Sunteți pe pagina 1din 10

Well-Test Analysis for Multiphase Flow

Rajagopa. Raghavan, * SPE, U. of Tulsa

;SYF-.. IL\ cA J(
Summary. This paper is a review of well-test analysis under multiphase-flow conditions. The contributions of Perrine and Martin
are noted, and the consequences of using their results are documented. The advantages of incorporating the heterogeneous nature of
the porous medium for solution-gas-drive systems with -the pseudopressure function are discussed. Methods to predict well-inflow per-
formance are discussed with regard to the works of Vogel and Fetkovich. The long-time-performance prediction of wells with decline
curves that incorporate the heterogeneous nature of the porous medium is addressed. The final part of this paper considers pressure-falloff
data. Theoretical developments are reviewed, and problems associated with the identification of fluid banks are listed.

Introduction
Although several hundred publications have examined the analysis Perrine-Martin Theory. As mentioned, virtually all analyses of
of well-test data, no more than 20 papers deal directly with well- pressure responses subject to multiphase flow are based on the em-
test analysis for multiphase flow. Perhaps most striking is that both pirical observations of Perrine. 3 He suggested that we can analyze
monographs 1,2 devote a total of four pages to this subject. In well- data for multiphase flow conditions if the mobility term in the diffu-
test analysis, the study of single-phase flow in porous media is the sivity equation for single-phase flow is replaced by the sum of the
norm. The lack of papers pertaining to multiphase flow may be mobilities of the individual phases and if the single-phase compress-
attributed to Perrine 3 and Martin. 4 ibility term is replaced by an effective or pseudocompressibility
The Perrine-Martin hypothesis permits wide latitude in analyzing that is a function of fluid properties and saturations. Martin 4 (also
data subject to multiphase flow. For all intents and purposes, these see Miller et at. 14) showed that Perrine's suggestions are valid if
works enable us to extend any technique valid for single-phase flow saturation gradients are negligible. This point is readily seen if we
to muItiphase flow. Obviously, there is a limit to which this theory consider flow in an oil/water system. We can expand Eq. 1 to
can be extended; thus, these limitations are explored. The past 10 krm 1 dkrm d
years have seen advances in analysis techniques that use concepts --Y'2p +----Y'pY'Sm+krm-[1I(J.LmBm)](Y'p)2
originally used to examine the flow of real gases through porous J.LmBm J.LmBm dS m dp
media. The application of pseudopressure functions 5 to analyze
pressure data in solution-gas-drive and gas-condensate systems is
discussed. 6,7 = k;m (SmCm : + a;;). ......................... (4)
Because the objective of every well test ultimately is to predict
well deliverability (or injectivity), the seminal works of Vogel 8 and If we ignore all second-degree terms, i.e., terms containing
Fetkovich, 9 which discuss the productivity of wells producing by (Y'p)2 and Y'pY'Sm, and add the resulting equations for each phase,
solution-gas drive, are examined. The results of these works are we obtain
compared to resolve the apparent differences.
Y'2p = (</>crlAt)(ap lat), .............................. (5)
Finally, this paper considers the application of the Perrine-Martin
theory to the analysis of injection well data. In particular, the utility where At=(klJ.L)t=k(krolJ.Lo+krwlJ.Lw) and Ct = So Co +Swcw' A
of methods given in the literature to identify fluid banks and to de- similar equation may be derived if we examine the simultaneous
termine the location of fluidlfluid interfaces 10, II is examined. flow of oil, gas, and water. The similarity between Eqs. 3 and 4
is readily apparent. The critical assumption made thus far is that
Discussion second-degree terms are negligible to obtain the left side ofEq. 5.
Basic Equations. The basic equations governing the multiphase On the basis of the works of Perrine and Martin, we can do the
flow of fluids were discussed first by Muskat and Meres 12 and following.
Wycoff and Botset \3 in 1936. We consider the flow of oil, water, l. We can calculate the total system mobility, (klJ.L)t (assuming
and gas in a uniform porous medium where the influences of gravity , three-phase flow), by
capillary pressure gradients, and rock compressibility are negligible. At == (klJ.L)t =(162.6/Im!h){ qoBo + 1,000[qg
Assuming that the {3-formulation is valid and neglecting capillary
effects, the differential equations that govern the flow of oil, gas, -0.001 (qoRs +qwRsw)]Bg +qwBw}' .................. (6)
and water are given, respectively, by Here, (klJ.L)t=(kolJ.Lo+kglJ.Lg+kwlJ.Lw), where k m (m=o, g, or w)
is the effective permeability of each phase. Although never explicitly
Y"[(~ )Y'p]=~~(</> Sm), ................... (I) stated, Eq. 6 also implicitly forms the basis for obtaining individual
phase mobilities, (klJ.L)m, from the appropriate rates.
J.LmBm k at Bm
2. The skin factor is given by
where subscript m represents oil or water, and
s=1.l511 l.lp IIhr -IOg[(~) _1-2 ]+3.231.......... (7)
C Iml J.L t </>CrTw J
Here I~ IIhr is the change in pressure during the test after one
hour of test time, and ct is defined by
=~~[</>(RsSo + RswSw + Sg )] ................... (2)
kat Bo Bw Bg Ct=So(B g aRs __1_ aBo)+Sg( __I_ aB g )
Bo ap Bo ap Bg ap
Eqs. 1 and 2 may be solved subject to the appropriate constraints
given by the initial and boundary conditions to obtain specific so-
lutions of interest. If we restrict our attention to single-phase flow +Sw( Bg aRsw __1_ aBw)+Cf' .................... (8)
(constant viscosity and slightly compressible liquid), then the partial Bw ap Bw ap
differential equation that governs the pressure distribution is given by
Eq. 8 usually is written as
Y'2p =(</>ctJ.Llk)(aplat). . ............................ (3)
Ct=Soco+SgCg+Swcw+cf . ......................... (9)
'Now at Texas A&M U. Here, we assume that cf is non-negligible. Although C t is known
Copyright 1989 Society of Petroleum Engineers as the "effective system compressibility," note that the right side
SPE Formation Evaluation, December 1989 585
W 24.----.---r-r-r.,-n,-----r-~-.-.-rrT~
C)
Z
J!,..., 22
00

W
~~20
0::
::10
(/) ......
::1310- 1
~ ...... "" '" "" f!3 ~ 18
WID
0::0::
o DRAWDOWN g:,
Q.~ o FIRST BUILDUP, 0';;; 16
O~ 1= 62.1 hrs WO-
W" 6 SECOND BUILDUP, ~OOrO' N -
!:::! ~lci2" :::i .;
rF~
1=869.75 hrs 14
...J- <1: ......
<I:~ 1=5DOhrs'
6 ::Eo.
::Eo. 6 0
0::<1 060 ~ <I 12
Oa.:-
Zo 6 0 6 0
60
Z
3 ",I ,,,I w
W15 10
!:t
0::
10'2 10" "" I 10 10
2 10
3 ~
0::
10

102
FLOW TIME,t,OR SHUT-IN TIME, at, hours 10
FLOW TIME, t, OR SHUT-IN TIME, at,hours
Fig. 1-Example application-drawdown and buildup
responses. Fig. 2-Expanded view of drawdown response from Fig. 1.

