You are on page 1of 20

.

Society of Petroleum Engineers


6200 North Centra/ Ekpwy.
Dallas, Texas 75206

;;z;ER

SPE6437

Wel1 TestAnalysis
of Hydraul
ical
1y Fractured
GasWel1s
by
Walter L. Dowdle,Member SPE-AIME,IntercompResourceDevelopmentand Engineering,Inc.
and Paul V. Hyde, Member SPE-AIME,ColumbiaGas TransmissionCorp.

THIS PAPER1SSUBJECT TO CORRECTION


@ Co~yrlght1977
American Institute of Mining, Metallurgical, and Petroleum Engineers, inc.
This paperwasprepared forthe1977 Deep Dr/llingandProduction Symposium of the SocietyofPetroleum
Enginee#ofAIME, heldinAmarlllo,Texas, Apri117-19, 1977.Permissiontocopyisrestrlcted toanabstractofnot
mofi:than300 words. Illustrationsmay notbecop/ed,T5e abstractshouldcontaln consplcuousacimowledgementofwhereand bywhomthepaper lspresented.Publication elsewhereafterpublication intheJO(LQNALOF
PETROLEWTECHNOLOG Yor:he SOC/ETYOFPETROLEUMENGINEERSJOURNALis usuaily?rantedupon
request to the Editor of the appropriate journal, provided agreement to give proper credit Is made Xscussion of
this paper is invited.

ABSTRACT

INTRODUCTION

Hydraulicfracturinghas been used extensivelyover the pact fifteenyears to stimulate


low permeabilityoil and gas wells. A considerable amount of fracturedwell performance
theoryhas accumulatedduring this period.
Transientdrawdownsolut%onsfor vertically
fracturedliquidwells based on numericalsimulationwere publishedin 1964. These solutions
establishedthe influence,
of verticalfractures
on traneientpressurebuildup.anddrawdown
testing. Othershave investigatedwell tests
of verticallyfracturedgas wells using both
analyticaland numericalmodels.Recent studies
have providednew informationfQr type-~urye
matchingof pressuredata obtainedfrom fractured (verticaland horizontal)wells. The
objectiveof thie paper is to illustratethe
applicationof numericalsimulationin evaluation of fracturestimulationof gas wells.
The previouslypublishedinterpretation
methods,
such as pressurebuildupand drawdownanalyses
and type-curvematching,form an extremelyimportantpart of the completeanalysis, Better
and more comprehensivewell test interpretation
can often be obtainedby using the so-called
conventionalmethodsend numericalmodeling
together.

prats, et.al.1, originallydevelopedens-,


lyticalaolu~o~ for the performanceof
verticallyfracturedreservoirsfor the compressiblefluid case. They consideredboth
the constantterminalpressureand constant
terminalrate cases. In the case of constant
rate, however,the early-timepressuretransient
solutionswere not fnvesti.gated.In 196+,
Russelland Truitt2publishedtransientpressure solutionsfor vertically-fractured
oil or
water wells based on numericalsimulation. From
their scluttonsthey developedmethods cifanalyzingpressurebuildup and drawdowntestswith
conventionalplottingtechniques. Clark3 later
appliedthe Russell-Truittresultsin analysis
of water-injection
well falloffdate. Analytical
solutionsand exampleapplicationsfor vertically fzncturedwells which produceslightly
compressiblefluidsalso were presentedby
van Everdingenand Meyer4.
.,

Referencesand Illustrationsat end of paper

More racen.tly
Gringarten,--
et.al.s,have
rev~ewedthe work of previousauthorsand publishednew solutionsespeciallyuseful for
type-curveanalye$s. They illustratedthe use
of their results (forwells with eithervertical
or horizontalfractures)in a companionpaper6.
However,their solutions,like thoee of Russell

10

WET.?.TEST
--------

ANAT.V!XT!SfW
.-..
-.-

v-

UVIM?ATIT.TCA1.
. ..-.--.-.

T.V

17RAPT?lQUn
+

k-w

-W*WU

@Ae

WA.

m??

T @

Gnu

WXIAJUU

0s

cf.~7
u-

Ja

and Truitt and van Everdingen-Meyer,


were.based
of the proper semilogstraighton a Horner graph.
on flow of a slightlycompressiblefluid such as Moreover,the entirepressurehistoryof swell
oil or water. It is clear that these authors
test now can be analyzedfrom type-cume matching
were aware that their 8oluticmscould be extended of field data and fracturehalf-lengthand
to the f?.owof real gases. In fact,Raghavan7
reservoirpermeability-thickness
can be deterrecentlyhas impliedwith practfcal
examplesthat mined.
it could be done. Hadinoto,et.al.~,have shown
explicitlythe relationshipb=w=n constant
Our main purposeis to demonstratethe improperlyllqu%dflow solutionsand constantrate portanceof numericalsimulationin the interdrawdownsolutionsin determinationof vertipretationof fraccuredgas well tests.
tally-fractured
gas well deliverability.
Conventionalmethods,rather than being replaced
by numericalmodeling,are extensivelyused in
The first practicalstudy of well test +nthe analysisprocedure. We will illustrate
alysisin verticallyfracturedgas wells was
our approachwith an examplewell test. First
publishedby Millheimand Cichowicz9. Using
we providea brief reviewof the real gas
previouslyderivedanalyticalmodels, they prepseudo-pressureand a discussionof the numersented the drawdownequationsfor ideal gas flow, ical model.
includingthe effectsof format%on<amageand,
turbulence,for both the linearand radial ilow
THE REAL GAS PSEUDO-PRESSURE
periods. The linear flow model was based on the AND DIMENSIONLESSSOLUTIONS
classicwork of van Everdingenand Hurst10while
radial flow was based on the model of Swift and
In 1966 A1-Hussainy,Ramey and Crawford13
Kielll. About a year later in 1969 Wattenbarger preiiented
the applicationof the real gas
and Ramey12 presenteda study of well test.inter- pseudo-pressure
m(p) to the flow of real gases
pretatlonin verticallyfracturedgas wells using throughporousmedia, One of their most in%
a numericalmodel which includedwellborestorage portantcontributions,like that of Aronofeky
and turbulence. Useful plots showingthe effects and JerJkins14
for flow of ideal gases,was that
of fracturelength,wellborestor&geand turbuthe m(p) functioncouldbe correlatedwith
lence on pressurebehaviorof a vertically
the liquid flow solutionsof van Evergingen
fracturedgas well were presented. One conand Hurst. They showed that
elusionamong others they reachedwas that the
drawdowntestingmethodspreviouslypublished
1.987x10-5khTsc
for verticallyfracturedwells couldbe extended
m(pi)-m(pw)]-s
PD(tD) =
qpscT
[
to real gas flow if the real gas pseudo-pressure
were used. In that paper,however;Wattenbarger
and Ramey gave no direct comparisonwith the
-Dq. . . , . . . . . . . . . . . .(1)
Russell-Truittresultsand the Gringarten,et.al.,
and van Everdingen-Meyer
solutionswere not~e~
pv%lished.
to an excellentapproximation.ne pD(tD)
here is the,vanEverdingen-Hurs-:
dimensionless
All of this previouswork has increased
pressuredrop functionand m(p) is definedas
significantlyour understandingof transient
fluid flow in hydraulicallyfracturedwe~ls.
p&Q$
(2)
The Russell-Truittstudiesshowed that convenm(p) = 2 ! pz
tionalpressurebuildup analysieof data from
Pb
thesewells would yield, under certaincircumstances,,
fairlygood estimatesof reservoir
For large ratiosof re/rw
permeability-thickness
and averagepressure.
They pointedout that the effect of a vertical
{3)
fracture(in the absenceof,other effectssuch
pD(tD)= l/2(fintD +@.80907) . . .
as formationdamageby fracturingfluids)is to
r 2
cause calculationof a negativepseudo-skin
factorfrom pressurebaildupanalyais. They also for 100 ~ tD S 1/4$
and
()W
provideda means of estimatingfracturehalflength. MillheimCichowiczdemonstratedthat
r 2
re
reservoireffectivepermeability,turbulence
pD(tD)=2n~-3/4+2~
~
**(4)
coefficient,effectivefractureflow area and
W
() e
fractureefficiencycould be obtainedfrom constant rate drawdowntestingof vertically
2
&
fracturedlow-permeability
gas wells. The
, where
for tD > 1/4 r
recentwork of Gringarten,et.al.,has
provided
()w

a sound bas%s for short-t%meanalysis. Using


0.000264kt
.(5)
*.*9**
type-curveprocedures,it is possibleto deter%=
mine the end of the linear flow period and start
$(uct)ir~

