Sei sulla pagina 1di 17

SPE 21513

SPE
Decline Curve Analysis for Variable.. Pressure DroplVariable Flowrate Systems
by T.A. Blasingame, T.L. McCray, and W.J. Lee, Texas A&M U.

This is a preprint -- subject to correction.

COpyright 1991, &.e;ety of Petroleum Engineers, Inc,

This paper was prepared for presentation at the SPE Gas Technology Symposium held In Houston, Texas, January 23-24, 1991.

This paper was selected for P!'esentation by a~ SPE Program Comn:'illlee following ~ew of inform~tion contained in an abslracl submitted by the author(s). Contents of the paper, as
p~ted, haw n~t been revtewed by t~ Socie.ly of Petroleum EngIneers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any
POSItion of the ~ty of Petr~~m Englneer:s, liS o~cers, or members. Papers presented at SPE meetings are subject 10 pubfication review by Editorial Committees of the Society of
Petroleum Engmeers. PermISsIOn 10 copy IS res~ed to an abs!ract of no! more than 300 words. Illustrations may nol be copied. The abstract should contain conspicuous
adInowledgernenlof whenl and by whom the paper IS presented. Wnte PubficatiOns Manager, SPE, P.O. Box 833836, Richardson, TX 75083-3836. Telex, 730989 SPEOAL

ABSlRACf

The motivation for the work: described in this paper arose from formed into an equivalent constant rate case for both gas and liquid
a need to analyze production decline data where the flowing flow data. Camach09 independently verified that this equivalent
bottomhole pressure varies significantly. The variance of the constant rate formulation is 'exact for the constant pressure
bottomhole pressure with time excludes the use of the exponential production of a slightly compressible liquid during boundary
decline model for conventional decline curve analysis (semilog dominated flow conditions.
plots and type curves). Using pressure normalized flow rate
rather than flow rate usually does not remedy this problem. The McCraylO sought to develop a method to transform variable-
method we present uses a rigorous superposition function to rate/variable pressure drop data into an equivalent constant
account for the variance of rate and pressure during production. pressure case. In doing this, McCray developed a recursion
This function is the constant rate analog for variable-rate flow formula to compute an equivalent time for constant wellbore
during post-transient conditions and can be used to develop a pressure production, t ep , that could be used with pressure drop
constant pressure analog for the decline curve analysis of field normalized flowrate to perform decline curve analysis using type
data. curves.

The constant pressure analog time function is computed from Although the approach suggested by McCray was verified
the constant rate function using the identity that cumulative using simulation, we sought a rigorous foundation for the
production for both cases must be equal. Using the cumulative application of this result As it turns out, a relatively simple proof
production identity, we solve recursively for the time function can be shown for the application of the t ep function during
using trapezoidal rule integration and, as an alternative, finite boundary dominated flow conditions. This proof is given in
difference formulae. We have also developed a constant pressure AppendixA.
analog time relation which is rigorous for boundary dominated
flow and serves as an accurate approximation for transient flow. In addition to the proof of McCray's result, we also provide
methods to compute the constant pressure equivalent time, tep,
We apply these relations to analytical solutions for verification using recursion formulae in Appendix B. In Appendix C we
and then use the boundary dominated flow relation on simulated provide relations which can be used to compute the constant
and field cases. These simulation cases include large and small pressure dimesionless rate solution given the constant rate
step changes in bottomhole flowing pressures, and periodic shut- dimensionless pressure solution. In the text we will prove that the
ins. Finally, we apply these relations to a gas well field case. computational methods we provide yield essentially exact results
during boundary dominated flow and give very good performance
during transient flow.
INTRODUCTION

The widespread use of type curves l -3 to analyze rate decline VERIFICATION USING DIMENSIONLESS SOLUDONS
data has motivated us to consider the implications of varying rate
and pressure drop production. Theoretically speaking, for the Our first goal is to establish that our new method actually
flow of a slightly compressible liquid, the'analytical stems on the transforms a variable-ratelvariable pressure drop system into an
Fetkovichl type curve are valid only for the constant wellbore equivalent constant pressure system. We begin with a proof of
pressure production case. 4 - 6 In previous works7 ,8 we have the validity of our solution by transforming a constant rate system
shown that variable-rate/variable pressure drop data may be trans- into a constant pressure system. Because these solutions are
frequently used in dimensionless format, we will perform this
References and illustrations at end of paper verification using dimensionless variables.
2 Decline Curve Analysis for Variable Pressure DropNariable Flowrate Systems SPE21513