of Eq. 8 cannot be derived from the basic definition of compressi- observations are applicable to the system under study. Note that
bility: c=[lIp(oplop)]t, where p is the appropriate density; i.e., the results of both buildup tests are in agreement with each other
the right side of Eq. 8 follows from the expansion and combination and with the drawdown response for I1t4.t. For larger values of
of terms on the right side of Eqs. 1 and 2 and not from ther- I1t, the buildup responses for the first test fall below the drawdown
modynamic considerations alone. One such derivation starting with response. The buildup responses for the second buildup test follow
the zero-dimensional material-balance equation is given in Ref. 15. the drawdown response for virtually all times considered here be-
A physical explanation of the various components in Eq. 8 [partic- cause of the large value of flowing time before shut-in.
ularly (BgIBo)oRslop] is given by Ramey. 16 Note that the devel- This example represents direct field verification, possibly the first
opment of Eq. 8 does not require that second-degree terms be for multiphase flow through porous media, that the pressure changes
negligible; all that is required is that saturation be expressed as a during the drawdown period, Pi -Pwj' are identical to the pressure
function of pressure. changes during the buildup period, Pws -Pwj,s' This suggests that
3 . We can use the Matthews et al. 17 functions to determine the Winestock-Colpitts 21 procedure for rate normalization
average reservoir pressure (see Earlougher et al. 18). (drawdown) and the Uraiet-Raghavan 22 method for normalizing
4. Although not confirmed, the deliverability equation for mul- buildup data can be used if pressure changes are normalized in terms
tiphase flow, following the analogy for single-phase pseudosteady- of qt. (It is possible that these results were obtained because a sig-
state flow, is given byl9 nificant amount of free gas was being produced.)
The upward bend in the drawdown response after 500 hours may
qt=i(p-Pws)/p., ................................ (10)
suggest that the system is being depleted, which was entirely un-
_ kh expected. I qualify this remark because a rapid decline in wellbore
where J = ................. (11) pressure (for multiphase flow) does not necessarily imply that the
141.2[1/2In(4AleT'CA r; )+s] system is being depleted. 24 Saturation changes at the sandface can
Here, we assume kh/p.=(kh/p.)t; CA is the shape factor 20 and qt produce a significant decline in the wellbore response. Fig. 2 is
is the total rate (the terms within the braces in Eq. 6). an expansion of the drawdown response (circles) shown in Fig. 1,
and the deviations are clearly evident. The unbroken line shows
Example Application. Here, we consider four tests-three buildup the trend of the data and does not denote the semilog straight line.
and one flow. The flow test was conducted for about 36 days. The The square data points are the results of the third buildup test, con-
first buildup test was conducted before the flow test, and the second ducted after the well was produced for 56 days (total producing
buildup test was conducted for 72 hours at the end of the 36-day time",,2,216 hours).
flow period. The objective was to determine reservoir PV. The well For the moment, let us assume that the deviation at t "" 500 hours
was then produced for 56 days, and a buildup test was conducted denotes the influence of the boundary. As mentioned, the well was
at the end of this period. put back on production after the 72-hour buildup test. On the basis
Fig. 1 shows the data obtained for the first three test periods. of single-phase flow theory, we know that if this is indeed a
Results are normalized by the appropriate test rates. The circular boundary effect and the well is shut in, then the well pressure should
data points represent the drawdown pressure response normalized equal p at I1t"" 500 hours. I With the analogy between buildup and
by the instantaneous total rate (i.e., by the rate normalization drawdown tests and the observation 1 that the buildup response (in
procedure 21 for single-phase gas flow); no smoothing procedure terms of the pressure change during the test) will follow the
was used. There is no theoretical basis to assume that rate normali- drawdown response until the well pressure reaches the static
zation will be valid for two-phase gas/oil flow. reservoir pressure provided that the well produced to pseudo steady
The square data points in Fig. 1 are the results of the first buildup state before shut-in, the buildup pressure was estimated at I1t = 500
test, which show a plot of I1Pbu1qt (I1t=O) vs. I1t to account for hours (the time to reach "pseudosteady state" during drawdown)
the differences in flow rate, where I1Pbu =Pws -Pwj,s' Ref. 22 as approximately 1,784 psi [12.3 MPa] after 2,216 hours of pro-
justifies the use of qt (I1t=O) to normalize the pressure change duction. (If the system is undergoing depletion and the Perrine-
during the buildup period. The triangular data points represent the Martin hypothesis is valid, then the difference between this value
buildup response during the second buildup test, again normalized and the initial reservoir pressure is useful in material-balance cal-
by the rate at I1t=O. culations. Average reservoir pressure could not be calculated by
The basic premise of all pressure-buildup analyses is that if conventional techniques because the reservoir volume was not
producing times are much greater than shut-in times (t~l1t), then known.)
a plot of (l1p )bu vs. I1t will follow the plot of (l1p)dd vs. t, where The prediction of the expected average reservoir pressure and
(l1p)dd=Pi-Pwj' If I1t</i,t, and the well is located in a closed or the time at which it was evident, subject to the above restrictions,
an infinite-acting system, then (I1P)bu < (l1p)dd for identical values was to ensure that the test not be terminated prematurely and that
of I1t and t (see Ref. 23). (If rates are different, then the pressure the pressure change, evident in the drawdown test for t > 500 hours,
changes must be normalized by the flow rate.) It appears that these was a result of depletion. An extended buildup test was conducted
586 SPE Formation Evaluation, December 1989
35
10 105 lOS
DIMENSIONLESS

~1
IAJ 30 FLOW RATE. qoD

~..J .!fw =428.8


._ 499 4910 60.054
..
8.
t!
~
. ~ 0.65
ZONE 0
I ~g 125 00.027
00.011

iii
~
(I) 0.45
40
~~
(llIf)
20
- LIQUID FLOW
cr 4960 ..J 60,100
4870 CI)CI)
::)
CI)
(5 t.days IAJIAJ _ 141.2~ojqo
0.25
If) 10' 10 10 2 ..Jcr 15 qOD -
kh Pj
IAJ
cr
DlSTANCE.r, f.et
~~
0..
z
eng
ZIAJ 10
SET I
4930 4830 5 =0
~
::)
IAJIf)
:!o..
:I:
CI)
0 5

4900 ?03 108


I 10 102 103 104
HORNER TIME RATIO, (Ht.t)/t.t

Fig. 3-lnfluence of saturation gradients on buildup data. Fig. 4-Drawdown response for solutlon-gas-drlve reservoirs.

after about three months of production. This test was terminated about the same as the slope of the second straight line for the case
after 361 hours; the buildup pressure at the end of the test was 1,779 when t=1O days. The values are Ao=3.66 md/cp [3.66
psi [12.3 MPa], and it was evident that depletion had occurred md/mPa 's], Aw= 12.06 md/cp [12.06 md/mPa's], and At= 15.72
(p i "" 1,900 psi [13. 1 MPa]). The square data points shown in Fig. md/cp [15.72 md/mPas]. These estimates are in excellent
2 are the responses of this extended buildup test. Again, note the agreement with the saturation profile in the reservoir and the re-
excellent alignment between the buildup and drawdown responses. sults are in accordance with the recommendations of Perrine and
Martin.
Influence of Saturation Gradients. The following limitations of The above example permits us to answer the following questions.
the Perrine-Martin theory are important: (1) I have made no mention Consider the situation where oil and free gas are produced at the
of individual phase mobilities; (2) determination of skin factor re- sandface and where, at distances far from the well, flow is above
quires knowledge of the total mobility and "effective or pseudocom- the bubblepoint. For this situation, (1) how does one determine Ao
pressibility"; and (3) the estimate of total mobility, (k/ p)t, need in the reservoir on the basis of the Perrine-Martin observation, and
not be constant with time. Perrine suggests that individual phase (2) does an estimate of the gas-phase mobility have any significance?
mobilities can be determined if appropriate flow rates are used; e.g., I believe the above results suggest that the total rate (oil plus free
individual phase mobilities for two-phase, oil/water flow may be gas) should be used to determine Ao in the reservoir, and the es-
estimated by timate of Ag has no significance.
Aj=(klp)j = 162.6qj Bj l[m[h, ........................ (12) The examples in Ref. 24 indicate that to determine individual
where j =0 or w. This view, although restated in Ref. 2, generally phase mobility, production of a given phase is a necessary but not
is incorrect. As shown earli~r. the Perrine-Martin approach assumes a sufficient condition to determine the mobility of that phase, and
that gradients are small. Chu et at. 24 examined the influence of reliable estimates of the mobility of a given phase can be obtained
saturation gradients on pressure data. They show that if saturation only if one of the following conditions is valid: (1) only one phase
gradients exist, then individual phase mobilities may be calculated is mobile in the reservoir, (2) saturation distribution is fairly
only for a restrictive set of circumstances. Before documenting their uniform, and (3) producing fluid ratios (e.g., the WOR for oil/water
conclusions, let us consider an example taken from Ref. 24. flow) reflect the flow rates of the appropriate phases in the zone
Let us assume that the initial saturation profile of a two-phase of interest. Both saturation profiles and pressure gradients govern
oil/water flow system is a step function as shown in the Fig. 3 inset. the mobility of fluids in the reservoir and the Perrine-Martin ap-
Initially, water is immobile up to r=36 ft [11 m]. The other curves proximation can only determine At. (This discussion also addresses
shown in the inset represent the saturation distribution as a function the philosophy25 of determining total mobility.) The reader,
of time. Fig. 3 presents the buildup responses for two values of however, carmot blindly extrapolate these results to other situations;
producing time-1O and 100 days. e.g., solution-gas drive.
Two well-defined straight lines are evident for the buildup
response at the end of 10 days. If the assumptions regarding well Application of the Pseudopressure Concept-Solution-Gas-Drive
performance in a composite reservoir are valid, then the first straight Reservoirs. Although this idea has been used in petroleum engi-
line should reflect the properties of Zone 1 and the second, the prop- neering for about 40 years,26 the impetus for using this concept
erties of Zone 2. Using the oil rate (no water was produced) and in recent times stems from the work of AI-Hussainy et aT. 5
the slopes of the straight lines, Chu et af. estimated Afor both zones If we restrict our attention to the steady flow of oil and gas, then
(with Eq. 6) as 47.2 md/cp [47.2 md/mPa's] for Line 1 and 15.44 following Evinger and Muskat,27 the oil flow rate is given by
md/cp [15.44 md/mPa' s] for Line 2. But do these values represent
Ao or At? For this system, the values of Ao in Zones 1 and 2 are
47.5 and 3.62 md/cp [47.5 and 3.62 md/mPa 's], respectively, and
the values of At in the two regions are 47.5 and 15.65 md/cp [47.5
qo(r)= kh re
~dp"
141.2 In(re 1r ) p(r) PoBo
................. (13)