WAT
- m711
*M*,

Q1
.-

T
J4

nnr.mt
u..-mw~u u

A1-Hussainy,et.al., also justifiedthe


use of superpositi~~ the real gas paeudopreseureresultsas an acceptableengineering
approximationof variablerate flow of real
gases %n a radialsyacem. To valop the
f!, considera
generalsuperpositionequation
well which flows at n differentrates as shown
in Fig, 1. ITkwell fnitiallyproducesat rate
ql for time tl. The rate is then changedand
rate q2 ie maintaineduntil time t2 and so on
until rate~to
time tn. There is no restrictionon productionrate: the well may be
producing,injectingor shut-in. Applying
the superpositionprincipleto Eq. 1, letting
s and D equal zero for simplicity,yieldsafter
rearranging
Tp8c
m(pw) = m(pi) -

! %J
10987x10-5khT~=
j=l

pD(t:-~j-l)

-pD(t; -t#

nAm

. . (6)

qTpsc

S*.*...

SPE 6437

mvnw
k6Lwu

---

.-

fracturedgae wells will be consideredin this


study since tha vast majorityof hydraulic
fracturesare vertfcalones.
THE NUMERICALMODEL
Coats,Dempsey,Ancell and Gibbs19originally describeda two-dimensional
formulation
of the mathematicalmodel used in this study.
We wI1l give a brief outlinehere of the model
adaptedto three dimensions. The effectsof
turbulence,pressure-dependent
permeabilityand
porosity,damage,wellborestorage,crossflow
and partialpenetrationare accountedfor rigorously. These relationshipsare coupledwith
a wellheadpressurecalculationwhich permits
simulationof observedtubingheadpressures.
AppendixA gives a detaileddescription.

The continuityequationin cylindrical


coordinatesis

+~bg).

...(8)

.. o........

(7)

The tem~t~ - 1$) is usqallyexpressedas


AtD and tD as (t+At)D.

.-kr~p
pg~r

. (9)

and similarlyfor v and Vz. In Eq. 8 bg Is


the reciprocal
gas tormationvolume factor i=
Mcf/cu.fi.
r

where

pl)()
1.987x10-5khT8c [b

-Pn(t:-t;)

w
v

4-cbvu

Brief Description

where superscriptshave been used on tD for


convenienceIn notation. If the well is produced at a eonstentrate q for time tl and
shut-inuntil tiae tn, Eq. 6 reduces to the
familiarexpression for pressurebuildup:

m(pw) = m(pi) -

Awn

-Vu

PT8C
bg =

The importanceof the A1-Hussainyset.al.,


work in justifyingthe linearizationof ~a~
gas flow based on m(p) goes beyond the use of
the van Everdingen-Hurst
solutionsonly.
Given the appropriatedimensionlesssolutions
representedby pD(tD)valuee,Eq. 6 can be
used to generatepreesureresponeefor any
eystem,s.~, a well at differentlocations
within variousdrainagepatternsand with
constantpressureor cloeedouterboundaries.
Ramey and Cobb16 clearlyexplain the use of
pD(td) in such a manner in their paper on
generalpressurebuildup theory. Ramey,Kumsr
and Gulati17and recentlyStrobel,Gulati and
Ramey18have illustratedthe applicationAn
field studies, The use of pD(tD) solutionsin
this fashionalso can be extendedto fractured
gas wells. We mentionedpreviouslythat
Hadinoto,et.al,have done so in determination
of the del~e~b$lity of verticallyfractured
gas wells. Althoughapplicationcan be made
to horizontalfracturesalso, only vertically

1000ZPSCT

.(10)

Darcys law modifiedto include turbulenceis


for radial flow

-$r .5k r Vr+lllpv;


3

. .(11)

whera for a real gas the equationof state is


p~~T

, . .

.* . . . .

, .(12)

-1
and @ is the turbulencefactorin ft .
For gae welle, Ramey20 showed the effect
of wellborestorageon the measuredwellhead
rate is

92

WELL TEST ANALYSISOF HYDRAULICALLYFRACTUREDGAS WELLS

SPE 6437

which representsthe fractureand rema~ning


eector spacingalso is given in .Appendlx
B.

VwTs=
%.
sf %h + 1000 pscTw @Z

.. (13)
Comparisonof Ntjmerical
and Approximate
&alytiCal Solutions

Wellheadpressurefrom the Cullenderrnith21


relationshipis

We foundmodel solutionsat relativelylow


productionratee for a well which has not been
fracturedto agree closelywith those computed
fromEq. 6 includingboth skin damageand
2-s
turbulence. At hi~her rates,however,the
e
bh
analyticalresults.donot
match the numerical
40,000d5
s
[
solutionsquite as well. This behavioris
consistentwith the conclusion of A1-Hussainy,
Iere
et.al.
They illustratedthat real gas cases

lead to dimensionlesspressuredrops which are


s * o.(-)375@T6za . , . . (15) lower than the liquid case and which are flow
rate dependezt. Ratz and Coats22also have
pointed out the limitationsof applyingthe
Eqs. 8-15 are combinedusing the real gas
van Everdingen-Hurst
solutionsfor the liquid
Seudo-prsssure
of Eq. 2 into one equationwhich case to real gas flo~ Nevertheless,we f?mnd ~
3 expressedin implicitfinitedifferencefozm. acceptableagreementof numericaland analytical
Ie resultingset of simultaneousequationsfor
resultsboth before end after outer boundary
m1 grid block and the wellboreare solvedby
effects for the non-fracturedcases. Furtherh~ct solution.
more, estimatesof reservoirpermeabilitythickness,skin factorand turbulencederived
from conventionalanalyeisof the numerical
fidSystem
resultswere re,~arkably
close to the actualvalues even at the higher rates. Before comparing
A generalizedschematicof the model grid
analyticaland numericalsolutionsfor the
rstemused to simulatevert~callyfractured
fracturedwell case some rev~.ew of the analytical
>11sis shown In Fis. 2. The fracture
solutionsis
worthwhile. We will confineour
}hadedarea) is modeled as a narrow cylindiscussionmainly
to the resultspresentedin
:icalsegmentwhich extendsto a radial
Ref. 5.
LstanceXf and has a height hf, Reservoir
For verticallyfracturedliquidwells,
id fracturepropertiesare aesignedto each
Gringarten,
et.al.,presentedsolutionsfor

:id.
block. PVT relationshipssndwell data
both
the
uniformflux
and infiniteconductivity
:einput along with any variationof permefracturecases.
We
have
reproducedtheir results
bllityand porositywith pressure. If
In the uniform flux case
as
Figs.
4
and
5.
rmmetryis assumed,only one reservoirquadmt as shown in Fig. 3 need be simulated(the (Fig.4) they stated that pressurewill vary
along the fracturelength except at early times.
~tal roductlonfrom one symmetricalelement
They
also noted that the pressuredrop is low
) q/45
and that come field data appear to match th%s
solutionbetter than the infiniteconductivity
In cylindricalcoordinates,fracture
solution.
Ldthis proportionalto radialdistancefrom
tewellbore. To m~del a constant-widthfracFor the infinite-conductivity
fracture
lre,it is necessaryto modify transmissibility
(Fig.5) pressureis uniform over the total
idpore volumes to accountfor this changein
fracturelength. To generatesolutionsfor
:Clength. The resultingexpressions,as dethis cbse Gringarten,
et.al.devaluatedthe
;Iopedin AppendixB are
uniformflux solutionat a dimensionlessdistance ~ (basedon half fracturelength)of
180wf
T;-:Tr~
. . . . . . . . . . .; (16) 0.732 for infiniteand closedrectangular
reservoirs. They justified,~yosteriori, the
choiceof this point for finiteresrev=s
by
desuperposition
of the;Russell-Truitt
results.
Indiscuseionof their solutions,Gringarten,
et.al.,comparedtheRussell-Truittresultswith

their inflnl~econductivitysolutionsin both


tabularad graphicalform(see Table 4 and
IereT; andpv~are the radial transmiesibtl%ty Figs. 11 and 12 of Ref. 5). Agreemenfi
was good
idpore volumemodifiedfor a constantwidth
except for the xe/xf = 10 case. Accordingly
they recommendedthat this Russell-Trultt
:acture. Determinationof the sector angle

pwh2
=

~h2GTazafx
es-~.(M)