The computational formulae are given in Appendices B and C functions. We will demonstrate the application of the boundary
and we will verify each. These include the following recursion dominated flow method on a simulated liquid production sequence
formulae; the integral method proposed by McCraylO and the 2- and a field case for a gas well that has been analyzed previously in
and 3-point backward difference methods developed in this work the literature.2.3.8
The recursion relations for this part of the verification are
developed in Appendix B and summarized in Appendix C.
APPLICATION OF THE BOUNDARY DOMINATED FLOW
We will also use the boundary dominated flow relations which MEUlOD TO A SIMULATED LIQUID FLOW CASE
result from equating the constant rate and constant pressure
analytical solutions. This development and the pertinent relations In this section we will apply the lep-Ier transform described in
for this part of the verification are given in Appendix B. the previous section to a simulated prOduction sequence in an oil
well. The reservoir data and flow history are given in Table 1.
Fig. 1 shows the log-log behavior of the qD functions versus
ID and lepD function for the case of a well centered in a bounded TABLE I'
circular reservoir (rd)=l03). Due to the number of methods being
considered. we will discuss the transient and boundary dominated Well and Reservoir Parameters
flow behavior separately. First we note that, during early times (Well Centered in a Bounded CircularReservoir)
(transient flow), all of the lepD methods yield a good
approximation to the qD(ID) solutIon, except at very early times B, RB/STB 1.00
(ID<20). This implies that all of these methods yield a reasonable Ct, psia- 1 15.0 x 10-6
approximation to the analytical solution during transient flow. h, ft 30
Obviously, the analytical solution for boundary dominated flow is ~ 0.30
not valid during transient flow as shown by the deviation of this Jl,cp 0.4
solution and the transient flow solution. rWlft 0.20
re, ft 745 (40 acre)
Now if we consider the late time (boundary dominated flow) k,md 1.0
portion of Fig. 1 (ID>3xlOS), we find that virtually all of the lep!> S 10
methods agree very closely with the qD(ID) solution. Athough this 4800
Pi> psia
scale precludes very close inspection, it does appear that IpcD2 for
the derivative method 1 does show significant deviation. This will CA 31.62
be investigated more closely when these data are replotted on a
semilog qD graph in Fig. 2. Using the well and reservoir parameters given above and Eqs. A-3
to A-5, we compute the m and b parameters and we obtain
Fig. 2 is a replot of Fig. 1 using a semilog scale for the qD
functions and a cartesian scale for the ID functions. We note that m 2.38709xlo-2 psi/STB/D/D
the results for derivative method 1 do begin to diverge from those b = 32.8948 psi/STB/D
of the other methods, which clearly overlay the correct solution.
Fig. 2 suggests that derivative method 1 should not be used in
practice, but that the integral method, derivative method 2, and the Flow Hislory
boundary dominated flow method should give accurate results.
flow sequence I, days q,STB/D Pw/o psia
Of these, the boundary dominated flow method is the easiest 1 180 50 constant rate
to apply since it does not require recursion calculations, but 2 180 constant Pll{ 2000
instead provides a direct transformation between ID (or lerD) and 3 0 shut in
lepD (Eq. B-2). lerD is the is the dimensionless material balance 3 180 constant Pll{ 4000
time function, introduced by Blasingame and Lee7 and later (and 4 180 constant PKf. 3000
independently) by Camach09 as an equivalent constant rate time 3 . 0- shut in
function for variable-rate/variable pressure drop flow conditions. 5 180 constant Pll{ 3500
6 180 constant Pwf 1500
One problem that the boundary dominated flow method might 3 0 shut in
be perceived to have would be that in field applications, ~ormation 7 180 constant Pll{ 2500
properties are required to compute the m and b constants m the lep- 8 540 constant Pll{ 1500
t er transform relation, Eq. B-1. However, these constants are
easily determined fro~ a cartesian pl.ot of L!plq v~rsus !cr<=Qlq)
during boundary d?mtnated flow. This procedw: Is.venfi~ for a The flow rate profiles that were obtained from the simulation
variety of cases 10 Ref. 7. Also, because thiS IS a ngorous cases are shown in Fig. 3. We have included an arbitrary base
formulation for boundary dominated flow, the m and b constants case (Pwt=3ooo psia) to orient the analysis of all of the data. That
are unique and will provide theoretically consistent results when is, when we have obtained the correct transformation of data, all
used in Eq. B-1. cases including the base case should overlay the same trend
Although it is conceivable that these rate profiles could be
Additionally, McCraylO reported inc0!1sistent results using the analyzed separately using decline curve analysis, the consistency
integral method to compute the lep functIon when large pressure of the results would be dependent on the ability of the analyst In
changes andlor shut-ins occured. McC~y was forced to ~se particular, the transient spikes caused by pressure changes and
emrirical extrapolation points to cause the.mtegral method to Y1e~ shut-ins would be difficult to interpret, and the non-uniform
correct results. Again, the boundary dOmtnated flow method will behavior after the effect of· the spike has subsided would
not have these problems because the m and b parameters ~ inevitably yield ambiguous results.
unique and the formulation of the lep-Iertransform (Eq. B-l) IS
rigorous. One method used to align variable rate pressure data with the
correct constant rate solution is rate normalization of the pressure
For these reasons, we recommend that the boundary drop. In decline curve analysis, many analysts use pressure drop
dominated flow method which uses Eqs. B-1 (field) or B-2 normalization of the flow rate profile in an attempt to obtain the
(dimensionless) be used for computation of the lep or tepD
SPE 21513 T.A. Blasingame, T.L. McCray and W.J. Lee 3