and 15.65 md/mPa' s], respectively. (The values of Ao and At in If we now define a pseudopressure function by the relation
Zone 1 are identical because Aw=O.) Thus, it is clear that the first
semilog straight line for this test reflects the value of Ao in Zone [p kro
pp(p)=J --dp', .............................. (14)
1 and the second semilog straight line reflects At of Zone 2. This
is an important result because even though we produce oil, it is
o PoBo
not possible to determine Ao in Zone 2. then we may rewrite Eq. 14 as
At 100 days, the oil (and water) saturation is uniform throughout
the reservoir, and the Perrine-Martin theory should yield excellent
results. Let us consider the buildup response after 100 days of pro-
duction. Unlike the response following 10 days of production, only
one straight line is evident because saturation gradients are Because k ro is a function of saturation, the integral can be evaluated
negligible. Note that the slope of the straight line for this case is only if the relationship between oil saturation and pressure is known.
SPE Formation Evaluation, December 1989 587
--LIQUID FLOW
11 Q
8 ............ RAGHAVAN
-'-BOE~~ SET I
1= 1011 hours SHUT-IN TIME,L1t, DAYS s= 0
co
VI
0.2
o
~
II:: 0.04 0 .09
=>
!c(
~ 0.1
<t
<.?

01 10 10 2
DIMENSIONLESS DISTANCE, ro
Fig .-Computatlon oom.._n. of dimensionless .... '
dopressure rise during buildup (after Aanonsen 33 ). Fig. 6-Gas saturation profiles during buildup.

Evinger and Muskat showed that the producing GOR, R, may be as a function of distance varies by about 10% during the boundary-
used to relate pressure and saturation. dominated flow period. This exceedingly important observation is
Note that the steady flow of hydrocarbon fluids has been examined the precursor to virtually all developments in this area of pressure
only recently. 7 ,28 These works establish the following points re- analysis. From my perspective, its only limitation is that Levine
garding steady flow of hydrocarbon fluids in porous media. and Prats did not consider the influence of the skin region on the
1. The overall composition at any point r is equal to the overall variation of GOR with distance. Note that the Levine-Prats obser-
composition at the external boundary of the reservoir. vation says nothing about the variation in producing GOR with time.
2. The mole fractions of components in the liquid phase, Xi, and All they suggest is that at any instant in time the variation in GOR
the gas phase, Yi, satisfy the flash or constant-composition ex- with distance may be considered negligible. Our ability to analyze
pansion (CCE) equations. data in solution-gas-drive systems by the Perrine-Martin theory and
3. The pressure distribution in the reservoir satisfies the single- the deliverability equations given by Fetkovich 9 follow directly
phase liquid-flow equation in terms of the pseudopressure. from the Levine-Prats observation.
4. Under steady flow, liquid and vapor do not flow at equal ve- A study of transient flow in solution-gas-drive reservoirs was con-
locities. ducted by Raghavan 6 in 1976. He considered constant-rate pro-
5. When two phases are in equilibrium, both phases are mobile; duction in a reservoir with P;=Pb' The well was located at the
thus, if two phases exist in portions of the reservoir, there would center of a cylindrical reservoir and the outer boundary was assumed
be a discontinuity in the saturation profile at the point where P = Pb to be closed.
(solution-gas drive) or P=Pdew (gas-condensate system). Fig. 4 shows, the drawdown response at a well producing a
6. The ratio of relative permeabilities as a function of pressure reservoir by solution-gas drive. The unbroken lines are the liquid
is given by solutions, and the data points are the gas-drive solutions expressed
in terms of dimensionless pseudopressure. By analogy with liquid
krolkrg=(Pglpo)(L/lo/V/lg), ........................ (16)
flow, the dimensionless pseudopressure, PpwD, is given by
where Pm is the molar density of phase m (m=o,g) and L and V
are the mole fractions of liquid and vapor that are in equilibrium
for a CCE (flash) process. Thus, Eq. 16 provides an alternative
method to obtain kg Iko for steady flow and does not require
knowledge of the producing GOR. Note that Eq. 16 implies that In Fig. 4, dimensionless time, tD, is based on initial system
for steady flow, krolkrg or kolkg is a single-valued function of properties:
pressure. Here, steady flow means all variables are independent
of time. 0.0002637kt
tD= ................................ (19)
In 1961, Levine and Prats 29 examined the performance of
~Cti /lit;'
solution-gas-drive reservoirs for the boundary-dominated flow
period. They suggested that the pressure distribution in the reservoir The basic data used to generate these results are given in Fig. 1
during pseudosteady-state flow can be estimated by of Ref. 6. To compute pseudopressures, it was suggested 6 that the
instantaneous producing GOR be used to express sandface satu-
kh [Pp(Pe)-Pp(P)] rations as a function oftime (pressure). Note that, in this case, the
qo=--' ..................... (17)
141.2 (r -r1 re)
2
---+In-
. parameters on the right side of Eq. 18 are functions of time.
Comparing PpwD with PwD values indicates that the difference
2rl r between two solutions is negligible during the transient flow period.
Note that the slopes of the straight line of the solution-gas-drive
Eq. 17 is analogous to the result for the flow of a slightly compres- and liquid solutions are identical; i.e.,
sible liquid during pseudosteady-state flow. Jones and Raghavan 7
dppwD Id In tD "" 0.5 ............................... (20)
have shown that an expression similar to Eq. 17 may be written
for gas-condensate systems in terms of the total rate (liquid and This result forms the basis for computing the pseudopressure with
vapor); i. e., the right side of Eq. 17 would be modified appropri- the instantaneous producing GOR because it suggests that it is pos-
ately. Again, the main problem at this stage is finding an expression sible to compute formation permeability if we plot Pp (p) vs. ton
that allows the integral on the right of Eq. 17 to be evaluated. Levine semilog paper. Similar results are obtained if s ~ 0, provided that
and Prats suggest that the conventional material-balance equation 30 steady flow conditions prevail in the skin zone. 6 Also, the fact that
may be used to obtain the pressure, saturation, and GOR at the ex- there is good agreement between the solution-gas-drive and liquid
ternal boundary. Furthermore, they suggest that at any given in- solutions during the transient-flow period indicates that the skin
stant in time, the GOR may be assumed to be constant throughout factor can be computed by the pseudopressure approach. The es-
the reservoir. By numerical simulation, they show that the GOR timate of skin factor represents the mechanical or actual (damage,