93

SPE 6437

WALTER
_.._AND PAUL V. HYDE
...-.. L;
. DOWDLE

mlutionnot be used further. We have reviewed


:hesepointshere becauseour resultsoffer an
hterestingcomparisonwith the Gringarten,
!t.al.,snd
Russell-Truittsolutions.
-

that is

To providea comparisonof analyticaland


wmerical solutionsfor a verticallyfractured
~aswell, we considerthe drawdownexampleof
hl-Hussainy
and Rsmey23. Pertinentreservoir
Iataand gas propertiesare given in Table 1.
h this instancethe well has an infiniteconductivity
verticalfracturewhich extends
wer the entire formationand has a half:racturelengthx of 22.21 feet. There is Iio
lkindamageor tur
% ulenceand wellboreetoragk
isnegligible. The well %s producedat a con\tantrate of 1600 Mcf/D for a period of 41.5
lours.

The fracturehalf-lengthiscomputed
from the time match (tD = 48.9, t = 10 hrs)
in a similarmanner

kh = 47.0 md-ft

2= 0.000264kt
f
o(vct)~t~

0)

2
(0.000264)(47.0)
(10)
(0.000404)
(0.77)(10)(48.9)
f = (0.1)(0.0176)
or,

As we have mentioned,symmetrypermits
me reservoirquedrentas illustratedin Fig. 3
:0be modeled. Block centerradii,block
~oundariesand
sectorangles used.in the sibilationare listed In Table 2. Permeability
:n the fracturewas assignedso that at any
}ointin time pressurealong the fracturewas
miform (infinite-conductivity
fracture).
lodelresultsare sumarized in Table 3.

= 21.5 ft

Althoughthese resultsprobablyare within


graphicaltolerance,we previouslypointedout
that the data at early times fall off the typecurve. This appearsto result from a longer
period of linear flow than the infiniteconductivity solut%onof Fig. 5 would predict.
Since flow geometryis essential linearat
early times,a plot of Ad(p) vs ft should give
A type-curve match of the model results
a straight-lineand establishthe durationof
~iththe infinite-conductivity
solutionof
this period9*25. Fig. 7 shows that the linear
tef.5 is shown in Fig. 6. Since for a verflow period lasts for about 0.051 hours (tD =
:icallyfracturedwell dimensionlesstime is
O.247). Such behavtoris more characteristic
lefinedas
of the Gringarten,et.al.,uniformflux solutions.
Matchingthe data o~T=le 3with the Fig. 4
0.000264kt
(18) type-curve,this time not fixingeitheraxis
**..*
*...
%=
but adjustinghorizontallyand vertically,gives
o(llct)ix:
the match shown by Fig. 8. The agreementwith
?e obtainedthe match point by fixing the time
the xe/xf = ~ curve appearsto be good at all
ixis so that tD = 4.888t and moving in a vertimes. From the pressurematch flow capacity
ie
calculatedas 50.4 md-ft. The time match
:icaldirectiononly until the best agreement
yields a fracturehalf-lengthof 24.2 feet.
rasobtained. As Ftg. 6 illustrates,data
aftera producingtime of about 0.8 hours
Actually,model-computedpressureslie
:loselymatch the xc/x$ = w curve. Pointsbefore
between the two analyticalsolutionsas Table 4
this time, except for the f%rst, fall noticeably shows.
The appropriatepD(tD) valueswere taken
above the tnf%nite-rat%ocurve.
from Ref. 5 and used in Eq. 6 to generatethe
analyticalresults. One possibleexplanation
Followingthe procedurboutlinedin Refs.
for differencesbetween the numericaland
j, 7, and 24, permeabilityIs computedfrom
analyticalsolutionsmay be a result of the
thepressurematch (PD = 0.35, Am(p) = 107
approximatenature of the analyticalsolution
?si2/cp)as follows
Ikeelf. This is well documentedinRef. 13.
We also are aware of inaccuraciesin numerical
OPscP~
U* s
(19) solutionswhich can result from time-stepsine
**,..
and ~%id-blockdefinition.Optimizationstudies
1.987x10-5TscAm(p
)
were performedbefore runningthis exampleto
eliminateany such differencesbetiieen
numerical
solutions. Itmaybe then, and Fig. 7 would so
kh= (3.600)(590)(14.7)(0.35)
illustrate,that the linear flow period for the
(1.987x1G-5)
(520)(107)
xe/xf =P case lasts longer than predictedby

..

UEl}.
AUAT.VQTf!
.-- TRST
---------

94

13U
.

tlVllQAUT.TfiAT.
. ...-.
..

T.V

??RA@FflQUli
a- -*
.-M

@AC
URT
Ca
-w-T

QDQ

W&a

Kh97
U-.?t

the Gringarten;
etial., solugions. A comparison and
of Tables 1 and ~o~Ref. 5 shows,quite interestingly~thatthe Russell-Truittxe/xf = 10
liquid flow solutionalso yields a longer linear
flow period than that of Gringarten,et.al.
Under certainconditionsoutliuedin Ref. 2,
This is true of the xe/xf = 5 curva a~w~l.
we can computefracturehalf-lengthfrom
We couldnot obtaina good match of the
Russell-Truitt
xe/xf =-10 curve using the data
-s
of Table 3, however. This supportsthe con=2rwe
. . . . . . . . . . . . . .(23)
f
clusionthat this Russell-Truittcase may be
erroneous.But it alsoseems that the
Thus,
Gringarten, et.al.,infinite
conductivity
solutionis n~~ecise
in the transitionregion
= (2)(0.5)e2*99
betweenlinearand pseudo-radialflow.
f
The drawdowndata of this exaziple
also can
be analyzedin the conventionalmanner. Fig. 9
is a semilogplot of m(pw) vs flow.timefor the
data in Table 3. The straight-lineshown on
this graph has a slope of 31.2 x 106 psi2/cp.
Flow capacityis computedas

= 20.0 ft

an extremelygood estimate.
VERTICAL-FRACTURE
GAS-WELLTEST EXAMPLE

Table 5 gives pertinentdata for the gas


well test examplewe considerhere. The test
k.h=
(21) consisteof three separateten-dayflow perfods
Tscm


each at a differentrate and beginningafter
reservoirpressurehas stabilized. Type-curve
analysisand the Millheim-Cichowicz
methodwill
be used to estimatereservoirpermeability~=
(57,927)(1600)(14.7)(590)
thickness,fracturehalf-lengthand turbulence
(520)(31.2X106)
factor. Fol)!owing
this teet,we will analyze
a three-cycledrawdown-buildup
test on thissame
well with computersimulation.
giving
u
Type-CurveAnalysis
kh = 49.5 md-ft
57,927~scT

A match of the second flow period data,


q = 765 Mcf/D, is shown in Fig. 10. Note that
herewe use Ap2 rather than the rei,lgas pseudopressurem(p). The type-curveused is the
uniformflux solutionof Ref. 5, our Fig. 4.
At time~before about 0.05 hours, the log-log
dpi) - dplhr)
slope on Fig. 10 ie unity, indicatingthat
6 = 1.151.
pressureresponseis
controlledby wellbore
m
[
storage. Followingthis period a transition
regionoccurs. The half-slopewhich is character~sticof the linear fl~w regime appeareto
-lo8(;,v;)+r:)+
3.23]. . . (22) occur around 0.8 hours at which time pointsbegin
to fall on the xe/xf= ~ curve.Data after about
twentyhours are linearon a conventionalsemilog
Obtainingm(plhr)from Fig. 9 and substituting plot
.
gives
From the pressurematch (pD ~ 0.2,
Ap2 = 105 psiz),
(434.65- 380.75)X106
s = 1.15
quzp8cTpl)
[
31.2x106
,~ m
. (24)
1.987x10-5TscAp . . 9
!%e pseudo-skinfactor can be used to approximate fracturehalf-length. From Ref. 23,
since there is na turbulence,

