WC b 61%J1U

M@wtJ@xan EnLmms

Decline Curve Analysis for Variable Pressu, J Drop/Variable Flowrate Systems

by T.A. Blaslngame, T,L. McCray, and W,J. Lee, Texas A&M U.

This is a preprint -- subject to correction.

cop~hhl t99\, S&ety of Pe!rdeum Engineers, Inc.

This fmper WA8 prepeti Psvpmsenlatim a? dm SPE Qat Tedmbgy Symposium held in Hwston, Texas, January 23-24,1691.

TM paprx was $elemed for presenlalicm by m SPE Program Commitrtoe following mtiew of inlormetkm conleined In an Mrnc4 wbmirted by the author(s), C%nmn,s of Um paper, ac
pmwnted, ham not bean revkwed gmesoc*volPlrk e o urn I%qineem tnd are subject 10 C&motion by tie eulhor(t). The materiel, : presented, does not necosaanly rellecr any
poslrim_Iof rho Sodaly of Petdeum nglneem, 11solfioerc, m members, Papera presented at SPE meetings are wbjact to publidon ravbw by Editorkil committees of the Sodary of
Permlwm &@neers, Permlmbn to copy is restdcled 10 an ab$tract o! not more thnn 303 words. Il[utl!arions may no! be copied. The abstract should conlain .%nsplcwous
~~t~ *m @ @ fim me paper is Pmmmtad. write puMic.atas hkwer, SPE, P.O Sex S3S9SS. Ridurd$on, TX T5C8S-21S6, Telex, 730369 SPEOAL

The motivationfor the workdescribedin this paper arose from
a need to analyze production dcclinc data where the flowing
bottomhole pressure varies significantly, The varirtncc of the
bottomhole rcssur@with time CXCIUS.ICS the use of the exponential
decline mJ c1for conventional dcclinc curve analysis (scmilog
plots arid type curves). Using pressure nonmalizcd flow rate
rather than flow rate usually does not remedy this problem. The
rncthmi wc present uscs a rigorous superposition function to
account for the variance of rstc and pressure during production.
This furwdon is the constant rate analog for vanablc.rate flow
during post-tmnsicnt conditions and can bc used to develop a
constant pressure analog for the dcclinc curve analysis of field
data.
formed into an cquivslcmconstantrate ca.wfor both gas and liquid
flow data. Camacho9indcpndcntly vcnficd that this equivalent
constant rate formulation is exact for the constant pressure
production of a sli htly comprtssiblc liquid during boundary
dominri:ti flow con#mons,
McCrayto sought to develop a method to transform variablc-
rate/vanablc pressure drop data into an equivalent constant
pressure case. In doing this, McC!raydcvclopcd a recursion
formula to compute an cquivdcnt time for constant wellbore
pressure production, tcp, that could bc used with pressure drop
normalized flowrate to perform dcclinc CUTVC analysis using type
curves.

The constant pressure antdog time function is computed from
the constant rate function using the identity that cumulative
production for both cases must be qud. Using the cumulative
production identity, wc SOIVC recursively for the time function
usin~ trapczoirlal rule integration and, as an altcrna!ivc, frnitc
diffcrcncc formuhtc, Wc have also dcvclopcda const;mtpressure
analog time rchuion which is rigorous for bounrky domimttcd
flow and stxvcs as an accuraIcapproximationfor tmnsicntflow,
We apply theserelationsto analyticalsolutionsfor verification
and then usc the boundary dominntcdflow relation on simulated
and ticld cascst These wmuhuioncases include htrgc rmdsmall
step chun CSin bottomhole tlowirtgpressures,and pcricxiicshut-
ins. Finn]! y, wc apply these relationsto a gus w.IIfield case,
Although the approach st,rggcstcdby McCray was verified
using simulation, wc sought a rigorous foundation for the
applicationof this result, As it turns out, a relativelysimple proof
can bc shown for the application of the rcp function during
boundary dominated flow conditions, This prcmf is given in
AppendixA.
In addition to the proof of McCraysresult, wc also provide
mcthmls to compute the constant pressure equivalent time, 2CP,
using recursion formuhm in Appendix B. In Appendix C wc
provide relations which can bc used to compute the constant
pressure dimcsionlcss mtc solution given the ccnsutnt rate
dimensionlesspressure solution, In the text wc will pmvc thtitthe
computatiortalmethods wc provide ylcld csscntittllyexact results
during boundarydominated flow and give very good pcrfornmncc
during transientflow.