588 SPE Formation Evaluation, December 1989


Pi =3175 psia
Pwf =2000 psia
0.95 , Cl R ji
o
01.180472.1 1946 3174
o 1009275.9 2046 2963
A 2043 187.6 2511 2815

0.80 - - SIMULATIONS
- . - FROM STEADY STATE THEORY
0, 0, A FROM PRODUCING GOR
0.75L-_---I_ _---1_ _--..l._ _--l.._ _---L.._ _...J
2000 2200 2400 2600 2800 3000 3200
PRESSURE,p, psia I 10
DECLINE CURVE DIMENSIONLESS TIME, 'dO
Fig. 7-Computatlon of saturation profiles, GOR, and steady-
state methods. Fig. 8-The Fetkovich type curve for decline-curve analysis.

stimulation) skin factor and not the existence of free gas. Jones and phase of the test as long as boundary effects are negligible. It is
Raghavan 7 obtained a result similar to Eq. 20 for condensate not clear whether the results of B~e et al. are applicable if S;C O.
systems except that pseudopressures are defined in terms of both B~e et al. 's observations that the producing GOR is independent
phases. They show that total mobility At(rw ) can be computed as oftime during the infinite-acting period and that the producing GOR
a function of flowing pressure, Pwj, by the relation during steady state is constant do not imply that the producing GOR
is constant during boundary-dominated flow as suggested by
1.151qt Whitson. 32 Any success Whitson claims for analyzing data during
At(rw)= ....................... (21) boundary-dominated flow should be attributed to the results of
27rCh IdPwj/d log t I
Levine and Prats. 29
Here qt is the total molar rate, and At is given by Although the pseudopressure concept appears to work well during
the transient flow period, application of the pseudopressure concept

+pg~).
to analyze buildup data requires that we resolve the question of how
At=(P}.::'" ............................. (22) one should compute pseudopressures during the buildup period.
/Lo /Lg Simulations suggest that the changes in saturation (at the sandface)
that occur after the well is shut in are not germane to the problem
where C=0.OO6328 ft3fD [0.7466 x 10- 5 m 3/d]. On the basis of
under consideration. Also, pseudopressures computed during the
Eq. 20, the effective mobility to the oil phase (as a function of
flowing period are not expected to be valid during the buildup phase
pressure) for solution-gas-drive systems can be computed along the
. of the test. Based on these observations, Raghavan 6 noted that the
same lines and kg(r w) (as a function of pressure) can be obtained
pressure and saturation distribution at shut-in must be incorporated
from the producing GOR. These computations are valid only after
into the computation of pseudopressures. He showed that the
the onset of steady flow conditions in the skin zone.
producing GOR at shut-in can be used to incorporate these con-
At late times, agreement between the solution-gas-drive solutions ditions. On the basis of Levine and Prats' results, we expect
and the liquid solutions is not good. Unfortunately, the slope of Raghavan's method to work well during the boundary-dominated
the P wD-vs.-tD curve on Cartesian coordinates is different from flow period.
27rrt; r}, the conventional result (see Fig. 4 of Ref. 6). This ob- In essence, Raghavan's6 procedure indicates that what is im-
servation also applies to gas-condensate systems, 7 which suggests
portant is the saturation and pressure distributions at time .:It=O;
that it is not possible to determine reservoir size by reservoir-limit e.g., the saturations at the sandface for .:It>O have no bearing on
tests in solution-gas-drive systems if the heterogeneous character-
our ability to determine formation properties from pressure-buildup
istics of the reservoir dominate the well response. (Chu et al. 24 data. Thus, even though the expression derived by B~e et al. pre-
discuss conditions under which it is possible to compute reservoir dicts the sandface saturations accurately, it is not useful in com-
PV for two-phase flow in oil/water systems.) Although Raghavan 6
puting pseudopressures for the buildup period.
was unsuccessful in correlating the solution-gas-drive responses with
The response shown in Fig. 5 is extracted from Ref. 33. Here,
the liquid-flow solutions during boundary-dominated flow, it is ex-
PpsD represents the dimensionless pseudopressure rise and .:ltsD
pected, in view of the Levine-Prats results, that the deliverability
represents the dimensionless shut-in time. They are defined by
equation 9 is valid:
qo =i[pp (ji)-pp(Pwj)]' .......................... (23) kh
PpsD= [pp(Pws)-pp(Pwj.s)] ................ (24)
Theoretical confirmation of this result is needed. 141.2qo
In 1981, B~e et al. 31 presented a theoretical analysis of solution-
gas-drive reservoirs. This is an important study for two reasons. 0.OOO2637k.:lt
and .:ltsD = ..................... (25)
First, they present an analytical expression to compute the change
ct (Pwj.s )/L(Pwj.s )r;'
in saturation at the sandface with pressure. Thus, they suggest that
the need to use the producing GOR can be eliminated and the errors The unbroken line in Fig. 5 represents the single-phase-liquid
associated with its measurement can be avoided. The expressions solution. The chain-dotted lines represent responses where Pp (Pws)
they derive may be used to relate saturation changes at the sandface is calculated with B~e et al. ' s relation. The dotted lines represent
for both drawdown and buildup'tests, provided that boundary effects the solution obtained where Pp (Pws) is calculated with the
are negligible during the test. The second important conclusion is producing GOR at .:It=O. It is clear that Raghavan's method ap-
that during the infinite-acting period, the change in the producing proximates the single-phase solution better than B~e et al.'s method.
GOR with pressure (time) is zero; i.e., dR/dp=O. Thus, the advantage of using the producing GOR at the instant the
By simulations, B~e et al. showed that the pressure/saturation well is shut in is clearly evident. (Fig. 5 represents the case where
relationships they had derived are valid for both drawdown and the B~e et al. method works best for the five examples considered
buildup and that the producing GOR stabilizes during the drawdown by Aanonsen. 33 )

SPE Fonnation Evaluation, December 1989 589


He proposed that this equation be used to predict deliverability in-
stead of the equation developed on the basis of the single-phase-
flow theory [Eq. 10 with (kh/JL)t=constant].
On the basis of field experiments, Fetkovich 9 suggested that the
inflow-performance relationship (IPR) may be written as

qo=J'(p2_py,j)n, ............................... (27)

where n is a constant in the range 0.5:s n:S I and J' is the stabi-
lized productivity index. For n = I, J 1 is given by

J 1 =J [1I(2p)](kro/ JLoBo)j5' ......................... (28)

To compare the results given by Eqs. 26 and 27, we rewrite Eq.


27 as follows for n= I (the two equations can be compared only
for this case):
qo/qo,max = I-(P;"f/ p2) . .......................... (29)
Responses given by Eqs. 26 and 29 are compared in Ref. 35.
For P<Pb, the variation in kro/(JLoBo) with pressure may be ap-
proximated by a straight line with the intercept equal to zero,9 then
the absolute OFP, qo,max, is given by
1 10
FLOW TIME, t, years qO,max =J(kro/JLoBo)j5 (p/2) . ....................... (30)

If we assume that the variation in kro /(JLoBo) is negligible, then


Fig. 9-Type-curve match of rate data-locke-Sawyer type
the absolute OFP, assuming that the conventional definition of J
curve.
is valid, is given by