49.5
log
((0.1)(0.0176)
)
(0.77)(0.000404)
(0.5)2(10)

+ 3.23

kh= <765)(&014)(0.845)(14.65)(570)
(0.2)
(1.987x10-5)
(520)(105)

.,

95

WAT.TRR Y.. lMMJllT.R


ANl) PAITT.V. HVTIE

SPR IM47

885 log A= -360

~Aving
1
and
kh =14.63 md-ft

A= 0.392
From the time match (tD= 0.091, t = 1 hr) and
Eq. 20, fracturehalf-lengthis

We can now computethe effectivewellbore


radius from Eq. 26 as

(0.000264)(14.63)
(1)

2=

(0.0405
)(0.014)(8.0x10-4)(0.091)(63)
t

&
<5.93X10-4)(
12.43)
r* =
w ((0.0405
)(0.014)(8x10-4)(0.392)(63)
)
That is

f = 38.5 ft
As shown in Table 5 fractureheight extendsover the perforated-interval
only. Since
the above reqult Is based on penetrationof the
totalreservoirheight,actual fracturehalfLengthshouldbe approximately81 feet.

r: = 25.6 ft
Since xf = 2r~ end fracturepenetration
extendsover 30 feet of the total zone, estimatedfracturehalf-lengthis 108 feet.

Althoughnot shown here, we also analyzed


The non-Darcyflow constantD can be dethe test data with the infiniteconductivity
terminedfrom the slope of the straight-lineon
Bolutionof Fig. 5. The match did not appear
to be quite as EO06. A slightlylower formation F&g. 12 as follows
permeabilityand fracturehalf-lengthwere obtained. A matgh w$t the recent type cl~rvee
8
0.87 mD = 0.285
of Reghavan,et.al$,was
not atteaqked.

Millheim-Cichowicz
Method

where m is the semilogslope from Fig. 11. Thus,

Fig. 11 is a semilogplot of Apz/q vs


flow time t. Our analysisfollowsalong the
lines recommendedby Millheimand Cichowicz.
From the slope of.the straightlines on this
graph, reservoirflow capacityis computed
ae
kh=

57,927pZTpsc~
T ~
. . .
ec

)
(
(0.%%85)

D=

3.70 X 10-4(Mcf/D)-1

The turbulencecoefficknt B is

* . (25)
v TschrwD
$=

@=
Thus
kb,= 12.43 md-ft
To determineeffact%vewellboreradius,we
plot Ap21q vs q at t = 1 hour from Fig. 11 as
shownby Fig. 12. Based on time in hours, the
apparentwellboreradius is

where the interceptat q = O on Fig. 12 Is


m(logA). Therefore,

..00..

(27)

2e715X10-k5PScMk

kh= (57,927)
(0.014)(0.845)(570)(14.65>
(520)(885)

-4
(0.014)(520)(63)
2(25.6)(3.70x10
(2.715x10-15)
(14.65)(29)(O.6)(12.43)

g%ving
13= 3.181x1013ft-l
The resultsof the type-curveanalysis
and Millhei-Cichowiczmethod are in fair
agreement. %heprocedurerecommsndedby
A1-Hussainyand Remey could also be used to
interpretthese test data. Our main objective
here was to obtain initialestimatesof kh, xf
and 8 for computersimulation.
;

.-

---

*M*

-76-.

..

M*.-

.-d*

.-

.-..

*W

.*

a-

. Aa

W-?an

ComputerSiunilation-

fracturelength,width and flow capacity. By


so doing, the analystcan eee beforehandthe type
Drawdownand buildupdata for the threeof data he will be reeponalbleto evaluate.
cycle test are given in Table 6. We simulated Thicishouldpermit abetter test designand inthis test in a manner similar to that preterpretationand may eliminateobtaininguseless
viouslyoutlinedin compaiinganalyticaland
data. Seeond,better and more comprehensive
numericalsolutions. In this case two vertical resultsat greaterconfidencelevels often can
layerswere used, one of 30 feet which conbe obtainedby combiningtype-curveanalysisand
tainedthe fractureand the other of 33 feet.
conventionalmethodswith numericalmodeling.
Without initialestimatesof reservoirand
Initialestimatesof reservoirpermeabilityfractureproperties,computersimulationof
thickness,fracturehalf-lengthand turbulence fracturedwell teat data wouldbe complicated.
coefficientwere based on resultsfrom the
Althoughan acceptablematch might be obtained,
tiillheim-Cichowicz
analysisabove. Using these the questionof uniquenesscertainlywould or
valuesand the rate data of Table 6, pressure
should%e raised.The analystshouldnot be
at each time incrementwas compted. A comcontentwith using one kind of plottingtechnique
parisonof model-calculated
pressuresand the
or procedureor only with the une of numerical
pressuresin Table 6 is shown in Fig. 13. Cal- ~dele .
culatedpressuresduring the latterstagesof
each drawdownare too high. The same is true
NOMtNXATURE
C& the builduppressures. Adjustmentof the
initialvalues of kh, Xf and @ gives the match
and reservoir
5.93x10-4k , diWn
Bhownby Fig. 14. Fiiacture
-A=
constant,
propertiesused in thiu simulationare sum$pctr~
marizedin Table 7.
.
elon; or drainagearea, acres
A comparisonof post-fractureand nofractureperformanceat a constantwellhead
reciprocalformat~onvolume
b=
pressureof 750 psig is shown in Fig. 14. The
factor,Mcf/D cu ft or STB/RB
no-fractureperformancewas computedwith the
-1
radialmodel without ths fracture. During
= compressibility,
psi
c
the firstyear, fracturingramlts in more
than a sevenfoldincreasein cumulativegas.
= insidepipe diameter,in
d
production.
D
= non-Dare flow constant
!
(Mcf/D)CONCLUDINGREMARKS
The precedingexampleshows that permeability-thickness
and fracturehalf-length
estimatedby the MillheimCichowiczmethod are
in errorby over 22 percentas comparedwith
the simulationresults. perc~t error in
turbulencefactor is about 44 percent. Reeultx
from the type-curveanafiysis
give even greater
errors in estimationof kh and Xf. It shouldbe
recalledthat these determinations
were based
on ten-daydrawdowntests. Had only the
three-cycletest data been available,interpretat,ion
by typs-curveanalysisor by the
Millheim-Cichowicz.method
would have been
difficultindeed. It is necessaryonly to
considerFigs. 10 and 11 to realiz,e
this is
true. Without drawdowndata beyond 10 hours, .M
like that of the three-cyclatest,well
forecastsbased on such analysesalone would
be precarious.
We believe this emphasizesat least two
points. First, design of pressuretests,on
vertically-fractured
gas wells i= critical.
Before tests areconducted,estimatedpreseure
transientresponseehouldbe calculatedusing
known or approximatereservoirdata and planned
..

= base of natural logarithm,


2.71828

frictionfactor

gas gravity (air=l)

= net formattonthickness,ft

= permeability,md

!Ln

= naturallogarithm,base e

S1O e of semilogstraight-line
8! /cp/cycleor p8i2/Mcf/D/cycle
psi
....
= nmlecularweight

m(p)

real gas pseudo-pressure,


psi2/cp
!
Am(p) = m(pl)-m(pwf),psi2/cp

pD(tD)=

fluid pressure,psi

dimensionlesswellborepressure
drop

... ----

-.

-.

--.. -

Pi

= Initialpressurein the system,


psi

Ap2

= P12 - pwf2~ or PW82 - Pvf2* Psi

Pv;

= modifiedradialpore volume for a


constantwidth fracture, Cu.ft.

