Beruflich Dokumente
Kultur Dokumente
By
Nino Hadinoto, Member SPE-AIME, R. Raghavan, Member SPE-AIME, and
G.W. Thomas, U. of Tulsa
.. .. ... -.
depends on the physicalpropertiesof the reser- flow prevails. To the best knowledgeof the
voir, the propertiesof the flowingfluid, the authorsthere has been no investigation
location,extent,and unless flow conditionsare regardingapplicationof this techniqueto
stabilized,u~n time. The exponentn depends fracturedgas wells where the flow regime can
on the flowingconditions,namely,Dsrcy or non- change during a test. The second objectiveof
Darcy flow. this paper is to test the applicabilityof the
0deh40nes method to fracturedgas wells.
Eq. 1 assumesthat the gas is ideal snd
flows under steady-state,isothermalconditions GEW3WlZONOF GAS-DEJXVIRABIIZTY
DATA
where pressuregradientsare small. If the well
has stabilizedthe flow exponent,n, the In this study,gas-deliverability data for
constantC may be evaluatedand the absolute the fracturedwell was obteinedusLng a two-
open flow potential(AOF),which is a measure of dimensional,single-phasemathematicalmodel
the quality of the well, can be obtained. that rigorouslyincorporatesvariationin fluid
properties. The well.is assumedto be located
Eq. lmaybe modifiedto accountfor the at the center of a vertical.fracturethat
flow of real.gases and cases where pressure totally penetratesa horizontal,homogeneous,
gradientsare not small by use of the real gas and isotropicsquarereservoir. The vertical.
pseudo pressure,jm(p), as: fractureis locatedsymmetricallyin a square
reservoirand is parallelto one of the sides
q =C[m(F)-p@pwf)ln, s . . . . (2) of the square boundary (Fig. 1). Two fractuxe
penetrationratios, xe/xf, (1 and 5), three
where pressure levels (I06L7, 5,CKX3,and 10,MIOpsi)
and two values of permeability(1 end 0.5 red),
P and two sets of PVT data (G .0.5$ andO.7)
pdp were examined. The fully penetratingcase
m(p) = 2 .***,...(3) (dq = 1) was chosen to investigatecases
~ where the entire deliverabiMty data are domi-
pb z
nated by the linear flow period. At time t X,
As alreadymentioned,Eqs. 1 and 2 are 0, gas is producedfrom the well at a consiaiii
based on the assumptionthat flow is radial; surfacerate.
that is, pressureis a Mneer functionof the
logarithmof flow time. Butit is well known This investigationmodeled the fracture
.
that in many instances,particularlyin low plane by assumingthe flux distributionalong
permeabilitywells, stimulationin the form of the fractureto be uniform. Followingthe
= hy&auMc fracturingor acid fracturingis nomenclaturein Ref. 4, this type of fracture
requiredfor wells to be commercial. In such will be referredto as a uniform flux fr~~cture,
cases, pressuresare more likelyto be functions This impliesthat pressurealong the frmtare
of squareroot of time (liiiear flow), transi- will vary and the fracturehas a high, but not
tional or logarithmicfunctionsof time (pseudo infinite,conductivity. For the purposesof
radial flow), each for differentflow rates.~ this investigation,this assumptionis not a
In addition,a good fracturingjob on some gas limitationas all featuresof transientflow
wells could lead to a linear flow period long behaviordue to a fractureintersectinga well.
enough to encompassthe entire deliverability are preserved. The generalresults are also
testingperiod. To date the applicabilityof applicableto wells intersectinginfinite
Eqs. 1 or 2 to fracturedgas wells has not been conductivityfractures. The reservoirpsrem-
investigated. However,field interpretationof eters and fluid characteristicsare given in
pressuredata from such wells based on radial Table 1.
flow solutionsin some instancesat least
appearsto provide~sonable answersfor An importantconsiderationin the solution
heavily fracturedwells. This difference of finite-difference equationsis the effectof
between field results and theoreticalexpec- discreti.zationin space and time. For exemple,
tati.onsbased on analyticalsolutionsimplies in modelingflow behaviorin a fracturedsystem,
that there is a need for study. Thus, the first space discretizationnear the tip of the frac+
objectiveof this paper is to examinethe ture becomes important. The followingprocedurf
applicabilityof commonlyacceptedsolutionson was adoptedto insure that discretizationerrors
estimatingdeliverabilityof fracturedgas wells. did not affect the solutionsobtained.