Aanonsen agrees that Raghavan's procedure for computing the qo,max,c=2qo,max' ................................ (31)
pseudopressure predicts the theoretical response well because it ac-
That is, if we assume that the variation in kro /(JLoBo) with pressure
counts for the correct initial condition (saturation profile at .:It=O)
is negligible, then we overpredict the absolute OFP by a factor of
needed to obtain a solution to the problem at hand. Aanonsen also
two.3 6 This result emphasizes the importance of considering the
showed that the difference between the solution-gas-drive responses
heterogeneous character of solution-gas-drive systems.
(dotted lines) and the liquid responses plotted in Ref. 6 is negligible
Unfortunately, the Vogel correlation assumes that s=O. To extend
if the responses are plotted in terms of pseudotime. 34
the correlation for s>O, Standing 37 suggests that one should first
Fig. 6 shows the gas-saturation profile in the reservoir after the
calculate the well pressure that would have been measured if s=O
well has been shut in, and demonstrates that, contrary to
and then calculate the productivity of the well with the Vogel
Whitson's32 claim, the GOR during the buildup period changes as
equation. His procedure assumes that the "infinitesimally thin skin"
a function of both distance and time. From this plot, it is clear that
concept of van Everdingen 38 and Hurst 39 is valid. Unfortunately,
Whitson's contention regarding the GOR during the buildup period
Standing assumes that the pressure drop across the skin region obeys
is false and that many of his strong assertions are incorrect.
single-phase flow theory and not the quadratic equation given by
Computing buildup pseudopressures by the producing GOR
Eq. 26. Thus, his results are erroneous because none of his solutions
method yields the correct pseudopressure to analyze buildup data.
As expected, this method works best during the boundary-dominated show that qot qEf=1 =Ef at PWj=O, and the problems with his
o,max
flow period and works well if SgCr) > Sgc, where ris the distance method are starkly evident if s<O, then in some cases, the equivalent
at which P (r) = p. On the basis of our knowledge of solution-gas- value of well pressure, Pwf, becomes negative. The procedures to
drive reservoirs, we do not expect this method to yield useful re- overcome the problem of negative pressures involve reverting to
sults at early times. It must be recognized that the only thing that Fetkovich's log-log procedure. 9 (In this case, the correct form of
is common to this method and the steady-state theory is the ex- the equation is used.)
pression of the producing GOR, which is a defining relation. The A serious impediment to Vogel's method is that additional as-
steady-state theory yields a single curve for the saturation/pressure sumptions must be made to determine the future inflow-performance
relationship. The producing GOR method yields saturation/pressure curve because the well's performance curve changes in a complex
relationships that are functions oftime. We see this in Fig. 7, which manner with saturation and pressure. Standing 40 suggested that
shows the results for a system produced at constant pressure. The future performance can be calculated by assuming that the produc-
unbroken lines represent the saturation profiles in the reservoir at tivity index at zero drawdown, J*, is a linear function of
the instant the well is shut in. The data points (circles, squares, [kro/(JLoBo)]p. J*= 1.8qo,max/ P (Vogel correlation) and 2qo,max/ P
and triangles) represent the saturation/pressure relationship obtained (Fetkovich correlation, n = I), respectively. Fetkovich 9 suggests
by Raghavan's method. For p<p, agreement is quite good. The that the permeability to oil changes linearly with p.
chain-dotted line represents the saturation/pressure relationship pre- Other techniques to predict future deliverability have also been
dicted by the steady-state theory. It is clear that the steady-state proposed. All techniques assume that there is a linear relationship
theory is inadequate for the simulations considered here; however, between kro/(JLoBo) andp (or p). They differ only in the value of
from the simulations of Ref. 7, the steady-state theory does work the intercept used to define the linear relationship; e.g., Fetkovich
adequately for the transient-flow period if Pi> Ps, where Ps is the suggests that the intercept be zero. Whitson 41 and Uhri and
saturation pressure (Pb or Pdew), and the well is produced at con- Blount 42 assume that the intercept equals [k ro /(JLoBo)]j5/9; Ref. 43
stant pressure (Pwf<Ps)' assumes that it is an undetermined function of time. When com-
pared with results predicted from simulations, these methods yield
Inflow Perfonnance of Solution-Gas-Drive Reservoirs. Probably inconclusive results. Additional fundamental work is required to
no study of solution-gas-drive reservoirs would be complete without predict future performance. .
reference to the Vogel 8 correlation. Vogel's objective was to The influence of non-Darcy flow is also ignored in Vogel's work.
obtain an expression to determine the deliverability of a well in Field examples presented by Fetkovich clearly indicate that non-
reservoirs produced by solution-gas drive. His deliverability Darcy flow can be significant. Depending on test conditions, it can
equation is given by be shown that Eqs. 26 and 29 will overpredict flow rates by 50%.
2 Details on the influence of non-Darcy flow are discussed in Ref. 41.
~=1_0.2Pwf _0.8 Pwf . . . . . . . . . . . . . . . . . . . . . . . (26) Whitson also presents methods to incorporate non-Darcy flow effects
qo,max P p2 in the inflow performance equations of Vogel and Fetkovich.
590 SPE Fonnation Evaluation, December 1989
~ 10 ~.-~"nn--.-nrrnrrr-'rT~TITrr--.-nrnmm
lQ.
W
1 I
C)
Z I I
<t
:I: I I I ooooEPi
(.)
I I1 oo~~
p .,. AI
~ I 0
::) ~o.~AA ~ A
(/)
I ~
In
a:: I1
0
1
Q. o~ 1
(/) 0<9 I TEST I TEST 3
(/)
W I RESPONSE - -GRID --GRID
...J
Z I PRESSURE o o
Q
(/) I DERIVATIVE A

z I I
'0 'OL_"'"2-'---'--LU-'-,L.l.'"'_I:--'---L...L...L..LLL.IJ.,_...L-.L.J..L..LJUJ,u--"'--'-'-..............
,02 w
0 0 ~ 162~~~~~~-'-~~llL__LJ~~ll-~~~~
SHUT-IN TIME, At, hours ~ 10- 2 '0- 1 I I 10 10 10 2
SHUT-IN TIME, t.t, hours
Fig. 10-Type-curve match of falloff response (pressure and
derivative) of Test 1. Fig. 11-0verlay of Test 1 response on Test 3 response.

Production Forecasting. The prime objective of a well test is to a wide variety of conditions (see Fig. 4 of Ref. 44 for a composite
produce a production forecast. For single-phase flow, the response plot). Fetkovich also provides the following guidelines for the ex-
during boundary-dominated flow for a well that is produced at a pected range of values of b for various producing mechanisms: b=O
constant pressure (for long-term forecasting this is the appropriate if the reservoir is highly undersaturated or for gravity drainage with
boundary condition) and is located at the center of a closed cir- no free surface; b=0.5 for gravity drainage with a free surface;
cular reservoir is given by44 b=0.667 for a solution-gas-drive reservoir if p is a linear function
of cumulative oil produced, N p ; and b=0.333 if p2 vs. Np is a
exp[ -2tD I(r;D In 0.472r eD)] linear relationship. Of course, reservoir heterogeneities and
.. (32)
In(0.472r eD)
producing conditions can also influence the value of b. Under no
circumstances is b> 1. If data suggest that b> 1, \ben Fetkovich 44
Here, reD=re1rw and tD is given by Eq. 19. rather explicitly states that the analyst is using transient data to
For solution-gas-drive reservoirs, single-well production forecasts predict boundary-dominated responses. Unfortunately, this point
are usually conducted as Arps45 suggested. The characteristics of is often ignored in many predictions presented in the literature. If
the late-time responses-exponential, hyperbolic, and harmonic-are one ignores this and uses transient decline rates (indicated by b> 1)
well known and will not be discussed here. In 1980, Fetkovich 44 to predict boundary-dominated flow, then prediction of production
suggested that Arps' results may be combined and expressed in the rates and cumulative production will be unduly optimistic (see Fig.
following form: 12 of Ref. 46). In essence, the Arps equations or Fig. 8 should
not be used unless boundary-dominated rate data are available.
q(t)=qi/[1 +bDitd] lib . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (33)
Decline-curve analysis cannot be conducted unless the well is on
for 0<b::51, and by q(t)=q; exp(-D;td)' ............. (34) decline.
Example Application. Fig. 9 is a type-curve match of rate data
for b=O. Here, b is the reciprocal of the decline-curve exponent
measured at a fractured well. (This example is discussed in Ref.
(first-difference-of-the-loss ratio). The generally accepted values
46.) Data are matched with the Locke-Sawyer 47 type curve, which
of bare b=O (exponential decline), 0< b< I (hyperbolic decline),
was developed for a single well intercepted by an infinite-
and b= I (harmonic decline); this point will be discussed in more
conductivity vertical fracture and located at the center of a square
detail later.
drainage region of length 2Lx . Here, tXfD is based on LXf. The
Note that the Arps equations were derived empirically. Fet-
Lx ILxf=oo line is the respon~e of a fractured well in an infinite
kovich,44 however, recognized that exponential decline was
sy~tem, and the responses corresponding to finite values of Lx IL xf
common to both the empirical relations of Arps and the theoretical
represent the influence of boundaries. Note that this type curve was
expression for single-phase liquid flow during the boundary-
developed for single-phase liquid flow. The type-curve match
dominated period (Eq. 32). Thus, he was able to combine the em-
suggests that this well is located in a system where Lx ILxf= 1.5
pirical relations of Arps with the expression for dimensionless rate
and that boundary-dominated flow begins at about 1 yea~. One can
during boundary-dominated flow. Fig. 8 is a log-log plot of the
readily predict future performance by extrapolating along the
Arps relations (Eqs. 33 and 34) for several values of b. Here qdD
Lx IL xf = 1.5 curve. For example, this match indicates that the
and tdD are defined as
flo~ rate at the end of II years would be 52 bbllmonth [0.28 m 3 /d]
re 1)
( In--- qD .......... (35)
oil and that at 21 years qo=10 bbl/month [0.053 m 3 /d] oil.
Because the Locke-Sawyer type curve is for single-phase flow
rw 2 and the data shown in Fig. 9 follow the Lx ILxf= 1.5 stem, we
tD
would expect the boundary-dominated flow data to follow the b=O
and tdD =Ditd = 2 ............. (36) solution of the Fetkovich type curve. Fig. 28 of Ref. 46 shows that
0.5(reD -l)(ln r eD-0.5) late-time data follow the b=O stem of Fig. 8.
Once a match with the Fetkovich type curve (Fig. 8) is obtained,
Note that for tdD <0.3, all curves in Fig. 8 coincide and are in- then the match points can be used to calculate formation properties,
distinguishable. Thus, it is impossible to determine b with data decline parameters, and original oil in place (OOIP); however, it
before tdD =0.3. A statistical analysis of data for tdD::5 0.3 will fit is beyond the scope of this paper to explore these aspects (see Refs.
any value of b in the range 0::5b::51 (see Ref. 44). 44 and 46 for details). Ref. 46 also presents details on how to ac-
As already mentioned, the b=O curve is common to the empirical count for changes in backpressure.
relations of Arps and the theoretical expression for flow rate for
single-phase flow; i.e., if the reservoir is highly undersaturated, Injection-Well Analysis. The principal objectives of injection-well
then b=O. This observation also lets us compare transient rate testing are similar to those of production tests. Because our purpose
decline for single-phase flow with boundary-dominated decline for is to ensure that injectivity is unimpaired, accurate determination