= surfaceflow rate, Mcf/D

r.

re

= externalradius,ft

rw

= :radiusof well, ft

r*
w

= apparentor effectivewellbore
radiuseft

R.

s=
s

----

-.-

- . ----

-1
turbulencefacto?,ft

= derivativeoperator

= angulardistance,degrees

= density,lbmlcu ft

= fluid Viscosity,

total porosity,fractionbulk
volume

radial distance,ft
Cp

Subscriptsand Superscripts
a

= average

bh

= bottomhole

dimensionless

fracture

gas

= initial

= jth time interval

= base

= nth time level

= oil

gas constant
skin factor,dimensionless

= constantin the Cullender-3niGh


equation,Eq. 15
8

----

= gas saturation,fzdctionpore
volume

time, houre

%)

dimensionlesstime based on wellbore radius rw or on a vertical


fracturehalf-lengthXf

At

shut-intime,hours

= radial coordinate

temperature,R

Sc

= referencecondition

Tr

radial transmissibility,
Mcf/D

Sf

= sand face

T;

= modifiedradial transmissibility
for a constantwidth fracture,
Mcf/D

= total

= well

= velocity,ftfsec

Wf

= well flowing

Vw

= wellborevolume,kft3
-.

wh

= wellhead

= verticalcoordinate

= angularcoordinate

lhr

= one hour

f
x
e

x
z=

fracturewidth, inches

= half-lengthof side of square


enclosinga vertically-fractured
wall, ft
= verticalfracturelengtk from
centerof well to tip of
fracture,ft
=

depth of well, ft
real gas law deviationfactor,
dimensionless(pv-ZnRT)

.-.

We are gratefulto INTERCOMPResource


Developmentand Engineering,Inc. and to
ColumbiaGas TransmissionCorporationfor
permissionto publ+shthis paper. We especiallythankK. L. Ancell,J. S. Nolen and
A. L. Spivak for their valuablesuggestions
and stimulatingdiscussion.
REFERENCES
1. Prats,M., Hazebroek,P. end Strickler,
of Vert%calFractures
W. R.: qEffect
on ReservoirBehavior- Compressible
Fluid Caser,Sot. Pet. ~. ~.
(June,1962).

2. Russell,D. G., and Truitt@N. E.;


TransientPressureBehaviorin
VerticallyFracturedReservoirst,
~.
Pet. Tech. (Oct.,1964), 1159-1170.
3. Clark,K. K.: TransientPressureTesting
of FracturedWater InjectionWells,~.
Pet. -Terh. (June,1968),639-643.
4. van Everdingen,A. F., end Meyer, L. J.:
Analyslsof BuildupCurves Obtained
After Well Treatment,~. Pet. Tech.
(April,1971) 513-524,Trm~, =,
~.

5. Gringarten,A. C., Remey,H. J., Jr., and


Raghavan,R.: Unsteady-State
Preesure
DistributionsCreatedby a Well with a
Single Infinite-Conductivity
Vertical
Fracture,Sot. Pet. ~. ~. (Aug.1974),
347-360;Tr=.,~=,
257.
6. Gringarten,A. C., Ramey,H. J., Jr., and
Reghavan,R.: AppliedPressureAnalysis
for FracturedWells,~. Pet. Tech. (July,
1975), 887-892;Trene.,Am, ~
7. Reghavsn,R.: Some PracticalConsiderations in the Analysf.s
of PressureData,
~,
Pet. Tech. (Oct.,2976),1256-1268.

10. van Everdingen,A. F. and Hurst,W.: The


Application
of the LaPlaceTransformation
.to Flow Problemsin Reservoirs,Trans.,
11. Swift, G. W. end Kiel, O, G.: The
Predictionof Gas-WellPerformance
.,
Includingthe Effect of Non-DarcyFlow,
~. Pet. Tech. (July,1962) 791-798,
Tram, =,
&&.
12. Wattenbarger,R. A. and Ramey,H. J., Jr.:
Well Test Interpretationof Vertically
FracturedGas Welle,~. Pet, ~.
(my,
1969) 625-632;Trane.,A~~246.
13. A1-HussainyaR., Remey,H. J., Jr. end
Crawford,P. B.: The Flow of Real Gases
ThroughPorousMedia,J. .
?et. Tech. (May
1966),624.
14. Aronofsky,J. S. endJenkins,R.: VA
SimplifiedAnalysisof UnsteadyRadial
Gas-Flow,Trans.,AIME,~,
i49.
15.

C. S..
end Russell.D. G.:
Pressure-Buildu6
end Flow Te&s in Wells,
Sec. Pet. Eng. MonoSraphSeries,Vol. 1,
SPE, Dalles (1967),14.

Matthews.

16. Remey,Henry J., Jr., end Cobb, W. M.:


A GeneralPressureBuildupTheory for
a Well in a ClosedDrainageAraa,~.
Pet. Tech. (December,1971),
1493.

17? Remey,H. J., Jr., Kumar, A., and Gulati,


M
Gas Well Test AnalysisUnder WaterD~~ve Conditions,AmericanGas Association
Monograph,1973. ~
18, Strobel,C. J., Gulati,M. S., and
Ramey,H. J., Jr.: tReservoir
Limit Tests
in a NaturallyFracturedReservoir--A
Field Case Study Using Type Curves$t,
~.
Pet. Tech. (September,1966), 1097.
19

Coats,K. H., Dempsey,J. R., Ancell,


K. L., end Gibbs,D. E.: Anslysisend
Predictionof Gas Well Performance,
SPE 3474, preeentedat the 46th Annual
Fall Meeting of the SPE of AIME, New
Orleans,Lousiana,October3-6, 1971.

8. Hedinoto,N. and Reghaven,R.: Deter-.


minationof Gas Well Deliverabilityof
VerticallyFracturedWells, SPE 6136,
20. Ramey,H. J., Jr.: Non-DarcyFlow and
presentedat the 51st Annual Fall Meeting
WellboreStorageEffecte
in Pressure
of the SPE of AIME,,New Orleans,Louisiana
Buildupand Drawdownof Gas Wells,
October 3-6, 1976.
~.
Pet. Tech. (Feb.,1965), 223-233;
9. Millheim,K. K., end.Cichowicz,
L.:
Teetingand AnalyzingLow Permeability
21. Katz, D. L., Cornell,D., Kobayashi,R.,
FracturedGas Wells\~. Pet. Tech.
Poettmann,F. H., Vary, J. A., Elenbaas,
(Feb.,1968), 193-198.

J. R. and Weinaug,L. F.: Handbookof


NaturalGas Engineering,McGraw Hill
Book Co. (1959),309-315.

-a

SPE 6437

WALTER L. DOWDLE AND PAUL V.HYDE

,.

22. Katz,D. L., and Coats,K. H.: UtiderjgroundStoragedf FluidssUlrZcheBooks,


Inc., Ann Arbor,=~
(196$),153.
23. A1-Hussalny,R. and Ramey,H. J., Jr.:
Applicationof Real Gas Flow Theory
to Well Testingand Deliverability
Forecasting,Trans.,AIME~

where f(p) 1s permeabilityat pressurep


dividedby permaab%lityat initialpressure.
!!INSS
O(P) is slmplya constant,ll(zki)~
times the real gas pseudo-pressure.So that
porositymay be allowedto vary as a function
of pressureg(p) will be definedas @(p)/$i,
where $i is porosityat initialPressures With
these definitionsit followsEhat

24, Ramey,H. J., Jr.: Short-TimeWell


Test Data Interpretationin the Presence
of Skin Effect and WellboreStorage,
~. Pet. Tech. (Jan..,
1970),97-104;
Tra~, ~,
~.

and, similarly

25. Ramey,H. J., Jr.? PracticalUse of


Modern Well Test Analysis,SPE 5878
presentedat the 51st Annual Fall
Meeting of the SPE of AIME, New Orleans,
Louisiana,October3-6, 1976.
26. Raghavan,R., Uraiet,A, and Thomas,
G. W.: VerticalFractureHeight:
Effecton TransientFlow Behavior,
SPE 6016, presentedat the 51st Annual
Fall Meeting of the SPE of AIME, New
Orleans,Louisiana,October 3-6, 1976.
APPENDIKA
The followingthree-dimensional
formulation of the model used in this study is
adaptedfrom the earlierwork of Coats, et.al.
SubstitutingEq. 9 into Eq. 8, neglect@%~Darcy flow for the presentyields

ei

MIA=_
00

M
IJ8Z ae

ep

kzip+~

(A-5)

(A-6)

g
SubstitutingEqs. A-4, A-5 and A-6 into A-2 and
noting that $ig(p)= 0 leads to

+&(arkri*)+*kff%i%P

I +.3$9-

q . a$i ~

(A-7)

(z )..