The widespread usc of ty~rcCUI-VCSIJ to analyze mlc dcclinc ~

data has motivated us to consldcr the implicationsof varying rate
and pressure drop production. 1hcorcticallys~aking, for the
flow of a slightly compressibleliquid, the analytical stems cmthe
Fctkovichl ty curve arc valid only for the ccm!ant wellbore
pressure pr z uction CMC,+6 In previous worksv,a wc have
shown that variablcwatdwtriablcpressuredrop data may bc trans-
Rcfcrcnccsand=tmtions at cnd of paper
Our first goaJ is to establish that our new mcttisxi actually
transforms a variable.ratclvariablc pressure drop systcm into an
equivalent constant rcssure systcm, Wc begin with a proof of
the validity of our soPution by transforminga constantrate system
into a constant prcssutv s stcm. Because these solutions arc
frqucntly used in dimonsrSMIMSformat, wc will perform this
verificationusingdimcnsionhw variabkx.
 2 DeclineCurve Analysisfor VariablePressureDrop/VariableFlowrateSystcn?s SpE 21513

The computationalformulaearc givenin AppcndiccsB and C functions. Wc will demonstrate the ap~licationof the boundary
and wc will verify each. Thcsc include the fcilowing rccuraion dominatedflow mcthcxion a simulatedhquid productioIIaqucncc
formulae; the integral method proposed by McC!ray10and the 2- and a fic!dcaac for a gas WCIIthat has been analyzedpreviouslyin
and 3-point brickwarddtifcmncomethodsdcvclopcdin this work the literature.ZM
The recursion relations for this part of the verification arc
dcvclopcdin AppendixB and summarizedin AppendixC,
We will also use the boundarydominatedflow relationswhich
result from tsating the constant rate and constant pressure
anal tics! SOIU7 or..,This dcvclopmcntand the pertinentrelations In this section wc will apply the tcP-tcrtransform describediri
for ti is psxtof the verificationam given in AppendixB. the previous section to a simulatedproduction squcrtcc in an Oil
WCI1.The mscrvoirdata and flow historyarc given in Table 1.
Fi . I shows the log-log behavior of the qo functions vcraus
tD ant tcp~ function for tic cssc of a well ccntcrcd in abounded TABLE 1
cimulaxreservoir (r~=NP), Duc to the number of mcthcds being
cansidcm we will discuss the transientand boundarydominated WellimdReservdr Parameters
flow behavior separately, First wc note that, during early times (Well Centered in alloundedCircular Resemoir)
(transient flow), all of the tcpp methods yield a good
approximation to tic q~(t~) sohmon, cxccpt at very early times B, RB/STB 1,00
(t@20). This implies that all of these methods yield a reasonable cl, psia-l 15.0 X 1o-I5
approximation to the analytical solution during transient flow, h, ft
Obviously,the analyticalsolutionfor tmundarydominatedflow is 4 0.::
not valid during transient flow as shown by the deviation of this \i, Cp
solution and the transientflow sohstion. rW, ft 02:
re, ft 745 (40 acre)
Now if wc consider the late time (boundary dominated flow) k, md 1,0
portion of Fi&.1 (rD>3x105),wc find that virtually all of the to D s
mcthds agrc.cwry closdy with the qD(lD) Solution.Athough &ls
pi, psia 48~
scale prccludcs very CIOSC
inspection,it dots appear that t D2 for
the dcnvativc mcth(xl1 dots show signi!lcantdeviation, f%iswill CA 31.62
bc invcstigsttcdmore closely when these dara arc rcplottcd on a
scmilog q~ gmph in Fig. 2. Using the WC]!and rcscswoirparametersgiven above and Eqs. A-3
to A-5, wc computethem and b paramctwsand wc obtain
Fig. 2 is a rcplot of Fig. 1 using a scmilog scale for the qD
fLSnctiOnS and a cancsian scsdcfor the tD functions. Wc nOtCthat m= 2,3&/09x10-z@/S~/D/D
the scsulta for dcnvativc method 1 do begin to diverge from those b = 32.8948 pSi&~@
r f the Othermcthtsds,which cIwly ovcr]ay the correct solution,
Fig, 2 sug~sta that dtivativc rrscthod1 should not be used in
racticc, but that the integraltncthw dcnvativc mcthcx!2, and the Flom History
L undarydominatd flow methodshould give accuratewsults.
flow scqucncc r, days q,STBfD p~, psia