SPE Fonnation Evaluation, December 1989 591


of the skin factor is of paramount importance. Also, in mjscible responses for Tests 1 and 3, and not to improve the type-curve match
flooding projects, the detennination of average reservoir pressure shown here. Estimates of (kh/ p.)t from the type-curve match agree
is of interest. Very little information is available in the literature with the semilog-straight-line analysis.
on the determination of average reservoir pressure when the mo- Test 3 was conducted after one cycle of gas injection (127,290
bility ratio, M, is different from unity. In most cases, especially Mcf [3604 x 103 m 3 ] of gas) and one cycle of water injection
in EOR projects, engineers want to defermine the location of fluid (17,259 STB [2744 stock-tank m 3 ]). The well was restimulated
banks. Although theoretical calculations and intuition suggest that during the water-injection cycle to improve injectivity. (A loss in
it is possible to do so by pressure analysis, in actual practice it is injectivity following the injection of a compressible fluid is not un-
difficult to locate fluid banks by falloff testing. In the following expected.) Fig. 11 is an overlay of the responses from Tests 1 and
we restrict our attention to the analysis of falloff data. 3. Here, the change in pressure during the test is expressed in dimen-
The composite reservoir model with a moving interface is the sionless form. The agreement between the two tests in terms of
most commonly recommended model for analyzing injection-well the pressure response and the derivative response is remarkable.
data. Most of the recent work in this area stems from the paper The implications of this observation are obvious. The displacement
by Ramey. 10 Using the ideas proposed by Ramey, Merrill et al. 11 in the time scales (dashed line represents grid for Test 1) reflects
examined falloff responses in a system similar to a composite the influence of increased compressibility and the change in fracture
reservoir. Their work indicates that if we neglect the unaffected length. From the displacement in the time scales, we may estimate
region and run the test long enough, then two semilog straight lines (ctL} h/(ctLth to be 3.83. This information can be used to
are evident. The slope of the first straight line reflects the flow ca- predi~t future Injection rates and to revise earlier estimates, if nec-
pacity of the injected fluid bank, and the slope of the second straight essary. (In this particular case, the flood was designed so that 0.9
line is a function of the mobility-thickness and the porosity- res bbl [0.14 res m 3 ] of rich gas would be injected per barrel of
compressibility contrasts between the two zones. Intuitively, one water. Because of the unanticipated reduction in water injectivity,
does not expect the porosity-compressibility product to influence this ratio was revised to 4: 1 gas/water. The economic consequences
the slope of the second straight line because the existence of a second are clearly evident.) Most interesting from my viewpoint is that
semilog straight line implies that "steady conditions" prevail in there is nothing in the trace corresponding to Test 3 to suggest that
the inner zone. Barkve 48 shows that the ratio of the slopes of the more than one fluid was injected. Similar results were obtained for
two straight lines should be a function of the mobility-thickness con- other wells. Experience indicates that this behavior is the norm even
trasts and not the porosity-compressibility contrasts. He suggests if fluids with small contrasts in compressibility (e.g., graded polymer
that the results of Merrill et at. are affected by boundary effects. slugs) are injected. Because much of what is in the literature does
Thus, their results should not be used. not appear to be directly useful for the type of tests considered here,
Sosa et at.49 examined the influence of saturation gradients on I usually use the well response before the injection of slugs as the
pressure-falloff data. They concluded that the Buckley-Leverett50 baseline solution from which mobility changes and injectivity losses
frontal-advance equation may provide a better estimate of the radius are evaluated. In essence, I use the field curve corresponding to
of the flood front than the correlations presented by Merrill et al. II the initial test (water) as the type curve.
Much work needs to be done in this area of pressure analysis, par-
ticularly in the determination of the flood-front location. One Concluding Remarks
analysis method for determining fluid banks that has been completely
From a research viewpoint, one must be careful in deriving con-
overlooked is Peaceman's probe-radius formula (Eq. 21 of Ref.
clusions of a general nature from a limited set of observations. For
51). If radial flow or pseudoradial-flow conditions prevail, then
example, based on the works in Refs. 6 and 7, it is readily obvious
the equation lets us compute the flow capacity to phase m, (kh)m'
vs. the distance from the wellbore, r, by the derivative method. that observations that apply for Pi "'" Pb (or Pdew) may not be ap-
First, we obtain instantaneous values of At as a function of pressure plicable if Pi > Pb (or Pdew)' As stated in Ref. 7, for the devel-
opment of a theory for multiphase flow, three concerns must be
by an expression similar to Eq. 7, where m represents the instan-
addressed: (1) the existence of a single-phase region must be in-
taneous slope. Then we use Peaceman's probe-radius formula to
corporated; (2) conclusions must be derived from simulations that
compute (kh)m as a function of the probe radius, r m'
incorporate the existence of the skin region (preferably a thick-skin
From a practical viewpoint, several factors inhibit our applying
region); and (3) conclusions must be based on solutions that incor-
the techniques discussed in the literature, particularly in the location
porate situations where critical saturations are greater than zero.
of flood fronts. These include (1) wellbore storage effects (the
In this area of reservoir engineering, we expect to derive conclusions
difficulty in analyzing data dominated by storage is exacerbated if
of general import or rules based on a limited number of numerical
the storage constant changes during the test, and the storage con-
experiments. It is imperative that the validity of such conclusions
stant, if it does change, usually increases during falloff tests), (2)
be demonstrated by considering exceptions that probe (not prove)
the existence of vertical fractures intercepting the wellbore (in the
such rules. Even though I say so myself, Ref. 7, in hindsight, seren-
U.S.A. the fractured injection well is the norm), (3) inadequate
test time, and (4) well spacing. Most of the theories discussed in dipitously fulfills these requirements.
the literature appear to be suitable only for mature waterfloods;
this situation, however, involves only the flow of a single fluid. Nomenclature
Example Application. We consider two falloff tests following A = drainage area, ft2 [m 2]
water injection in a tertiary miscible project. Data pertaining to this b = reciprocal of the decline-curve exponent
dolomite reservoir are as follows: depth=4,9oo ft [1494 m], net Bm = FVF of Phase m, RB/STB or RB/scf
pay = 100 ft [30 m], k=2.1 md, T=105F [41C], Pi=I,690 psi [res m 3/stock-tank m3 or res m 3 /std m 3]
[11.6 mPa], crude viscosity = 1.93 cp [1.93 x 10-3 Pa' s], oil c = compressibility, psi - I [kPa - I]