As is well known, flow after flow data can Wattenbargerend Ramey6 have shown that
be used to calculateformationflow capacity, constantrate drawdownsolutionsfor fractured
kh, skin effect,s, and the non-Dsrcyflow gas wells may be correlate with constantprop-
coefficient,D, by use of the 0deh40nes tech- t along the lines
erty liquid flow solutions
nique.5 Like deliverabilitytests, this tech- suggestedby A1-Hussainyet ala Thus, for eacl
nique is also based on the premise that radial flow system considered,&%w~wn solutionswere
SPE 6136 NICO HADINOTO.R.
. . RAGHAVAN. end G. W. THOM4S
various flow periods for cases examinedin this higher than that obtainedfrom the draw-down
study are given in Table 3. Comparisonof flow equation,but AOF vakes are much lower since
times used in this study end the various flow n<l.
periods indicatesthat the durationof the iso-
chronaltests is such that all flow regimes Determinationof StabilizedPerfoimsnce
[linearflow period, transition,pseudo radial CoefficientFrom TransientData
flow pericd and transitionto pseudc-steady
state [stabilized)]have been investigated. As is well known, deliverabilitytests
are often not run long enoughto reach
Isoc!monalTestin~ of FracturedWells stabilizedflow conditions. The stabilized
flow coefiicientis usually obtainedfrom the
A sampleof the syntheticisochronaltest well lmown relation suggestedby Poetmsnnand
data generatedduring the course of this study Schilson,12which is based on plane radial flow
using the single-phasereservoirsimulatorare considerationsand is given by
shown in Figs. 2 through1+as circles,tri-
angles?crossestand square data potits. Cl In @ ~n
DetAls of these tests may be found in Ref, 11. c= ***,.*, (13)
The stabilizedbackpressurecalculatedfrom ~, ln(+ K)n
10 is shown as a broken line in each of these
where a/a is defined as
figures. For all cases except (~/~ = It pi =
1061.7psi) in this study,the isochronaltests .i
In
yield straightlines with a flow exponentequal C2
to unity. For Xe/~ = 1, pi = 1061.7,the flow
exponentwas equal to 0.911+for flow periodsof lln
2 (cl - c2in )
1 and 3 hours. For longer flow periods the
flow exponentwas 1 (See Fig. 2). This implies 2
a=
that~ even though flow regimes change, in T lln . , , . (14)
generalthe flow exponent,n, of the back- cl
pressurecurve can be obttied from a short-time
isochronaltest of a fracturedwell will be no l/n 1/n
differentfrom that of an unfracturedwell. 2 (cl -C2 )
Thus, it is impossibleto distinguishan un- tl
fracturedwell from that of a fracturedwell
and Cl and C2 are performancecoefficients
by exsminingdeliverabilitydata. The results
correspondingto a multipointflow test with
indicatethat if the floi~ tests are run long flow periods equal to tl and t2! respectively?
enough,the flow exponent,n, csn be assumedto
unity. and C is the desiredperformancecoefficier~i
correspondingto flow period,t,
Followingthe proceduresuggestedinRef.
If pressurebehavioris dominatedentirely
1, the absoluteopen flow potentialswere
calculated. The results sre summarizedin by the linear flow period, the deliverability
correspondingto the end of the linear flow
Tables 4 and 5, and are comparedwith values
period may be obtainedfrom13
predictedby Eq. 10 for app~opriateflowing
times. Calculatedvalues of the performance
c1 ( /T-l)n
coefficient,C, obtainedfrom isochronaltest
c= ~*.**.**** * (15)
data also are given in Tables 4 and 5. For
small times? isochronaldata indicatevalues ( a)n
differentfrom that of the drawdownresults. where Cl is the performancecoefficientcorre-
For some cases, values of AOF are lower, but in spendingto a multipointflow test with a flow
other cases they are higher than that predicted period of tl and C is the desiredperformance
by the drawdownequation. ior long times, coefficientcorrespondingfor any time during
agreementbetween drawdownand isochronaldata the Mnesr flow period. It shouldbe noted,
is fair. However, all values of AOF and C are however,that the meaning of the deliverability
higher than from the stabilizedvalues due to curve is not obvious in this case as k and xf
the short flow periods invnlved. It is also csnnot be determineduniquelyif data is domi-
seen that, if flow times are comparableto nated completelyby the linear flow period.