"correct" constant pressure solution. It can be seen in Fig. 4 that TABLE 2


pressure drop nonnalization does not yield a constant pressure
analog solution. Well and Reservoir Parameters
(Assumed Geometry: Well Centered in
Clearly, we must use other techniques which are more a Bountied Circular Reservoir)
rigorous than pressure drop normalization for field applications of
decline curve analysis. The method of choice will be the one B. RB/MSCF 0.70942
proposed by Blasingame and Lee7 which converts variable- Ct. psia- l 1.870 x 10-4
rate/variable pressure drop data to the equivalent constant rate h. ft 70
case: From this analysis we will obtain the m and b parameters 0.06
~
requIred by Eq. B-1 for transformation to an equivalent constant p,cp 0.02167
pressure system. rW> ft 0.354
Fig. 5 shows the cartesian plot of lJp/q vs. te,{=Q/q) required
S -5.30 *
k. rod 0.07865 *
to det~ne. the m and b parameters. '!' is the slope of this plot p\4fpsia 710
and b IS the Intercept. Although there IS some data scatter, it is
clear that the m and b parameters do represent a best fit trend of Pi> psia 4175
the data. Therefore, the step of detennining the m and b CA 31.62
parameters is illustrated as a simple and straightforward process. G, Bscf (ref.2) 3.360
Fig. 6 shows the log-log plot of lJp/q versus ter that could be used G, Bscf (ref.3) 3.035
for type curve matching on constant rate type curves. Fig. 6 also
shows that the concept of using L1p/q and t er appears to also be * Average of values obtained from ref. 2 and 3.
valid for transient flow, given the agreement between the constant
rate and constant pressure base case (PwF3000 psia) during From the results of ref. 8 we have
transient flow (ter<5Odays).
ma =
2.05536x10-3 psi/MSCF/D/D
The next step is to use the m and b parameters in Eq. B-1 to ba =
1.3094 psi/MSCFID
convert from ter (constant rate analog time) to tel! (constant G = 2.6281 Bscf
pressure analog time). This is also a simple and straIghtforward
procedure. Once tep is computed, a log-log plot of q/IJp vs. tep is Fig. 8, which is a log-log plot of .1fJa/q versus t er a. is taken
made. Fig. 7 is such a plot and we immediately note that all cases directly from ref. 8 and shown here for completeness: We have
overlay the same trend during both transient and boundary included the computed response during boundary dominated flow
dominated flow. Obviously, the analytical solution for boundary as pres<:ribed by Eq. A-I. The tJiJa/q and ter,a variables are the
dominated flow (exponential decline) will not agree with transient pseud~tIme and pseudopressure as defined and computed in ref.
flow solution. 8. This nomenclature may seem awkward, but defining these
variab~es in this manner allows us to use liquid flow equations for
Fig. 7 represents the endpoint of our effort to determine an lI:J1alysls. Therefore, any equations we present are valid for either
equivalent constant pressure transformation for variable-rate/vari- lIqUId or gas flow as long as the correct time and pressure
able pressure drop flow data. We are satisfied that this is a logical variables are used.
and consistent procedure that should yield accurate results when
applied to field data. The verification of this method is that all We note that the boundary dominated flow solution does not
cases overlay the base case (P~3000 psia), where q/L1p and t model the transient flow behavior of the data in Fig. 8. This is
were used as the plotting functIons for the base case. At this expected and we only state this observation for completeness. We
point, Fig. 7 can be used for decline curve analysis using type could use Fig. 8 in a type curve matching analysis with constant
curves such as the one presented by Fetkovich.l rate type curves. However, it is our objective to analyze these
data with a constant pressure type curve so we must proceed with
transforming the ter,a variable to yield tep,a.
APPLICATION OF mE BOUNDARY nOMINATED FLOW
METIiOD TO A GAS WELL FIELD CASE This analysis requires that we modify the variables in Eqs. A-
I, A-3. A-4. and B-1 for analysis of gas well test data using
In this section we will apply the tep-ter transform (Eq. B-1) to pseudopressure and pseudotime. For gas analysis Eq. A-I
the analysis of variable rate gas well data as described in becomes
references 2,3, and 8. Unfortunately, the complexity of this
analysis is compounded because gas well analysis requires the !!.p (t)
~=matcr,a+ba (1)
used of pseudopressure and pseudotime. We will refer to ref. 8
for the pseudopressure and pseudotime functions as well as for Eq. A-3 becomes
the results of an iterative procedure to detennine the m and b
parameters and the gas-in-place, G. This iterative procedure, as
ma=.J..- (2)
described in ref. 8, simulataneously determines these parameters cp
because the pseudotime function requires knowledge of the gas-
in-place for material balance computations. Eq. A-4 becomes

The reservoir data and flow hil'tory are given in Table 2. ba = 70.6
Bg
kh
J.L 1 4A
'W'"
e/\"Arw
2
)
(3)

For a closed cylindrical reservoir Eq. 3 becomes

ba = 141.2B::[,~)_~] (4)
4 Decline Curve Analysis for Variable Pressure DropNariable Flowrate Systems SPE 21513

And finally, for gas well test analysis Eq. B-1 becomes Also, for a bounded circular reservoir, the formation permeabi-
lity, k, can be computed using
=~~1 +~tcr,a}
141.2~[1!:L)_3:j(;a)
tcp,a (5)

We have used Eq. 5 to compute the tep,a functions used in Fig. k=


9. Note that Fig. 9 is a log-log plot ot qftJpa versus t ep a and that h r.' 4 qdD (10)
w
the the boundary dominated flow solution (computed qlfJpa
func?on) agrees very well with the data during boundary Using Eq. 8, we can estimate the gas-in-place, G. This calcula-
domxnated flow but not during transient flow. This is expected tion gives
and we should not be concerned about this difference. G 1 [(~)(~'l
(1.87Ox10-4) (1.0) (l.O)/J
Once we have created Fig. 9 using the qltJpa versus tep,a data,
we will want to match this data upon the Fetkovich 1 type curve. G =2.6278 Bscf
Fig. 10 represents this type curve match. Note that the data agree Using Eq. 10, we can estimate tthe formation permeability, k.
with the type curve during the transition from transient to This calculation yields
boundary dOmit;tated flow and throughout boundary dominated
flow. The scarcxty of data for t e/y1<100 days (4 points) limits our k = 141 2 (O.709 42XO.021 67) [1n(20) _11 (0.78»)
interpretation of the transient now portion of the data, but an . (70) 4] (1.0)
estimate of re/rw'=20 for the transient stern seems reasonable. k = 0.05432 md
Once the data are overlain and matched to the type curve, we The computed values of gas-in-place and permeability
will determine a match point from the coordinates of both plots. compare very well with those obtained in ref. 8. In fact, the
The match point for Fig. 10 is results are virtually identical. The reason for this is quite simple.
The constant rate analog method (using tJpalq and ter,a) and the
Rate Function Match Point: constant pressure analog method presented in this work are
qD= 1.0 rigorously related. Therefore, if we are consistent, both methods
should yield the same results.
qltJ.p =0.78 MSCFID/psi
Our estimate of gas-in-place, G, is 22 percent less than the
Time Match Point: estimate of Fetkovich, et al. 3 and is 14 percent less than the
teiD= 1 estimate given by Fraim and Wattenbarger.3 However, we
tep,a =630 days express confidence that our estimate is as accurate as the ones
given by the other investigators. And we feel that our approach is
more rigorous, because of the pseudopressure and pseudotime
Type Curve Solution Relationships solution formulation.

