thckness, h :15 ft [4.5 m]
Y = volume (fi~), [m3] . End point relative permeability :0.7
x, = mole fraction of component i in liquid phase ● Porosity index :2
Y1 = mole fraction of component i in liquid phase
7.1 = mol fraction of component i

subscript
o = oil
g = gas
w = water

Greek symbol

= fluid density, (lbm.f13), [kg/m3]


; = porosity
v = differential operator

104
A New Transient Well Testing Method for Gas Condensate Wells 9
SFIE 39967

Gas composition

Component mole(%l

co* 14.735
Nz 0.307
c1 71.852
Cz 5.53
C3 2.527
iC4 0.612
nC4 0.812
iC5 0.4
nC5 0.292
Clj 0.452
C7+ 2.476

— ——
1.80
(reD= 3000 qT=2000 Ibm /d k=l Omd)
~=3000
?)

1.20 ~=6 ,
. . . . ----------------- ------- t
-..
-.. . ... ------
---- ,
---- . ,
--- , NEW METHOD
, ,
, -. .
‘t ..
....~=3mo ‘$, ~ SIMULATOR
‘.
0.60 $
, ‘\ “$,<
=lfim
‘. , zone of uncertainty
$ i.
\ ‘\ . ,, arising from simulation
\
‘. ‘. ,. , results
. ,,
“., t;= 6 \
. *;
\. pres > Pdew
\ \ ,,~
. \
.
‘. .!
.$
‘. i-. I .,, !, I
0.00
1 10 100 1000
DIMENSIONLESS RADIUS, r D
Figure-1
10 W. JATMIKO AND T. S. DALTABAN SPE 39967

1.80 .—. .-.

NEW METHOD SIMULATOR ~


......

.I;.=w

1.20 ,
,

(~=3DJq,=lCJ3Xl lkrn/d k=lClhd) :


t
,
; zone of uncertainty
0.60 ~ arising from simulation
~=moo : results
--------
----------------------------------
.
\ ‘\
. . .
‘. ~.fxl !
\r
i Pres > Pdew
.. ,
----
---- A’
-.. .
..-
I ----- d 1
0.00 J
1 10 100 1000
DIMENSIONLESS RADIUS, t’ D
Figure-2

5.00 —

NEW METHOD SIMULATOR A


.- --- ----- .
-..
, --- 1

4,00 ,,,,.
-.+
=.
,. ... (~=3XD qi9XJXl [@/d k=lCOmi)
,\ ..
,. ‘.
,, ‘. . ,
, ~,
‘, ., — ‘.
3.00 , to= 30(XI ‘\
; zone of uncertainty
\ ~, ~ arising from simulation
,. “.&ZC#J results
t,, = 65’,, “, \ ,
\ \
\ \ ~ %es > Pdew
. .
2.00 ._
\
tD.3cKx13 “..
-r \ ‘+
?5 tD=30m
.’T–— ___ . \
\ \ ,
t,, =65 ‘.\ ‘. \
1.00 . ‘.
\
\. \
‘. ,;
“.
. ‘.
‘,:
‘.
‘. %N
..
-. -.
1
-. 1
.- ,
0.00 I

1 10 - 100 1000
DIMENSIONLESS RADIUS, r D
Figure-3
11
SPE 39967 A New Transient Well Testing Method for Gas Condensate Wells

“ ● =. . . . . . .
0.0 ~~ ..
.

.
i
UJ .
z
a
-1
<
.
m -15.0 “- .
-1

(rm=~q,=~ Ih /d k=lo~) .

1 I
NEW METHOD
n
SIMULATOR
.
1
. I

-30.0 ‘ I E+4 1 E+6


IE+O 1E+2

DIMENSIONLESS TIME, ~D
Figure-4

4,380

,- 4,375
a
w-
Ct 4,370 .
3 .-
.“ CALCULATED
m
U)
r.u
Of 4,365 *
n .
g * t;= 3000
0 *
> 4,360 *
z
w to=lm

K!
M 4,355
k=lOnd)
/
I I I

4,350 >
10 100 1000
DIMENSIONLESS RADIUS, r D
Figure-5

107
12 W. JATMIKO AND T. S. DALTABAN SPE 39967

4,380 —:
0000

.- 4,375
0
Q
w-
W 4,370 .
3
@ CALCULATED —
RI .
4,365 .
K .
. ~=60 Cl

. ..=

(~=XX13qFlCUll lkrn/d

—— .—
4,350
1 10 100 1000
DIMENSIONLESS RADIUS, r D
Figure-6

—.
4,380
o
0
0

0
4,360 .
e=””

*
CALCULATED
*
4,340
*
~ :65
* .

* ~=3000
4,320 /

*
to.2axKl

*
*
4,300 ****
[ L 1

1 10 100 1000
DIMENSIONLESS RADIUS, r D
Figure-7
13
SPE 39967 A New Transient Well Testing Method for Gas Condensate Wells

...-.
10
CONSTANT RATE, LIQUID
(r@.3000q T.2000hn/d k=l md)

8 (-)

c]
WD
o
m 0
6 NEW METHOD o
.

SIMULATOR
o
4

0
.
c1
~ooo u
t n

f’ ‘“’’’”““

E+O 1E+2 ‘“’’’”
“’” ‘ “’’”~“’<’” ‘“’”
I E+4 1E+6
DIMENSIONLESS TIME, ‘DI~ ~

Figure-8

L— CONSTANT RATE, LIQUID SOLUTION


(r@. X03qFlUXKl [km/d k.lCIlrrd)
(>

o
c)
NEW METHOD ‘9
. c1
SIMULATOR o
c1
i, o

L,
I ! I .__. _..
v.

/E+O 1 E+2 1E+4


DIMENSIONLESS TIME, t.,,r
Figure-9
14 W. JATMIKO AND T. S. DALTABAN SPE 39967

25
CONSTANT RATE, LIQUID SOLUTION
(t&=3333~~ltn/d k=l(l%rd)
20 (1

u) 15 NEW METHOD
(c

co . @
u! ,.)
SIMULATOR ,)
c-, c1
10 ,.)
r,
r,

~
,.,
1 I I
,)

I
~E+O 1E+2 1E+4 1E+6
DIMENSIONLESS TIME, to,,~
Figure-10

110

You might also like