times for stabilization,then the isochronal
results comparefavorablywith stabilized In the following,we shall demonstrate
performance(Table2). errors that can result if the various flow
regimes controllingflow behaviorof fractured
Before proceedingfurther,one other point wells are not taken into account. Values of
deservesmention. As alreadymentionedfor the the performancecoefficientcalculatedfrom the
~/xf. 1, pi= 1061.7case, it was found that isochronaltests are comparedwith that pre-
the flow exponent,n, was differentfrom unity dieted by Eq. 13 in Fig. 5 for the.cases
for small times. Thus, values of C are much
SPE 6136 NICO HADINOTO,R. RAGHAVAN,and G. W. THOMAS
examinedhere. The solid lines in Fig. 5 were than that obtainedfrom longer-timeisochronal
obtainedu~ing I@. 13. Values of Cl, C2, tl, data.
end t2 used to predict the performancecoeffi-
cient correspondto that of the two lowest time Flow-After-FlowTests for FracturedWells
periods shown in Tables 1+and 5.
A few of the backpressurecurves obteined
The resultsin Fig. 5 indicatethat, if in this investigationfrom normal sequenceflow
the durationof the isochronaltest is such that after-flowtests are shown in Figs. $ and 9.
it is greaterthan the Iinea flow period~ The square,circular,end triangulardata
valuesof C predictedby %. 13 are in fair @nts representtest periods of 1, 3, end 24
agreementwith long time isochronaldata; see hours, respectively. The stabilizied back-
data for p. = 5000 psi; k= 1 md (star data pressurecurve calculatedfrom Eq. 10 is shown
points) an~pi = 1061.7 psi; k = lmd (square as a dashed line in each of these figures. In
data points). Howevertif the test duration general,the flow exponentfor the normal
is smallerthen the time for end of linear flow$ sequencecurves is equal to one. The only
then values by IQ. 13 are in poor agreementwith exceptionis xe/xf = 1 case for small times
long-timeisochronaland stabilizedflow data, (Fig. 8)0 Thus, in this instsncealso it would
Noting that the ordinatein Fig. 5 is l/C and be impossibleto detect a verticalfracture
not C, determinationof values of C using ~ 13 from a deMverabil.itycurve.
~ be larger (see, for eyle* %1~ = 5, pi =
10,000and 5,000,k = 0.5 md or smal er (see, Figs. 10 and 11 present sampleback-
for example,~/~ = 1, pi = 5,000). Thusf the pressurecurves obtainedfrom reverse sequence
AOF calculatedon the baszs of ~. 13 can be flow-after-flowtests in this study. The most
smalleror larger. Thus, it is quite possible surprisingand importantresult is that, for
that in some instancesdeterminationof AOF for small flowingtimes, flow exponentsfor the
heavily fracturedwells using radial flow reverse sequenceflow-after-flowtests are much
considerationscan lead to values which appear greaterthan unity. This phenomenonwas
reasonable. observedfor all penet~ationratios end} as
shown in Figs~O end 11, is not restrictedto
The applicabilityof ~. 15 is shown in the l-hourtests. We have carefullychecked
Fig. 6. The solid lines in Fig. 6 were obtained this result end are certainthis is not due to
by Eq. 15 using the results of the l-hourtests truncationerrors, etc. At long times the flow
for x~xf = 5, The square and circle, stex snd exponentis unity. To the best knowledgeof
triangledata @nts shown are values of C the authors,this type of behaviorhas no
obtainedfrom the isochronaltests. The results counterpartin plane radial flow. On the basis
indicatethat, even for the case where much of of Figs. 10 end 11, it can be concludedthat
the data is dominatedby linear flow, Eq. 15 is reverse sequenceflow-after-flowtests are
not adequate. As predicted,values of C are inadequatefor estimatingdeliverability.