Of these, the boundary dominatti flow method is the easiest 180 50 constantrate
to apply since it dots not rcqtlirc recursion calculations, but ; 180 const~t p~ 2000
instead provides a dhtct Uansf{mationbctwccn tD (or tc,D) and .. shut in
tcp (Eq.B-2), tc~o is the is the dimensionless matcnaI balance 1:0 constantp~ 4000
time function, introduced by Blasingamcand UC7 and later (and : 180 Constyt p)@ 3000
indcpcndcntly) b Camacho9as an quivalcnt constant rate time .. 3 shut in
function for variai lc-ratchriablc pressuredrop flowconditions. 5 3s00
6, 1500
Onc problcm that tic boundarydominatedflow method might .. shut in
b pcrccIvcdto have would bc that in field applications,formation 7 180 constantp~ 2500
proptics arc rquircd to computethe m and b constantsin the tcr 8 S40 const~x p~ 1500
tcftransform relation, Eq, IV1. However, these constants arc
easily dctcm~incdfrom a cartesian plot of A@Qversus tc+=Q/q)
during boundarydominatul flow. This proccdurc is verified for a The flow mtc profiles that WCIEobtained from the simulation
variety of cases in Ref. 7, Also, bccausc this is a rigorous cases arc shown in Fig, 3, Wc have inchsdcdan arbitrary base
formulation for boundary dominated flow, the m and b constants case (p~30Mt psia) to orient the analysisof all of tic data, That
arc unique and will provide theoreticallyconsistent results when is, when wc have obtained the correct transformationof data, all
used in Eq. B-1, cases including the base case should overlay the same trend,
Although it is conceivable that these rate profiles could bc
Additionally,Mcthylo reportedinconsistentresults using the analyzed separatelyusing dcdinc curve analysis, the cawistcncy
integral method to compute the rc function when Iargc pressure of the rcsuhs would b dcpcndcnton the ability of the analyst, In
changes ancl/or shut-ins occurc# McCray was forced to usc particular, the transient spikes caused by pressure changes and
cmpisicalextrapolationpointsto cause the imcgralmethodto yield shut-ins would be difficult to interpret, and the non-uniform
correct results, Again, the boundarydominated flow method will behavior after the effect of the spike has subsided would
not have these roblcms Iwausc tic m and b parameters arc inevitablyyield ambiguousrcstdts.
unique and the formulation of the tq+ transfo~ 0%. B-1) iS
rigorous, Onc method used to align variable rate pressure data with the
correct cccnstantrate solution is rate norrnahsstionof the pressure
For these rcmons, wc rccommcnd that the bounda drop. In dwlinc curve analysis, many analystsusc prcsstsrt dmp
dominated flow rncthod which uscs Eqs. B-l (field) or B-T normalisation of the flow mtc profile in an attempt to obtain the
(dimcnsionlcss) be used for computation of the tcp or ?CPD
 SPE 21513 T.A. B!usingamc,T.L. McCrayand W,J, La 3