Once the match point is determined for a data set on the Our estimate of permeability is within 31 percent of the
Fetkovich 1 type curve, we can compute the volume of fluids in the average of the estimates given in refs. 2 and 3. We feel that our
reservoir and the formation permeability, k. We need permeability estimate is consistent with these values of the other
computational relations for liquid (oil) and for gas where the investigators and certainly reasonable, even given the sparse data
correct pseudopressure and pseudotirne functions are used. trend in the transient flow region. These data are used to identify
the re/rw ' value from the transient stem type curves.
Li([.Uid (oj/) Relatjonshius
The oil-in-place,N, can be computed using SUMMARY AND CONCLUSIONS
This paper introduces a method that can be used to analyze

N ~i[(:'L l:l,j (6)


variable-rate/variable pressure drop production data using a
constant pressure analog time function. The most significant
result is that of the boundary dominated time transformation for
constant rate or constant pressure flow. This transform, given by
The formation permeability, k, can be computed using Eq. B-1, allows an analyst to compute an equivalent time for
constant pressure production, quickly and easily, based on the

k = 70.6
BJl
h
1 4A
'Vro'
e/L.Arw
)(;)
2 qdD (7)
parameters m and b, obtained using Eq. A-I. For gas wells this
procedure is less straightforward and requires an iterative solution
developed in ref. 8.

Gas Relationships We considered four different methods to transform variable-


rate data into the constant pressure solution profile. We
The gas-in-place, G, can be computed using considered three recursion formulae which compute the constant
pressure equivalent time function, tep. by panel summations based

G~i[(:l(t;LJ
on trapezoidal rule integration lO and finite difference expansions.
Each of these relations was applied successfully to convert the
(8) constant rate dimensionless solution into the constant pressure
dimensionles solution. However, McCraylO found that
The formation permeability, k, can be computed using application of the trapezoidal rule integration method gives poor
results when applied to erratic data or data with extencded shut-
ins. This behavior makes the general application of these

k=70. 6BgJ1
h
1 )(;aJ
e
4A
rc'
Ar
2
w
qdD (9)
recursion formulae difficult at best.
SPE 21513 T.A. Blasingame, T.L. McCray and W.J. Lee 5

The fourth method developed was a rigorous identity which cp constant pressure or constant pressure analog
equates the boundary dominated solutions for constant rate and D = dimensionless variable
constant pressure production. The resulting two-parameter mp match point on a type curve
relation (Eq. B-1) may be used for dimensionless solutions or
field data applications. When the m and b parameters are
determined using the methods developed in ref. 7, data scatter ACKNOWLEDGEMENTS
should have little effect on the values of the parameters because a
best fit trend is established. These characteristics make the We gratefully acknowledge the assistance of Elizabeth
boundary dominated flow method the most useful product of this Barboza and Jennifer Johnston for their help in the preparation of
work. this manuscript.

Applications:
REFERENCES
We recommend using the methods presented in this work for
the type curve analysis of variable-rate/variable pressure drop 1. Fetkovich, M.J.: "Decline Curve Analysis Using Type
production data. The method is relatively simple and should be Curves," JPT (June 1980) 1065-77.
applicable to a wide range of well test problems, including the
analysis of gas well test data demonstrated in this work. 2. Fetkovich, MJ., et al: "Decline-Curve Analysis Using Type
Curves--Case Histories," SPEFE (Dec. 1987) 637-56.
Conclusions:
3. Fraim, M.L. and Wattenbarger, R.A.: "Gas Reservoir
1. The recursion formulae discussed in this work should not be Decline-Curve Analysis Using Type Curves With Real Gas
applied in practice due to problems associated with the erratic Pseudopressure and Normalized Time," SPEFE (Dec. 1987)
nature of field data, which could cause poor results. 671-82.