much lower then that obtained from isochronal
data. Thus, Eq. 15 does not appear to be Analogousto the isochronalcase values
particularlyuseful.,even for predicting of AOF and C calculatedfrom the flow-after-
deliverabilityduring the Mnear flow period. flow tests were comparedwith the drawdown
Fig. 7 shows the correspondingresults for ~~ equation. For small values of time, the normal
% = 1 case. For this case also predicted sequenceflow-after-flowcurves predicted
values of C are much lower then that obtained lower values of unstabilizedopen flow poten-
from isochronalend, more important,do not tial thsn drawdowndata. For largertimes
follow the trend predictedby Eq. 15. Thus, agreementwas good. However, it was found that
Eq. 15 doesnot appear to be particularlyuse- in all cases isochronaltests were supericrto
ful even for predictingdeliverabilityduring flow-after-flow(normalsequence)tests for all
the linear flow period. times.
Reservoir Characteristics
TABLE 2
-5
Fracture Penetration Ratio Initial Pressure Permeability Stabilized AOF Stabilized C 10
3
%=/Xf pi, pai m~ q, MCF/D MC F/D/psi fcp
TABLE 4
(for xe/xf = 1)
-5
Flowtime AOF, MCF/1) C, MC F/ D/psi2/cp x 10
t, hours Isochronal Test Equation 10 Isochronal Test Equation 10
TABLE 5
(forxe/xf= 5)
-5
Flowtime AOF, MCl?/D C, MCF/D/psi2/cp x 10
t, hours From Isochronal Test Equation 10 From Isochronal Test Equation 10
(for x=/xf = 5)
TABLE 7
(for xe/xf = 1)
Initial Permeability
Pressu~ s Duration Thickness Skin
pi, psi Hours Type of Teat kh, md ft Factor
BOUNDING SURFACES
OF DRAINAGE VOLUME
FKACTUkE
I I I 1 1 # I I 1 I I I I 1 I I I 11
J!. I ..
FLOW TV4E,t,hours
/4
1
#t i
a td10 1 1 I 1 I 1 Itl
lo?
1 t 1 1 I I t 1
4?
Fm RATE,q,hiCF/O
x./xf plsloooo
5,@ ,/
&.0.5md */
A
Kfl- , 1 1 1 1 1 1 1 I t 1 I t 1 1 I t
d d d
FL(W RATE,q,MCF/C)
= 105 1 t 1 1 I 1#11 t 1 1 1 I 1, k
K! 10 102 lo~
L
FA(131.
12- Fig, 4 - Back pressurecurve: Isochronaltest (xe/xf 5)
10-
e-
4 -
X@/xf#s
I
o~m
FLOW TIME,t,
how
Fig. 5 - Compari
son of values of C predictedby Eq. 13
and isochronaltests
~RMMMANCE COEFFICIENT,
C X K?, MCFID@ %p
w a m a 5
0 I 1 , 1 I : I
I I r I I I I 1 1
.
a
o
~n
I I n I r I rI 1 1 I I 1 Ill-
I
0:
FLOW TIME,t, hOUrO 0
41 t # 1 , 1 1 1 t 1
03 ,/ 0.s t
x 12
RECIPROCALSQUAREROOT FLOWTIME,
-- sT~lL@W CAM /(
1~, hours-2
~/x, 1, p, ro61.7pd
/
kml~
/
/0
3
Ii!1($ 1 , 1 * 1 , I I 1 I , t m, L
to 102 103
1
X 24
la - STABILIZED CASE /
/
x, /xf 5, pi=iomo p8i 4
fj
k o,s md / / 1
I
.1
FLC?4/ RATE, q, MCF/D
FLOW TiME,t,houra
F Al
03 J
0 .24
-s, pi~sooopti
=imd
i&Lw1 ~1 . , ,
I io i&
FLOW RATE, q, MCF/D
I 07~
106~ A
1
FLOW TIME, t, hours
al
) 03
0 24
Xe/Xf * 1, pi 4000 psi
k= Imd
1
I 05
t 1 t I I I 1 1 1 1 t I I 1 1 1 I
,.2 103
10
FLOW RATE, q, MCF/D
- 14
- [u
3 -
-8
N .-
UI
a
2 -.6 ft
Q
x
1-
1
I
I I I 1 1 1
00 0.1 0.2 0.3 0.4 of
f ~LoG(tn-tj-l)
j-l n