II prcs:
correctconstant pressure solution. It can be seen in Fig. 4 that
xc drop normalization does not yickl a constant nrcssure
analog solution,
TABLE 2
Well~Reservoir par~ters
(Assumed Geometry: Well Centered in
Clearly, wc must usc other techniques which are more a Bounded Circular Reservoir)
rigorous than psessumdrop normalizationfor field applica{{,ons
of
decline curve analysis, The method of choice will be the onc B, R13/MSCF 0.70942
proposed by Blasmgamc and Lec7 whit: converts variablc-
ratchriablc rcssure tip data to the equiwdcm constant rare et, psia-l 1.870 X 104
h, ft
case. From t is anal sis wc will obtain the m and b parameters
! B-1 or transformation to an equivalent constant
rquired by Eq. # 0.;:
pressure system, & Cp 0,02167
Fw,ft 0,354
Fig, 5 shows the cartesian plot of alp/q vs. tcr(=Q/q) rquircd -5,30 *
to determine them and b parameters. m is the slope of this plot ~, md 0,0786S *
and b is the intcrcc t. Although thcte is some data scatter, it is p.~ psia 710
ckar that them an1 b psramctcra do rcprmcnt a best fit trend of pi, psia 4175
the dam. Therefore, the step of determining the m and b C* 31,62
parameters is illustrtttcdas a simple and straightforward process. G, Bscf (!ef,2) 3.360
Fig. 6 shows the log-log plot of Ap/q vetmssrc, that could be used G, Bscf (rcf,3) 3.035
for type curve matching on constant rate type curves. Fig. 6 also
shows that the concept of using Aplq and fcr appears to also be * Averageof valuesobtainedfrom ref. 2 and 3.
valid for transientflow, given the agreementbetweenthe constant
rate and constant pressure base case (pW~3000 psia) during From the results of ref. 8 wc have
transientflow (rcrc50days).
ma = 2,O5536X1O-3ps@lSCFP/D
The next step is to usc them and b parameters in Eq. B-1 to b. = 1.3094psi/MscF/D
convert from tcr (constant rate analog time) to tcp(constant G= 2.6281 Bscf
pressure analog time), This is also a simple and straightforward
procedure. Once tc is computed, a log-log plot of q/@ vs. Gp is Fig. 8, which is a log-log plot of @a/q versus tc~,a,is taken
made. Fig. 7 is suci a plot and wc immcxiiatclynote that all cases directly from rcf, 8 and shown here for complctcmess. Wc htwc
overlay the same trend during both transient and boundary included the computedresponseduting boundarydominated flow
dominated flow. obviously, the analyticalsolution for boundary as prescribed by Eq. A-1, Thc Apa/4 and tcr,a variables arc the
dominatedflow (exponentialdecline) will not agree with transient pseudotimc and pseudopressureas defined and computed in ref.
flow solution. 8, This nomenclature may seem awkward, but defining these
variablesin this manner allowsus to usc liquid flow quations for
Fig, 7 represents the endpoint of our effort to determine art analysis, Therefore, any equationswc preaem are valid for either
quivalent anstant pressuretransformationfor variablc-rate/vari- liquid or gas flow as long as the correct time and pressure
~blcpressuredrop flow data, Wc are satisfiedthat this is a logical variablesarc used.
and consistent procedure that should yield accurate rcsuhs when
applied to field data, The verification of this method is that all Wc note that the boundary dominated flow sohstiondoss not
cases overlay the base case @~-30fXt ]sia), where ~IAPand t model the transient flow behavior of the data in Fig. 8. This is
were used as the plotting functions for I$C base case. At this cxpectcdand wc only state this observationfor completcncss, Wc
point, Fig, 7 can & used for decline curt c analysis using type could use Fig. 8 in a type curve matching analysis with constant
curves such as the one presentedby Fctkovich,l rate type cwves. However, it is our objective to analyze these
data with a constant pressure typeCUIVC so wc must proeecd with
transfomningthe tcr~variableto yieldCcp,a.
This analysisrquircs that wc modify the variables in Eqs, A-
1, A-3, A-4, and B.1 for analysis of gas WC]]test data using
In this section wc will apply the ICrtc, transform @q. B-1) to pscudoprcssurc and pscudotimc, For gas analysis Eq. A-1
the analysis of vttriablc rate gas well data as described in becomes
rcfcrcnccs 2,3, and 8. Unfortunately, the complexity of this
analysis is compounded because gas WCIIanalysis requires the
used of paeudoprcssurcand pscudotime, Wc will refer to ref. 8 (1)
for the pseudopressure and pscudotimc functions as WCIIas for
the msulta of an iterative procedure to dctcrminc the m and b Eq, A-3 becomes
parameters and the gas-in-place, G. This iterative procedure, as ~a=_&
described in ref. 8, simulau+ntmuslydctcrmincs these parameters (2)
tsauac the pseudotime function requires Icnowlcdgcof the gas-
in-phw for matcdal balancecomputations. Eq. A-4 becomes

The reservoirdata and flow hiworyarc given in Table 2.