2. The boundary dominated flow method is the method of choice 4. Ehlig-Economides, C.A. and Ramey, H.I, Jr.: "Transient
to transform the constant rate analog time function into a Rate Decline Analysis for Wells Produced at Constant
constant pressure analog time function. This method is Pressure," SPEJ (Feb. 1981) 98-104.
consistent, easy to apply, and should give accurate results for
a wide range of problem types. 5. Ehlig-Economides, C.A. and Ramey, H.J., Jr.: "Pressure
Buildup for Wells Produced at a Constant Pressure," SPEJ
3. The boundary dominated transform method can be used to (Feb. 1981) 105-114.
model constant wellbore pressure production behavior exactly
during boundary dominated flow and should give accurate 6. Blasingame, T.A. and Lee, W.J.: "Properties of
results during transient flow. Homogeneous Reservoirs, Naturally Fractured Reservoirs,
and Hydraulically Fractured Reservoirs from Decline Curve
Analysis," paper SPE 15018 presented at the 1986 SPE
NOMENCLAWRE Permian Basin Oil and Gas Recovery Conference, Midland,
TX, March 13-14, 1986.
Dimensionless Variables
bD dimensionless constant defined by Eq. B-4 7. Blasingame, T.A. and Lee, W.J.: "Variable-Rate Reservoir
CA = dimensionless shape factor Limits Testing," paper SPE 15028 presented at the 1986
mD dimensionless constant defined by Eq. B-3 SPE Permian Basin Oil and Gas Recovery Conference,
PD dimensionless pressure (constant rate case) Midland, TX, March 13-14, 1986.
qD dimensionless rate (constant pressure case)
r Euler's constant (0.577216 ... ) 8. Blasingame, T.A. and Lee, W.J.: "The Variable-Rate
Reservoir Limits Testing of Gas Wells," paper SPE 17708
Field Variables (Formation and Fluid Properties) presented at the 1988 SPE Gas Technology Symposium,
B formation volume factor, RB/STB Dallas, TX, June 13-15, 1988.
b constant as dermed by Eq. A-4 (liquid) and Eq. 3 (gas)
Ct total compressibility, psia-1 9. Camacho-V., R.G.: Well Performance Under Solution Gas
G = gas-in-place, MSCF (or Bscf as in the field example) Drive, Ph.D. Dissertation, University of Tulsa, Tulsa, OK
h total formation thickness, ft (1987).
k formation permeability, md
m constant as defined by Eq. A-3 (liquid) and Eq. 2 (gas) 10. McCray, T L.: Reservoir Analysis Using Production Decline
N oil-in-place, STB Data and Adjusted Time, M.S. Thesis, Texas A&M
Pi = initial reservoir pressure, psia University, College Station, TX (1990).
Pwf = flowing bottomhole pressure, psia
q = flowrate, STB/D or MSCF/D for the field example
Q = cumulative production, STB Appendix A: Derivation of an EQuiYalent TIme. tfJl.. for Constant
rw wellbore radius, ft Wellbore Pressure Analysis of Yariable-RateNariable-Pressure
re reservoir drainage radius, ft Drop Test Data
rtf) re/rw , dimensionless reservoir drainage radius
~ = porosity, fraction We will start with a relation which allows for the constant rate
Jl viscosity, cp analysis of variable-ratelvariable-pressure drop test data during
boundary dominated flow. This relation has been verified for
Subscripts liquid flow by Blasingame and Lee7, and Camach09. This relation
a "adjusted" variable for gas well test analysis. Use of has also been verified for gas flow by Blasingame and Lee8. The
these variables in gas well test analysis yields an constant rate analog relation is given as
equivalent liquid system.
(7 constant rate or constant rate analog
Decline Curve Analysis for Variable Pressure DropNariable Howrate Systems SPE 21513
6

!¥J(t) We need to prove the right-band-side (RHS) ofEq. A-8. This is


q(t) =mtcr + b (A-I) done by integrating Eq. A-7 so it is the same form as the RHS of
Eq. A-8. This gives

1
where
tcp

i
t

Jorr q('t)d't = Q(t) q('t) d't =1 cp ExP(,~)


tcr=---L (time in Days ) ' d't
q(t) q(t) !J.p('t) bOb
(A-2) o .
or
m = 5.6l5--.lL-
tPhcrA tcp
(A-3)
..l.[1 -T1vn1=D1. tCP1~
i
q('t) d't=
B !J.p('t) m ~11 b ~
b = 70.6 jJ. In( 4A )
o (A-ll)
kh e1CAr~2 (A-4)
and Combining Eqns. A-7 and A-II gives

i
tcp
(A-5) q('t) d't =..l. [1 _ b q(t) )
We will also need the general solution for a well producing at a o 6.p('t) m !J.p(t)
constant pressure during boundary dominated flow. This solution (A-I 2)
is given by Ehlig-Economides and RameyU and later by
Blasingame and Lee.6 This solution is Notice that the right-hand-sides of Eqs. A-IO and A-I2 are
identical. This result proves that Eq. A-8 is exact for boundary
q(t») = 1 Ex,,(.:lll t ) dominated flow.
( t.p b t'\ b
cp (A-6)
Appendix B: Determination of the ('12 Function
Our objective is to develop a general time function that allows us
to use Eq. A-6 to model a variable-rate/variable-pressure drop
Bounda1y Dominated Flow Method
process. This general relation is The objective of this section is to develop a method to compute the
t ep function. A relatively simple relation is obtained by equating
q(t) =1 EvnI.:lll t )
Eqs. A-I and A-7 and solving for tep' This gives
!¥J(t) b ~1'\. b cp
(A-7)
1b Exn/::l11t )-
1'\. b cp - mter + b
1
where t ep is the time at which the constant pressure solution is
valid for a general variable-rate/variable-pressure drop response. or
In this sense, t ep is an unknown which must be determined. tcp =;} l~ I + r.g- to-) (B-1)
or in terms of dimensionless variables
McCray 10 proposed the following relation as a defining relation
for t ep tcpD =~rn(I I~~ taD) (B-2)
tcp

i
where
Q(t) = q('t) d't mD =...2- = 21C r w2
!¥J(t) 0 !¥J('t) reD2 A (B-3)
(A-8) and
McCray developed Eq. A-8 via induction from the constant bD=lnr~_l=lmf 4A )
pressure solution where tJ.p(r) would be constant and could be 4 2 \e7CAr~2 (B-4)
factored out. where
Although McCray proved Eq. A-8 empirically using simulation, r' ~
eD e-S
we sought a more rigorous justification. In the following
derivations we provide a rigorous proof ofEq. A-8 for boundary Although Eqs. B-1 and B-2 are strictly valid for boundary
dominated flow. dominated flow, we find that, for short times, these relations
reduce to
First we require the constant rate analog result, Eq. A-I, and
Eq. A-2. Combining Eqs. A-I and A-2 gives tcp == t cr (B-5a)
!¥J(t) = m Q(t) + b and, in dimensionless form
q(t) q(t) (A-9) tcpD == tcrD (B-5b)
Although we do not present a rigorous proof, Eqs. B-5a and B~5b
Solving Eq. A-9 for Q(tY!!.p(t) gives
do suggest that the t ep function cali be approximated by the ter
Q(t) = .1.[1 _b q(t) ] function during trailSient'flow: 'General application ofEqs. B~5a
and B-5b requires further investigation.
!¥J(t) m !J.p(t) (A-lO)
Eq. A-lO provides us with the left-hand-side (LHS) of Eq. A-8.
SPE 21513 T.A. Blasingame, T.L. McCray and W.J. Lee 7