gravity=30oAPI [0.88 g/cm 3], B o =1.16, and miscibility pressure cf = PV compressibility, psi -I [kPa -I]
::: 1,100 psi [7.6 mPa]. Test 1 was conducted before the start of C m = compressibility of Phase m, psi-I [kPa- l ]
the miscible flood and after the injection of about 123,600 STB [19 C t = effective or pseudocompressibility, psi -I [kPa -I]
651 stock-tank m 3 ] of water. The well had been acid fractured
C = concentration of free gas (0.006328), ft 3 1D [m 3/d]
without a propping agent. Fig. 10 is a type-curve match of the falloff
response with the uniform-flux type curve (Test 1).52 The inverted CA = geometric shape factor
triangular data points are the derivatives of the pressure/time Di = initial decline rate, day-lor month-I
response. The derivative response is plotted as 2flp't:.t vs. t:.t, where Ef = flow efficiency
t:.p'=d(flp)/t:.t. For data plotted in this manner, the pressure- h = formation thickness, ft [m]
response curve and the derivative-response curve are identical at I = productivity index, defined by Eq. 11, (RBID)/psi
early times. Thus, plotting data this way helps to match pressure [(res m 3/d)/kPa]
and its derivative simultaneously. The bottom line in Fig. 10 is a j = defined by Eq. II
plot of 2p;"DtxfD vs. txfD, where P;"D =dpwD /dtxfD' The purpose I' = stabilized productivity index, (STBID)/(psi2)n
in computing the derivative response is to compare the derivative [(stock-tank m3/d)/(kPa 2)n]
592 SPE Formation Evaluation, December 1989
J* = productivity index at zero drawdown, (STBID)/psi Subscripts
[(stock-tank m3 /d)/kPa] bu = buildup
k = absolute permeability, md c = critical
k m = effective permeability of Phase m, md dd = drawdown
krm = relative permeability of Phase m f = flood front; future conditions
L = moles of liquid phase g = gas
Lx = distance from well center to external boundary i = initial
e (square drainage region), ft [m] j,m = index for phase of fluid (0, w, or g)
L Xj = fracture half-length, ft [m] 0 = oil
m = slope of semilog straight line, psi [kPa] P = present conditions
M = mobility ratio r = relative, ratio
n = exponent of inflow-performance curve t = total
Np = cumulative oil produced, STB [stock-tank m3 ] w = water
P = pressure, psi [kPa] 1,2 = zone
p = average reservoir pressure, psi [kPa]
Pb = bubblepoint pressure, psi [kPa] Superscript
Pdew = dewpoint pressure, psi [kPa]
= average
Pe = external boundary pressure, psi [kPa] Acknowledgments
Pi = initial pressure, psi [kPa]
The financial support of the Dept. of Petroleum Engineering at the
pp(p) = pseudopressure function (Eq. 14), (psi/cp)/(RB/STB)
U. of Tulsa is appreciated. The assistance of colleagues Tai Tu Cao,
[(kPa/Pa's)/(res m3 /stock-tank m3 )] Somporn Vongvuthipornchai, and Nai-Shyong Yeh is gratefully ac-
PpsD = dimensionless pseudopressure rise knowledged. I take this opportunity to express my thanks to M.J.
PpwD = dimensionless pseudopressure Fetkovich for many fruitful discussions on multiphase flow through
Ps = saturation pressure, psi [kPa] porous media over the past 10 years.
PsD = dimensionless pressure rise
PwD = dimensionless wellbore pressure References
P;"D = derivative of PwD with respect to tXlD 1. Matthews, C.S. and Russell, D.G.: Pressure Buildup and Flow Tests
Pwj = wellbore flowing pressure, psi [kPaJ in Wells, Monograph Series, SPE, Richardson, TX (1967) 1; 22-24.
Pwj,s = wellbore flowing pressure at shut-in, psi [kPa] 2. EarJougher, R.C. Jr.: Advances in Well Test Analysis, Monograph
Pws = shut-in pressure, psi [kPa] Series, SPE, Richardson, TX (1977) 5, 74-89.
ilp' = derivative of t.p with respect to t.t 3. Perrine, R.L.: "Analysis of Pressure Buildup Curves," Drill. & Prod.
Prac., API, Dallas (1956) 482-509.
q = flow rate, STBID and MscflD [stock-tank m3 /d and 4. Martin, J.C.: "Simplified Equations of Flow in Gas Drive Reservoirs
std m3 /d] and the Theoretical Foundation of Multiphase Pressure Buildup
qdD = decline curve dimensionless rate Analyses," Trans., AIME (1959) 216, 309-11.
qD = dimensionless rate (Eq. 32) 5. AI-Hussainy, R., Ramey, H.J. Jr., and Crawford, P.B.: "The Flow
qi = flow rate at time decline begins (reference rate) of Real Gases Through Porous Media," 1PT (May 1966) 624-36;
qo,max = maximum flow rate
Trans., AIME, 237.
6. Raghavan, R.: "Well Test Analysis: Wells Producing by Solution Gas
r = radial distance, ft [m] Drive," SPEl (Aug. 1976) 196-208; Trans., AIME, 261.
rD = dimensionless distance 7. Jones, R. and Raghavan, R.: "Interpretation of Flowing Well Responses
re = external drainage radius, ft [m] in Gas Condensate Wells," SPEFE (Sept. 1988) 578-94; Trans., AIME,
reD = dimensionless reservoir radius 285.
r w = wellbore radius, ft [m] 8. Vogel, J.V.: "Inflow Performance Relationships for Solution-Gas Drive
R = producing GOR, scf/STB [std m 3 /stock-tank m 3 ] Wells," 1PT (Jan. 1968) 83-92; Trans., AIME, 243.
9. Fetkovich, M.J.: "The Isochronal Testing of Oil Wells," paper SPE
Rs = solution GOR, scf/STB [std m 3 /stock-tank m 3 ] 4529 presented at the 1973 SPE Annual Meeting, Las Vegas, Sept. 30-
Rsw = solution gas/water ratio, scf/STB Oct. 3.
[std m 3/stock-tank m 3 ] 10. Ramey, H.J. Jr.: "Approximate Solutions for Unsteady State Liquid
s = skin factor Flow in Composite Reservoirs," 1. Cdn. Pet. Tech. (Jan.-March 1970)
Sm = saturation of Phase m 32-37.
Sor = critical oil saturation 11. Merrill, L.S., Kazemi, H., and Gogarty, W.B.: "Pressure Falloff
Analysis in Reservoirs With Fluid Banks," 1PT(July 1974) 809-18;
Sw = average saturation behind the front predicted by the Trans., AIME, 257.
Buckley-Leverett theory 12. Muskat, M. and Meres, M.W.: "The Flow of Heterogeneous Fluids
t = time, hours, days, or months Through Porous Media," Physics (Sept. 1936) 7, 346-63.
td = decline-curve time, hours, days, or months 13. Wycoff, R.D. and Botset, H.G.: "The Flow of Gas-Liquid Mixtures
tdD = decline-curve dimensionless time Through Unconsolidated Sands," Physics (Sept. 1936) 7, 325-45.
tD = dimensionless time based on r w 14. Miller, C.C., Dyes, A.B., and Hutchinson, C.A. Jr.: "Estimation of
Permeability and Reservoir Pressure from Bottom-Hole Pressure Build-
tXjD = dimensionless time based on L Xj up Characteristics," Trans., AIME (1950) 189,91-104.
t.t = shut-in time, hours 15. Macias, C.L.: "Multiphase Multicomponent Compressibility in Pe-
t.tD = dimensionless shut-in time based on r w troleum Reservoir Engineering," PhD dissertation, Stanford U.,
t.tsD = dimensionless shut-in time defined by Eq. 25 Stanford, CA (1985) 134-36.
T = temperature, of [0C] 16. Ramey, H.J. Jr.: "Rapid Methods for Estimating Reservoir Compres-
V = moles of vapor phase sibilities," 1PT (April 1964) 447-54; Trans., AIME, 231.
17. Matthews, C.S., Brons, F., and Hazebroek, P.: "A Method for De-
Vp = PV, bbl [m 3 ] termination of Average Pressure in a Bounded Reservoir," Trans.,
xi = components in the liquid phase AIME (1954) 201, 182-91.
Yi = components in the gas phase 18. Earlougher, R.C. Jr., Miller, F.G., and Mueller, T.D.: "Pressure
'Y = Euler's constant, 0.57721 ... Buildup Behavior in a Two-Well Gas-Oil System," SPEl (June 1967)
A = mobility (k/JL), md/cp [md/Pa's] 195-204; Trans., AIME, 240.
JL = viscosity, cp [Pa' s] 19. Odeh, A.S. and Jones, L.G.: "Two-Rate Flow Test, Variable-Rate
Case-Application to Gas-Lift and Pumping Wells," 1PT (Jan. 1974)
JLm = viscosity of Phase m 93-99.
p = molar density, Ibm mollft 3 [kmollm 3 ] 20. Dietz, D.N.: "Determination of Average Reservoir Pressure From
</> = porosity Buildup Surveys," 1PT (Aug. 1965) 955-59; Trans., AIME, 234.