Now using the chain rule of partialdifferentlon

(p) $)=*

(-)

A-8)

and lettingc* = d(g(p)p/Z)/dQ(p)s


a singlevalued functionof Q(p)> Eq. A-7 becomes finallY

Mark%+$))+++@ei*)

+=(akzl*b=aoic*
(A-)

Ueing the definitionof bg fromEq. 10, letting


a= T8C /1000p8cTgives

Referringto Fig. 2, an implicitfinitedifferenceapproxtmsttonto Eq. A-9 is

+(+$%)-q
++)

A-2)

- i,j,k

.*

%!....

. (A-1O)

Radial transmissibility
for flow between
To remain consistentwith the development .
blocks ri and ri+l, where ri are block center
of Coats,et.al.,the
real gas pseudo-pressure
definedb~E~2
will remodified slightly
radii, 3s definedas
and retermedQ(p) as follows
0.00633(2ma)A0
Azkkri~ , k
..(A-n)
T
360 kn(ri+l/ri)
ri+%,j,k=
.
(A-3)
Q(p)=lp~dp
. . . . . . . .
P~ 8

20

WELL TEST ANALYSI$or-HYDR LICALLYFfiCTUREDGAS WELLS

where Ae is the jth angularsector in degrees


and wher~ the effectiveinterlock permeability
must be
ri+%,j,k

SPE 6432

%.3acc~ulation or capacitycoefficient
(Vp c*)

i,j,k =

&

i
to correctlyrelate steadyflow re~e and pressura drop in the intervalri+l, ri fo~ the
case of a given permeabilitydistribution
k(r).
Theta-directiontransmissibility
between
angularsegmentsj and j+l is
e

fl+?-g,k
. (A-13)

%/360J@j+~ji-1)

e alue
fnEL-%
or=1si

w e

Finite-difference
notationis

where the effectiveinterlock permeability


must be
0.5(A$ +AO+l)
e
i,j+%,k

Singleterm qt k is the productionrate from


gridblock i,j~i ~ Each term in Eq. A-10 has
unitsMcf/D.
..

i,jiwc=

0.00633a(rmi+1-rmi)A%kei,

Ls a chord slopewith respectto # of the term


representinggas-in-placein the grid block.

(A-14)

..0
~ej+l+

A ~n+l
r r r i,j,k
~

-T ri*, j,k(@i+l,j
,k-Qi,
j ,k)n+l
,k-@i-l~,k)n+l. . .(A-18)
-T ri-%,j,k(Qi,l
?

andzIi*is the log mean radiue


(r
i-l-l
- ri)/~n(ri+l/ri),
Transmissibility
for verticalflow between
layersk snd k+l is

Eq. A-10 is written for teachblock in the finite


differencegrid. Letting,NR, NY, and NZ
representthe number of grid incrementsin the
(r,O,z)coordinatesystem,the totalnumber
of such equationsis N - NR*NYDNZ.
The additionalpotentialdrop at the wellbore surfacein each layer due to skin is

T
zi,j,ki%=
0.00633(2?ra)(rm
2- rti-~)kz
361)(Azk
+ Azk+l)

ktiAe
j ..(A-IS)

with effectiveinter~lockpermeabilitydefined
as
o.5(Azk+Azk+1)
(A-16)
kzi,j,ki%=
k+l dz

.
kz (Z)
r

k
..(A-19)
D
a
i,j,k. W = 0.00633(2ra)(krAz)kqi,j,k

where@w is wellborepOtt3ntid and Sk %s the


skin factor for layerk. This equationis
writl:en
for each completedlayerwith qi,j,k
set to zero everywhereelse. The total number
of such equationsis NQ.
One f~nal equationaccountsfor wellbore
storageand is simplya mass balanceon,the
wellboregas-filledvolume. A staticwell
pressuregradient.isaseumedin integrating

SPE 6437

WALTER L. DOWDLEAND PAUL V* ~E.

)1

the statichead equationto obtain


1441T;~
TSR
Gwb
G

wb

1000 MPs
=Cp

.(A-20)

bheo

where

144mr~T8R
c=

(pbh - Pwh)

1000 Mp~

In the case of.turbulentflow Darcys law


is xnodif%ed
to

MD
- 144 ZaRTwa

l-e

-k$gv:
dr
krr

.(A-24)

.**.=

for~adialflow. Integratingthis equationfor


a constantflow q Mcf/D from rl to r2 Yields

The mass balanceon the wellboreis


N+l +

and the right-handside (capacity)coefficient


involvesthe chord slope of the function
g(P) be(p) with respectto 4.- The coefficient
C in the counterpartto Eq. A-20 relating
wellboreoil volume to bottomholepressurecan
be easily derivedfor the two cases of a freely
flowingor pumpedwell.

...qmq=cl+~
q= Trt(01-Q2) . . . . . . . . . ..(A-25)

or
qN=l +
=

where subscriptt deontesmodificationof the


transmissibility
due to turbulence

00 q~NQ = q + Cdo

a,kr

q+c(@w +1-0:) . . . . . .(A-21)

T=
rt

where

2
Rn+
1

. . .(A-26)

*r
a,RTp (< -$)

where
dpl..
. . . . . . . . . . . .(A-22)
c =C~@
a, . 2n~z
w
and q is.wellheadproductionrate. Either q or %nd likewisefor Tet and Tzt.
Qw~ybe
specifiedas a boundary condition.
At the beginningof each time step the transEquationsA-10, A-19 andA-21 are N+NQ+l
missibilityT-. can be evaluatedusing in the
equatiom which containN unknownvalues of 0,
demoninatort~~ value of q existin~at the end
NQ unknownvalues of ql ~ and an unknownvalue of the previoustime step. We found a more
of either total rate, qj 6t wellborepotential, stableand satisfactoryprocedureis to expaiid
is solved directly Eq. A-25 as a quadraticin q, use the value
#w. This systemof equations,
by.Gaussianelimination,making use of handing . of @l _@2
existing
at the beginningof the
characteristics
of the coefficientmatrix.
time step (timen) to calculateq and use that
The chord slope coefficientsdefinedby Eqs.A-10
value to evaluateT . Of courseTrt can be
and A-17 can be approximatedat the beginning
1
updatedusing iteratons
similarto the chord
of each tiu Step by the SIOpeSat @i.,.k as
slope treatment. We have found iterationon
determinedfrom tablesof the functi&fithe q tern in Trt to be unnecessary.
g(p)p/Zand p versus Q. For large pressure
(potential)changesover the time step,twoor
three outerIterationscan be performedwhere APPENDIXB
the chord slopesare reevaluatedand the
ScalingRadial Transmissibility
equationsresolved. We have found on tha
great majorityof problemsthat no iteration
Transmissibility
in a radial (pie-shaped)
is necessary,i.e., the answer is not sigfractureis
givenby
nificantlychangedby iterating.
The above equationsapplywith minor
changesto the case of single-phssa011 flow.
The potentialfor the oil case is definedas
or
P f(p)bo(p)dp
o-r
pb

PO

.(A-23)

kf21Trefhf
Tr .

.(B-2)

SPE 6437

WELL TEST ANALYSISOF HYDRAULICALLYFRACTUREDGAS WELLS

09
-.

Transmissibility
in a llnear fractureis
given by
I
kfl
Tx=~.
. , . . . . . . , . (B-3)

h a linear fracturepore volume is

Vx-wf%li+l-rmi)$hf

(B-7)

For flow it>a fractureof width Wf and


heighthf, Eq. B-3 gives
~ =%wfhf
x
L

Pore volume in a radial fracturemodified


to.simulatea ltnear fractureis from Eqs. B-6
and B-7
360wf
*
.(B-4)

+rmi) . . . . . (B-8)
r - vr~

From Eqs. B-2 and B-4,


!