General Methods
~nd combining Eqns. B-8, B-9, and B-ll, and solving for Atep,i
Integral Method gives
McCraylO proposed to compute the tep function by approximation M i = !¥J(ti) [Q(ti-V _4Q(ti-O + 3Q(ti)]
of the integral in Eq. A-8 using the trapezoidal rule. This
ep
, 2q(ti) !¥J(ti-V !!.p(ti-O !¥J(tj)J (B-13)
essentially results in a recursion formula where the tep function is
computed as
n Appendix C: Computational Formulae Used to Compute the
tep = L !!.tep i Constant Pressure Dimensionless Rate Solutions froID Constant
i=1 ' (B-6) Rate Dimensionless Pressure Solutions
The t..tep,i terms are computed using individual trapezoid panels of
In the calculations, the constant pressure dimensionless rate
solution is defmed as
the -q(t)
- f uncbon.
. 1:'
.ror • di 'd al
an m 'd h
VI u trapezol we ave qepD=_1
!!.p(t) PD (C-I)
The equivalent dimensionless time is
Ij = !!.rep,; [q{t j) + q{ti-I)] n
tepD = L MepD i
2 !!.p(tj) !!.p(ti-I) i=l ' (C-2)
where the McpD,i are given for the integral method as
also

Ij = [Q(tj ) _Q(ti-I) ] tD i tD
2 -'---'-
i-I]
!!. . _ [PD,i PD,i-1
!!.p(ti) !¥J(ti-I)
for a given panel.
tcpD,1 - [-L I I 1
PD,i PD,i-IJ (C-3)
and for derivative method I as
Combining and solving for !!.tep,i gives A tD,i PD,i
iltepD' ,1 -- tv·
" - PD,i-1
2[~-~] and for derivative method 2 as
(C-4)
!¥J(tj) !!.p(tj-I)
A PD "i (tD ,'-2 4tD ,i-I 3tD ,i]
tepD,i = 2 PD,i-2 - PD,i-1 +pj)j (C-5)
(B-7)
We can also use the boundary dominated flow method to compute
Derivative Methods
the tepD function. In this case, tepD is obtained using Eqs. B-2,
Other methods, which are based on the derivative of Eq. A-7, can
be developed to compute the t ep function using Eq. B-6. Differen- B-3 and B-4.
tiation ofEq. A-7 with respect to t cp yields
q(t) _ -fL (Q(t) )
!!.p(t) - dtcp !!.p(t) (B-8)
Recalling the general finite difference expansion for the first
derivative
(x) =-AL
a.!!.xi (B-9)
where
( (x) = fIrst derivative
M = difference expansion
a = constant for a given difference expansion
t..xi = panel width for the difference expansion.
In our case we will use the fIrst and second backwards difference
expansions. These are
case I: N=fi - kl ,<x=l (B-IO)

case 2 : !!.f=k2 - 4fi-1 + 3/i, a=2 (B-ll)


Combining Eqs. B-8, B-9 and B-1O and solving for Mep,i gives
Me i = !¥J(ti) [Q(tj} - Q(ti-I) J
p, q(ti) !!.p(ti) !¥J(ti-I) (B-12)
SPE 21;1,

1 2 3 4 5 6 7
10 10 10 10 10 10 10 108
10°-=! " " 11 "1 "'" 11 11 "'11'''1 • I II II "! I I II III'! • ""111! 111111111:-10°
(~l/PD) vs. tepD (all tepD methods)
i i

10- 1 _~~:Js.-to for(constaht pressure a~a1Y~:a1 s O l u t \ o n + - 1


boundary dominated flow) ! i
10- 1

CI}
c
o
',::l
u
10-2
topD4 = (bolmo) 1n[1 (mofbo)toJ
mD = 2/reD2
J I I
··+···································t·····
I
·..· ··· ··...1·· · · ·· ·
It · · ·..· · , 10-2
~
. . .
bD = In r D-3/4 ! <ID(=l/PD) vs. tepD4 (computed
e e 1
:
from given relation)
j ; ~

IqcpD vs. tD (~onstant presiure analytical solution)-+


10- 3 . . . . . . . . . . . . . . . . . . .! j · ··..~(=l/PD) vs.; tepD1 (Integr~l method)~·l1 · ·..· .. 10-3
i <ID(:.l/PD) vs. tcpD~ (Der~vat~ve ~ethod 1)~
I
i 1 <ID(-l/PD) vs. tepm (DenvatIve method 2)/1
qcpD ~s. tD (consta~t pressure solution for bo~ndary domin~ted flow)~
10-4 1 i i i i i i i if
.
i i i i i i 'ii i
.
I I i II iii
.
I i i i i II if I
.
i i i 11'1i I i I I II Iii I
. : \ i I I 1111
~-4
10
1 2 3 4 5 6 7 8
10 10 10 10 10 10 10 10
tDand tepD functions