SPE Formation Evaluation, December 1989 593


21. Winestock, A.G. and Colpitts, G.P.: "Advances in Estimating Gas Well 40. Standing, M.B.: "Concerning the Calculation ofInflow Performance
Deliverability," J. Cdn. Pet. Tech. (July-Sept. 1965) 111-19. of Wells Producing Solution Gas Drive Reservoirs," JPT(Sept. 1971)
22. Uraiet, A.A. and Raghavan, R.: "Pressure Buildup Analysis for a Well 1141-42.
Produced at Constant Bottomhole Pressure," JPT(Oct. 1980) 1813-24. 41. Whitson, C.S.: "Reservoir Well Performance and Predicting Deliver-
23. Raghavan, R.: "The Effect of Producing Time on Type Curve ability," paper SPE 12518 available at SPE, Richardson, TX.
Analysis," JPT (June 1980) 1053-64: 42. Uhri, D.C. and Blount, E.M.: "Pivot Point Method Quickly Predicts
24. Chu, W.C., Reynolds, A.C., and Raghavan, R.: "Pressure Transient Well Performance," World Oil (May 1982) 153-64.
Analysis of Two-Phase Flow Problems," SPEFE (April 1986) 151-64. 43. Kelkar, B.G. and Cox, R.: "Unified Relationship to Predict Future
25. Hazeltine: "Discussion," Drill. & Prod. Prac., API, Dallas (1956) 509. IPR Curves for Solution Gas-Drive Reservoirs," paperSPE 14239
26. Muskat, M.: "The Theory of Potentiometric Models," Trans., AIME presented at the 1985 SPE Annual Technical Conference and Exhibition,
(1949) 179, 216. Las Vegas, Sept. 22-25.
27. Evinger, H.H. and Muskat, M.: "Calculation of Theoretical Produc- 44. Fetkovich, M.J.: "Decline Curve Analysis Using Type Curves," JPT
tivity Factor," Trans., AIME (1942) 146, 126-39. (June 1980) 1065-77.
28. Chopra, A.K. and Carter, R.D.: "Proof of the Two-Phase Steady-State 45. Arps, J.J.: "Analysis of Decline Curves," Trans., AIME (1945) 160,
Theory for Flow Through Porous Media," SPEFE (Dec. 1986) 603-08. 228-47.
29. Levine, J.S. and Prats, M.: "The Calculated Performance of Solution- 46. Fetkovich, M.J. et al.: "Decline-Curve Analysis Using Type Curves-
Gas-Drive Reservoirs," SPEI (Sept. 1961) 142-52; Trans., AlME, 222. Case Histories," SPEFE (Dec. 1987) 637-56; Trans., AIME, 283.
30. Muskat, M.: "The Production Histories of Oil Producing Gas-Drive 47. Locke, C.D. and Sawyer, W.K.: "Constant Pressure Injection Test
Reservoirs," J. Appl. Phys. (March 1945) 147-59. in a Fractured Reservoir-History Match Using Numerical Simulation
31. Bl'e, A., Skjaeveland, S.M., aDd Whitson, C.S.: "Two-Phase Pressure and Type Curve Analysis," paper SPE 5594 presented at the 1975 SPE
Test Analysis," SPEFE (Dec. 1989) 604-10. Annual Meeting, Dallas, Sept. 28-0ct. 1.
32. Whitson, C.S.: "Topics on Phase Behavior and Flow of Petroleum 48. Barkve, T.: "A Study of the Verigin Problem with Application to
Reservoir Fluids, " PhD dissertation, U. of Trondheim, Norway (Aug. Analysis of Water Injection Tests," Report No. 78, Dept. of Applied
1983). Mathematics, U. of Bergen, Norway (March 1985).
33. Aanonsen, S.I.: "Nonlinear Effects During Transient Fluid Flow in 49. Sosa, A., Raghavan, R., and Limon, T.J.: "Effect of Relative Perme-
Reservoirs as Encountered in Well Test Analysis," PhD dissertation, ability and Mobility Ratio on Pressure Falloff Tests," JPT(June 1981)
U. of Bergen, Norway (1985). 1125-35.
34. Agarwal, R.G.: '''Real Gas Pseudo-Time'-A New Function for 50. Buckley, S.E. and Leverett, M.C.: "Mechanism of Fluid Displacement
Pressure Buildup Analysis of MHF Gas Wells," paper SPE 8279 in Sands," Trans., AIME (1942) 146,107-16.
presented at the 1979 SPE Annual Technical Conference and Exhibition, 51. Peaceman, D.: "Interpretation of Well-Block Pressures in Numerical
Las Vegas, Sept. 23-26. Reservoir Simulation," SPEI (June 1978) 183-94; Trans., AIME, 265.
35. Fetkovich, M.J. et al.: "Case Study of a Low-Permeability Volatile 52. Gringarten, A.C., Ramey, H.J. Jr., and Raghavan, R.: "Unsteady-
Oil Field Using Individual-Well Advanced Decline Curve Analysis," State Pressure Distributions Created by a Well with a Single Infinite-
paper SPE 14237 presented at the 1985 SPE Annual Technical Con- Conductivity Vertical Fracture," SPEI (Aug. 1974) 347-60.
ference and Exhibition, Las Vegas, Sept. 22-25.
36. Bertuzzi, A.F. et aI.: "Wellbore Hydraulics," Petroleum Engineering
SI Metric Conversion Factors
Handbook. H.B. Bradley (ed.), SPE, Richardson, TX (1987) bbl x 1.589 873 E-Ol m3
30-34-34-35. ft x 3.048* E-Ol m
37. Standing, M.B.: "Inflow Performance Relationships for Damaged Wells psi x 6.894 757 E+OO kPa
Producing by Solution-Gas Drive," JPT(Nov. 1970) 1399-1400.
38. van Everdingen, A.F.: "The Skin Effect and Its Influence on the
'Conversion factor is exact. SPEFE
Original SPE manuscript received for review March 14. 1986. Paper accepted for publi-
Productive Capacity of a Well," Trans., AIME (1953) 198,171-76. cation July 11. 1989. Revised manuscript received Nov. 29. 1988. Paper (SPE 14098) first
39. Hurst, W.: "Establishment of the Skin Effect and Its Impediment to presented at the 1986 SPE Inti. Meeting on Petroleum Engineering held in Beijing. March
Fluid Flow Into a Wellbore," Pet. Eng. (Oct. 1953) 25, B-6. 17-20.

594 SPE Formation Evaluation, December 1989

S-ar putea să vă placă și