:- 180 Wf
T:yTr7

Determinationof FractureSectorAngle
. . . . . , , . .(B-5) and SectorSpacing

Baned on a fracturewidth w= %U inches


and
choosingan
arbityaryradialLdistance
r,
where T$ is the transmissibility
in a radial
the
fracturesectorangle
is
given
by
fracturemodifiedto simulatea linear fracture.f
The angle (3is in degreesand represents.the
lswf
fracturesector.
ef=~..
. . . . . . . . . . (B-9)

ScalingPore Volume

It is convenientto selectr such that


Of equals some small angle, say 0.01 degrees.
Remainingsector anglesare than spaced logarithmically. If symmetryperdts one
reservoirq&adrant~o be ;odeled,then Wf
shouldbe multipliedby O.J.

Pore volume in a cylindricalelementis


computedas
2
pv IT(r
- r~~hfgf
mi+l
r
360

. . . . . . (B-6)

TABLE 1
RESERVOIR DATA AND GAS PROPERTIF7
FOR COMPARISON OF ANALYTICAL l!JSD
NUMERICAL SOLUTIONS (FROM REF. 23)
-WSERVOIR DATA

GAS PROPERTIES
,,

Pi

== 2,300 psia

= 10 fit

r
.

= .5 ft

r=

= 3,oOO ft(640 acre spacing!

@
9
T

Sc

P Sc

m(p) 2
MM psi /cp

2,0f20

0.95
0.90
0.86
0.81
0.80

0.0117
0.0125
0.0132
0.0146
0.0163

14.45
57.38
128.17
224.85
340.95

2,400

0.81

0.0180

400

800
1,200

L.

130F

=5md
= 0.1 (fractionof bulk volume)
= 0.77 (fractionof pore volume)
= 60F
= 14.7

psia

1,600

468.79

TABLE2

TABLE 3
MODEL RESULTS FOR EXAMPLE COMPARISON
WELL WITH AN INFINITE CONDUCTIVITY
VERTICAL FRACTURE, Xf = 22.21 ft.

MODELGRID BLOCK CENTERS, BLOCK BOUNDARIES


AND SECTOR ANGLES USED IN EKAMPLE COMPARISON
OF ANALYTICAL AND NUMERICAL SOLUTTONS

OF A

BlockBoundaries
ft.

Block Center
Radii, i%

0.62
1.03
1.91
3.53
6.52
12.03
22.21
41.01
Ion,02
if5.78
258.14
476.62
0s0.01
1,624.82
3,000.00

0.50
O*75
1.38
2.56
4.72
8.72
16.09
29.71
54.s5
164.99
187.03
345.33 637.60
1,177.24
2,173.60

o
:.:;:
0:030
0.040
0.042
0.062
0.102
0.142
0.202
0.302
0.415
0.615
0.815
1.015
2.015
3.115
4.151
8.151
12.151
20.152
28.153
36.154
41.505

SectorAnglcs
Degrees

No.

1
2

0.06
0.11
C.31
0.P7
2*5O
7.15
20.46
58.54
m

hrs

In(pi)-ln(pwf)

m (Pwf)

Pwf

MM Psi2/cp

M14Psi2/cp

w
2390.0
2271.5
2258.2
2248.8
2241,4
2240.3
2229.6
2214.1
2202.8
2190.2
2175.4
2163.7
2150.7
2141.2
2133.6
2105.9
2085.1
2071.0
2039.2
2021.3
1998.9
1984.1
1972.8
1966.5

434.65
425.20
420.82
417.75
415.32
414.97
411.50
406.4S
402.86
398.83
394.14
390.47
386.21
383.17
380;72
371.S6
365.29
360.87
351.02
345.54
338.75
334.31
330.95
329.09

TABLE 4
COMPARISON OF ANALYTICAL AND NUMRRICAL SOLUTIONS
FOR THE VERTICALLY FRACTURED WELL EXAMPLE

t
hrs

0.01:
0.020
0.030
0.04G
0.042
0.062
0.102
0.142
0.202
0.302
0.415
0.615
0.815
1.015
2.015
, 3.115
4.151
.-.
8.151
12.151
20.152
28.153
36.154
41.505

Bottom-hole Pressuret psia


Infinite
Uniform
Numerical
Flux
Conductivity
2300.0
2270.6
2260.6
2253.4
2247.6
2246.8
2237.7
2224.2
2214.0
2202.3
2188.0
2175.9
2160.5
2149.2
2140.2
2111.3
2092.4
2079.9
2049.9
2031.9
2008.0
1994.1
1983.6
1977.8

2300.0
2268.5
22!55.5
2245.8
2237.9
2236.9
2224.8
2207.7
2195.6
2182.1
2166.1
2153.0
2136.6
2124.7
2115.3
2085.5
2066.2
2053.4
2022.9
2004.6
1983.0
1969.0
195s.4
1952.4.

2300.0
2271.5
225S.2
2248.8
2241,4
2240.3
2229.6
2214.1
22a2.s
2190.2
2175.4
2163.7
2150.7
2141.2
2133.6
2105.9
2085.1
2071.0
2039.2
2021.3
1998.9
1984.1
1972.8
1966.5

m (Pwf),
Infinite
Conductivity

434.65
424.89
421.58
419.23
417.34
417.08
414.11
409.73
406.47
402.70
398.13
394.31
389s47
385.91
383.10
374.18
368.43
364.62
355.63
350.29
343.52
339.06
335.70
333.83

MM p?;i2/cp
Uniform
Numerical
*
434
424
419
416
414
413.84
409.94
404.44
400.57
396.25
391.21
387.10
382.00
378.31
375.42
366.32
360.51
356.67
347.64
342.28
335.50
331.03
327.67
325.80

434.65
425.20
420.82
417.?5
415.32
414.97
411.50
406.48
402.86
398.83
394.14
390.47
386.21
383.17
380.72
371.86
365.29
360.87
351.02
345.54
338.75
334.31
330.95
329.09

9.45
13.83
16.90
19.33
19.68.
23.15
28.17
31.79
35.82
40.51
44.18
48.44
51.48
53.93
62.79
69.36
73.78
S3.63
89.11
95.90
100.34
103.70
105.56.

TA8LE5
VERTICAL-FRACTURE
GAS-WELLEXAMPLE
TEN-DAYDRAWDOWNTESTS

Initialreservoirpressure,p%,.psis

1625

Drainagearea,
A,acres

160

Porosity,
$ fractionof bulk voluma

.064

Thickness,h, ft

63

Formationtemperature,
R

570

Gas

SS

saturatim.

0.633

-1
Totalsystemcompressibility,
Ct, psi

-4
8.0 X 10

Viscosity,p, cp

0.014

Co!npreesibility
factcc,Z

0.845

Reference
presswre,
Pat,pSii3

14.65

Referencetemperature,
Tee, R

520

Gas gravity,(air-1)

0.6

Flov stringdiameter,inches

4.5

Flow strin~veight,iQ8/ft

11.6

Wellb.ieradius,ft

0.35

Averagewellborete~perature,
*R

545

ft
Well depth,

4930

Perforated
interval,
ft
Fracture
width,
in
Fracture
height,
ft

0.1

4725-4755
30

TEN-DAYPRIXSUREDRAWLIOWN
DATA

q = 525 Mcf/D
I

Ap2
!!&&

0.02
0,03
0,04
0.05
0.07
0.09
0.20
0.30
0.40

I
I
I
I

O*9O

9
10
14
40
50
70

100
150
200
240

0.0078
0.0114
0,0149
0.0185
0.0250
0.0295
0.0314
0.0534
0.0733
0.0905
0.1065
0.1338
0.1568
0.1669
0.2282
0.2743
0.3101
0.3390
0.3849
0.4193
0.4339
0.4791
0.5300
0.5923
0.6384
0.6767
0.7368
0.7839
0.8043
0.8863
0.9478
0.9873