Figure 1 - Log-Log plot of qn versus tn Functions for a Finite Circular Reservoir


3
Produced at Constant Pressure (rcD= 10 ).
SPE 2J 51 ,

6
xl0
o 5 10 15 20
0
10 -=I , , , I=- 10 0
tepD4 = (bdmD) In[1 + (mdbD)tD]
mD = 2/reD2
bD = In reD -3/4
10- 1 10- 1

It~~~=:~,::~,(~J:~;=~:~:; i
I
C/}

§ ClD(=l/PD) vs. tepD2 (Derivative method 1)


'J:!
u
l:: 10-2 ................................................................j............................................ . J 0& • • • • • • • • • • • 1 1••••••••••••••••••••••••••••
10-2

~elation)
t.S
a
'1o(=l/pol vs. topo. i(cOmputed from given I
10- 3
.......··· ··qepD vs. tD (const~nt pressure analytical ~olutionr · · · · · ·j~~ _·..·, · . · . 10-3
ClD(=l/PDj vs. tepD3 (Derivative ~ethod 2)~
'lcpD vs. tD (consiant pressure analytical ~olution for boundary dominated flOW/
10-4
-l i i i ~ 10-4
6
o 5 10 15 20xl0
tDand tepD functions

Figure 2 - Semilog plot of qD versus tD Functions for a Finite Circular Reservoir


3
Produced at Constant Pressure (reD =10 ).
SPE 21513

1 1 2 3 4
/0- 10° 10 10 10 10
10 8~
p~3000 psi. ~ 10
111111 I I 1111111, I I I I I I II I 3
"I , I " , I ,
6 Pwr2000 psi. i "'" " I

4 (flow 2) (flow 4)

~
m.\mm+~
2
2
10 :
Pwf,b"",=jUC: .
·. 10
2

e 4 I
~
Cf.l
a

2 2

t, days

Figure 3 - Plot of rate versus time for liquid simulation cases.


seE 21513

1 1 2 3 4
10- 10° 10 10 10 10
100gd I!! I ! ! I I ! !! I I ! I I!' I I I ! I ! ! !, ! I I I ! ! ! I ! !!! I ! I ! ! ~ 100

6
pwf=2000 psia pwf=3000 psia pwf=1500 psia
4 (flow 4) (flow 6)
(flow 2)

.....
rJ:l
.~ \ ~ . 10- 1

~ pwf=1500 psia
(flow 8)
~
t; . ~,/
Cf.)

Ii q=50 STB/D 1 : pwr=3500 psia .


~
c::r 10-2~.........
6
(flow 1)
I
1 .lI. ~.~~~. ~~ ---t-!----~ I
pwf=4000 P~~J · ·;:J7~~ ·r "·· ·..·.· · · . 10.2

4
i
1
!
:
i pwf=2500 psia l
(flow 3) ! :
i i i (flow 7) :
2 i PWf,base=3000 psia i l ·2
- I i j ! i ~
3 .,
10 10- i i i i i i ii,'
° I I I I Ii "Ii
1
II I I I i IIIi
: i
I I Iii iii
!:
I I I
\
I I I II
~ 10·3
1 2 3 4
10 10 10 10 10
t, days

Figure 4 - Plot of pressure drop nonnalized rate versus time for liquid simulation cases.
SPE 21513

o 500 1000 1500 2000 2500 3000


120 I ! ! ! ! ! I 120

I
100 .....................................1""" ··················t·····································t······································(····································t····················· .--L 100
1 i j PWf,base=3000 psi- :

~ 8°11! 1 1 + 80
:a
Q.. 60 tj :1....... Pwf--(3500
flow 5)psia · f· i · · · · ·~ 60
...:.! i 1
~.
Q.. i : : Pwr=2500' pSla iI - :i
<I 40 .. ~i ~ pwr=1500
~ (flo~ 7)
(flow 8)psia!r····························T·················_······ ij..... \- 40

~A
' i i PwF3000 psi·l Bound"; Dominared Solution FIOL
. 20 ....... .~............................
:. .. t.
.
: (flow
: 4) .,.............
i Ap/q-- m t cr + b
Ll I- 20
..
2
i
q=50 STB/D pwr=2000 psia ! pwf=1500 psia i m = 2.38709xlO- psi/STB/D/D
(flow 1) ! (flow 2) i (flow 6) iI : :
b = 32,.8948 psi/STB/D
i i
o 500 1000 1500 2000 2500 3000
t cr (=Q/q), days

Figure 5 - Cartesian plot of ~p/q versus to' for the liquid simulation cases.
SPE 2151 ,

10- 1 100 10 1 102 103 104


103~ I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I hI 0 3

6 Boundary Dominated Flow Solution


4 t.p/q=m tcr + b
2

!
m = 2,38709xlO- psi/STB/D/D: .
b = 32.8948 psi/STB/D j Pwf.base=3000 pSla- _~

102~ ···Im jm jm Pwl(ri~~o3~siaJ .-..,


8.
~
~
..... :~ I I 1
CI.l
~

.i
" l' PWf,base=3000 ps~a

~ 1o18~ q=50 STB/D


(flow 1) i( . i

6 i
4 pwr=2000 psia'
(flow 2)
2 pwr=1500 psia pwr=3000 pshi pwr=1506 psia -2
(flow 6) (flow 4) (flow 8)
10o~ I i i i i I Ii iii j I I ill".~
I I I i I I iii I I i i I i iii I j i I I I Ii 10 °
10- 1 100 10 1 102 103 104
t er (=Q/q), days

Figure 6 - Plot of i\p/q versus to" for the liquid simulation cases.
SPE 21513

1
10- 1 10° 10 10
2
10
3
10
4

0 . .
10 10°
8
6
pwr=1500 psia i pwf=3000 psia i
pwf=1500 psia pwr=3500 psia 8
(flow 6) 1 (flow 4) ! (flow 8) (flow 5)

Pwrc~:O~~ ~ ~
4
pwf=2500 psia
(flow 7)

..... 10- 1 ........ q=50STB/D~"~~""""~"""""""\"""""""~"""'''''''''''''''''''''''''''''''''' ....·k10- 1


tI}
~ (flow 1) I
~
~
4 'Pwf,base=3000 psia I
\
~

B~undary
CI)
.,
~
"" DOminated Flow Solution
~
O"i 10.2
8
........................................ qjL\p=(1/b)Exp[-(m/b) tep] =..···l........····....·····..·....··........·....·+~, . ·.. . ·.·.. . ·.. . . . b.1 0.2
m =2.38709xlO- psi/STB/D/D
2
6
4
i b = 32.8948 pSi/SfB/D
J :

Itep =(b/m) 1~[1 + (m/b)terJ I I - --- \ 1-2

10-3~ ~10·3
i :
i i i i iii i i I i i i iii i i I I I I I "'i I i i i iii i i I \
i i i iii i

10. 1 1 2 3 4
10° 10 10 10 10
tep' days

Figure 7 - Plot of q/~p versus tcp for the liquid simulation cases.
tcp computed using the given equation.
sPE 2151'

0 1 2 3 4
10 10 10 10 10
102:d I I I I I I I I , I I I I I I I I , I I I I I I I I , I I I I I I I I b: 102
8 . . . ~

6
4

2 ·2

1 . . . ..............""8 101
10
~
:~=:l:~=~t::;~=C:;i~:-r
8
U 6

~ 4 + tlpJq=m a t er a ba i
rn
0-

~
~
2

...........................................ej
! 'J .
~·!······ ··············..·······
.
·I · · . · · · ·h 10 0

: Results of Gas Material Balance Iterative


I Solution ( Blasingame and Lees}
i rna =2.05536x10-3 psi/MSCF/D/D
i ba =1.3094 psi/MSCF/D 2
i G =2.6281 Bscf
-1-1
10 I I i i i i i Ii
iI I I
iI ii
I i Illii i 1IIIi iii I I ill
~ 10 -1
0 1 2 3 4
10 10 10 10 10
tcr,a' days

Figure 8 - Plot of L\pjq versus tcr,a for data of Fetkovich, et at?


Normalized gas functions computed using methods of
8
Blasingame and Lee .
SPE 21513

1 2 3 4
10° 10 10 10 10
10 1gd "'"11'! "IIII"! "1111111 '111111'1::.10 1
6
4

.~
(/.)
10°
~

~
U
~ 2
~
~
c:r
10-1 8~ q/.1.pa=[rn a tcr,a + bar!
j ·················..·· ······.. i·..··..·······..· ····· [i _ . 10- 1
6 Results of Oas ~aterial Balance Ite~ative !
4
8
Solution ( Blasingame and Lee ).
3
I
rna =2.05536xlO- psi/MSCF/D/D 1
2 ba =1.3094 psi/MSCF/D _---;::--:---:-::..:.:i;-:--~-=-:---=-
0=2.6281 ~scf . Itcp,a= (balrna,) In[1 + (rnJba,)tcr,al I
-2 I : : : I -2
10. i tilliili i ((Iiiilj i iiili'li i iiiiiiirlO
1 2 3 4
10° 10 10 10 10
tep,a' days

Figure 9 - Plot of q/L\Pa versus tcp,a for data of Fetkovich, et al?


tcp,a function computed using tcr,a and the equation shown.
SPE 2151 3

10.4 10-3 10.2 10- 1 10° 10 1 102


102 -d I" I I I I I ! I I I , I I I I ! '" I I I I I , I" I I , I I , " I I I I I I ! J I I , I I I I ~ 102

! 100 i 101 i 102 i 10


3
: 10
4

rJew' = 161 i i i ! 101


10 1
-, ····IXl0· · ·.. ! :· . ·.· · . · ·..··-1-·····..1..· ·· ··· ·..·.. ······.+ 1·· •· • · · ·t I·..·:..· · ·· ·t......... . . 10
1

1 i . 1 Match Pomt i
! 4 i ! (q/~pJmJ. = 0.78 MSCF/D/psi
, i ! (qdD)mp ! = 1.0 !
! : (tcp,Jmp: =630 days i
L : i: ii (tdD)mp i = 1.0 !:
°~ lxlO
4_lxl0~
-lxl0~
L--.......L
ij..·...·............t.......·j"· ·
,:
·rJr w i = 20.0 ! ..
~ 10. !6
........ ....•.
..
~ ~

:ese- I I
~ ~ + .
I 10°

i 4 :
:
:
:
:
!

I 2
\

I
10- 1
"""I!l'::llI1: L
::::l
"",,~L::::: 10- 1

10-
2 . -I Hf
I i i i i i iii
~ I: i II O. I i i iii I
.
11 i I I iii
;',,
II.21
'i
I i
I
I I i
, I
III
I "1
31 i
I
I
\IV'''~IV'~
\ . ''{IV..'
I ill'
.\
41
X'''" ~V'7
Ii I I
rif- 10 2 ...

4 3
10 2
10 110 10 1 10 2
10- 10- tcp,a' days 10- ItO· 10° 10 10
dD

2
Figure 10 - Type curve match of q/tlPa versus tcp,a for data from Fetkovich, et a1.
Fetkovich type curve for radial flow.

Potrebbero piacerti anche