2Ap/q
p~
14.85
21,64
28.44
35.22
47*55
56.18
59.87
101.6
139.5
172.4
202.8
254.S
298.8
318.0
434.7
522.4
590.7
645.S
733.2
798.7
826.5
912.6
1010
1128
1216
1289
1403
1493
1532
1688
1s05
1881

~=
Apz2
MMusi
3.0114
0.0169
0.0224
0.0253
0.0343
0.0440
0.0485
0.0835
0.11.44
0.1426
0.1679
0.2126
0.2505
0.b675
0.3626
0.4351
0.4932
0.5404
0.6134
0.6680
0.6907
0.7574
0.8329
0.9243
0.9907
1.046
1,133
1.201
1.230
1.350
1.438
1.494

765 Mcf/D
.

q=

1000Mcf/D

l@2

Mkrurli

.Ap2/q
psiz/Mcf/D

0.0127
0.0195
0.0266
0.0337
0.0472
0.0601
0.0665
0.1144
0.1575
0.1967
0.2325
0.2954
0.3469
0.3705
0.5067
0.6109
0.6946
1..7596
0.8613
0.9373
0.968S
1.063
1.164
1.285
1.373
1.445
1.559
1.648
1.686
1.842
1.955
2.029

12.66
19.46
26.58
33.69
47.24
60.10
66.53
114.4
157.5
196.7
232.5
295.4
346.9
370.5
/ 506.7
610.9
694.6
759.6
861.3
937.3
968.8
J.063
1164
1285
1373
1445
1559
1646
1686
1842
1955
2029

,2

plQz/iicfjD

14.85
22.06
29.25
33.06
44.89
57.s4
63.43
109.2
149.6
186.5
219.4
278.0
327.5
349.7
474.0
568.8
644.7
706.5
801.9
873.2
902.9
;::;1
1208
1295
1267
1482
1570
1608
1764
1879
1954

TASLE 6
VERTICAL-FRACTUREGAS-WELL EXAMPLE
THREE-CYCLE TEST

Flow rate,

q, Mcf/lJ

345 (first flow)


765 (second flow)
1105 (third flow)

Flowingtime,t, hours

6(for each flow)


6 (first and eecondbuj.ldup)
12 (third buildup)

Shut-in time between flow


period, At, houfs
.

Wellhead Pressure Data

First Cycle
t or At

hours

Pwfs Psia

PW8V psia

0.5
1.0
1.5
2.0
2.5
3.(3

1427.1
1418.2
1410.7
:404.3
1399.0
1394.5
1390.7
1387.5
1384,6
13S2.0
1379.7
1377.6

1493.7
1402.4
1407.7
1411.4
1414.2
1416.4
1418.2
1419.7
1421.0
1422.1
1423.1
1424.0

:::
4.5
5.0
:::
6.5
7.0
7.5
S*O
s.5
9.0
10.0
10*5
11.0
11.5
12.0

Third Cycle

Sacond Cycle
Pwfg

Psfa

1396.6
1375.1
1354. s

1337.0
1321,7
1308.7
1297,6
1288.2
1280.0
1272.8
1266.4
1260.6

TABLE

PW8Y psia

Pwf, Ps~c

1306.4
.1334.2
1351.3
1362.4
1370,7
1376.9
1381.7
1385.8
1389.2
1392.1
1394.6
1396.8

1356.8
1323.9
1290.4
1259.2
1232.4
1210.2
1191.0
1174.3
1159.7
1146.7
1135.1
1124.7

RESULTS OF COMPUTER SIMULATION


OF THE VERTIC!AL-FRWTURE
GAS WELL
THREE-CYCLE
TEST

I
$

kh,

Reservoir perrneability-thickness,
ntd-ft

10.14

Fracbure half-length, Xf, ft

140

Turbulence coefficient, 6, ft-1

5;640x1013

Fracture flow capacity, kfwf, red-in

100,000

Fracture height, ft

30

pws, psia
1200.3
1250.3
12S4.3
1306.1
1320.7
1331.3
1339.4
13.fi%e9
13:
4
1356.1
1360.1
1363.9
1367.3
1370.3
1373.0
1375.4
1377.6
1379.6
13s1.5
1383.2
1384.7
1366.2
1387.5
1388.8

4
?

#
qs

2
4

tl

t;

tj

tn.1

n~, t

FM

Fig. l-Productimhtstoryof
agaswll
which proAces at n diffmnt
ratea.

I
Wdlbom

t-

2
1

;
Iez

1234

...

NR
I

Fi .2-6eneralized
Ca! Iyfrar.turedweil.

achemtic modal gAdayntakforavertf-

Fi . 3-%haxatic
we! I exqles.

grtdsyatam for vertically

*
Urmnnl F&x Fmctum

Id

t~,4!2Qw#

xf

10

+++
Fig. 4-~OvstD

@#=

foraunfforn

flux vert,cal fracture, frm Ref.5.

td

10

?0=
Fig. 5-~uvstD

foraninfinite

con&ictfvity verfical fracture, fr@v R6f. 5.

fractured

.------?-------?-------5$

-----X-----J!

~bl+
+---L--L----ti5Fti--&---ti5Fti
I

I@

Ii

Ii

I
~,

I
1

TMAiCti
FOINTi
Itslohm
~

ae
dss

ld$

I
@fil%p

IF

UP

rasultanith

the inftnitecw-

Flowllm*,t
Ism
Fig. 6- C+arison-ex@a
typa-curveeatchof
ductivity colutim of Ref. 5.

thamwwical

to

Flmv Time, t ma

Fi . 7-Am(p)
vssquars-rcotof
ti.m for theanaIyiciil-mmerical
!
crmpariscnexmmle.

:7

--

400

w)

m(plht) x W0,7S aloe Pcizmp

j
Z*

ms S1.2x 10SFSI%CP
SEMILOO
STRAIQNT UNE

240 -

220 ~

10

Flowllme,t
Fig. 9- Smilog plot cf rn(p~)
twmfcal cmparkwOexa
@e.

vaflw

100

t.hm

hrs

tim fcr drawdowndata of cnalytical-

Ff . 1O-T
!?
aoutionof

e-curvainctchof varticell -fractured gas-ttell exacplenit


af. 5-ten drcvdmn, q=7d lkf/O.

hunifornsflus

1.

2400 -

-OoD

t
2000

n
: .-

.
Intarcept

:?Zoo
1
< la

A
.
0

-m

-200

S40 -

F@ 1 Met t D

~$

-300

*%..

%.

%.
~

a26PelftsaflD
-m

~-

I
1

Oim

I
fo

00

I
foo

low

2od

400600

lao

800
Flow Iseta, (!

I,hra

Fig. ll-Hillheic-Cfchwicz
plot forthethree
fractmedgas-uell
excmple.

tendcydrawdmrnaof

thev8rtically-

(P8it Mcf/D)*

400 -

s 0.262

Fig. 12- cIWJplCIt


of Ap2/q vs q at td
fractured gas-well ex~la.

;\,
tooo

,
1200

s4DD

Mcfm

hr. from Ftg. 10 for the vartically-

!!+t
1200 -

i
noo

Ossmwd
Cakulsfmd

MctlD
CIZs 7d2 MctlD
QJ * 110S Mctl D
q

.245

I
-o

slo

15202520

Kl(loo~6

2tj40

~
4a

TIME. hre

71ME,

Fig. 13- Ccmparieon ofcbserved end mdekalculated


welI head pressures for tha verticall y as-well exa#ple thrae cycle test besed on resul te frcm tha Millheitiiciw
I od.

Fig. 14- Ikmparison of observed and madel-calculated wel!heed pressures for tha vertfcal I y-fracturd gas+ell ex~le
thrap cycle test based on final nwical
aimletfcm rasulta.

UiCZ mt

t t

1000

S403

I
i

WHP x 760 p9i!3

1200
t

,.

l
200,

hn

No Fraofura
K . ------------------------

TIME.

months

Fi .15- @artson of fcr~asted prcdcctfcn parformnca


k!ore and after fracturing for the vertical Iy-fractured
*e-uell axaaples.

You might also like