(3)
For a closed cyhl.rkxd swemoir Eq, 3 becomes

(4)

I
 4 DcclincCur-wAnalysisfor VariablePressureDrop/VaririblcFlowrateSystetm
And finally,for gas weil test analysisEq. B-1 hxomes
(5),
We have usedEq. 5 to eomjwtcthe tc~ functionsused in Fig.
9. Note that Fig, 9 is a log-log lot ot q fAP~ versus tcp,~and that
the the boundary dominated h ow sOIWiOn(Computccfq14Pa
function) agrees very wcli with the data during boundary
dominated flow but not during transient flow. This is expected
and we shouldnot be conccrncdabout this difference.
Once we have created Fig, 9 using the q/Apa versus rCPadata,
wc will want omate} this data upon the Fetkovichl type curve,
Fig. 10 reprc%ms this type curve match. Note that the data agree
with the t curve during the transition from transient to
boundary r ominated flow and throu hou; boundary dominated
flow, The scamity of data for tc ~<1h days (4 points) iimits our
interpretation of the transient Fiow portion of the data, but an
estimateof r#W=20 for the tmnsientstem seemsreasonable.
Once the data are overlain and matched to the type curve, we
will determine a match point from the coordinates of both plots,
The match point for Fig, 10 is
RateFunctionMatchPoint:
qD= 1.0
q{dp = 0,78 MSCFiD/psi
Time MatchPoint:
t&=l
tcP~= 630 days
,.
Cwve~
Once the match point is determined for a data set on the
Fctkovichl type curve, we can computethe volume of fluids in the
reservoir and the formation permeability, k. We need
computational relations for liquid (oil) and for gas where the
correct pseudopressureand pseudotimefunctionsare used.
The oil-in-place,N, can be computedusing
~ This paper introduces a method that can be used to analyze
N=.~ & h constant pressure analog time function, The most significant
ct [()(1]
9dD mp rdD rnp (6)
The formationpermeability,k, can& computedusing
(7)
The gas-in-place,G, can bc computed using
=*[(~)mp(~~)mj (8)
The formationpermeability,k, can be computedusing
(9)
SPE21513
Also, for a bounded cirat!ar reservoir, the fosmation permeabi-
lity, k, can be computed using
-g_
k=14L2~
Polo
~-z
~w 4
*
qdD
Using Eq. 8, wc can estimate the gas-in-place, G, This calcula-
(lo)
tion gives
G=&[(w)(Hll
G = 2.6278 B~f
Using Eq, 10, wc can cs:imatc tthe formation permeability, k,
This calculationyields
~ ~ ,41 ~~0.70942 0.02167
(;0) +@o)-ii(w)
k= ().()5432md
The computed values of gas-in-place and permeability
compare very WC]] with those obtained in ref. 8, In fact, the
results arc vimsally identical. The reason for this is quite simple,
The constant rate analog method (using @a/q and tcr,a) and the
constant pressure analog method presented in this work arc
rigorously related, Therefore, if wc are consistent,both methods
shouldyield the same results.
Our estimate of gas-in-place, G, is 22 percent less than the
estimate of Fctkovich, et al,q and is 14 percent Icss than the
estimate given by Fraim and Wattcnbarger.j However, wc
express confidence that our estimate is as acclqratt as the ones
given by the other investigators. And wc feel that our approachis
more rigorous, because of the pscudoprcssurc and pscudotimc
soiutionfommiation.
Our estimate of permeability is within 31 pcrccnt of the
average of the estimates given in refs. 2 and 3. We feel that our
permeability estimate is consistent with these vahsesof the other
investigatorsand ccstainlyreasonable,even given the sparsedata
trend in the transient flow region. These data arc used to identify
ths rJrWc value finm the tmnsicntstem type curves.
SJMMARYANDCOKLJJSIONS
variable-rate/variable pressure drop production data using a
result is that of the boundary dominated time transformationfor
constant mte or constant pressure flow, This transform, given by
Eq. II-1, allows an analyst to compute an equivalent time for
constant pressure production, quickly and easily, based on the
parameters m and b, obtained using Eq. A-1, For gas wclis this
prcdurc is lCSSstraightforwardand rquircs an iterativesolution
developed in ref. 8.
Wc considered four diffcmnt methods to tratyfomrtvariable.
rate data into the constant pressure solution profile, Wc
considered three recursion formulae which compute the constant
pressure~uivalcnt time function, tc,by panel summationsbased
on trapezoidalrule intcgrationloan{fmitc diffcrcnccexpansions,
Each of these relations was ap lied successfully to convert the
constant rate dirncnsionlcss soPution into the constant pressure
dimcnsionlcs solution. Iiowcvcr, McCraylo found that
application of the trapezoidal rule integration method gives poor
results when a plied to erratic data or data with cxtcncded shut-
ins, This bc[ avior makes the general application of these
rccrnsionfwrrwlacdifficultat best,
 SPE21513 T,A. Blnsirsgime,T,L. McCray and W.J. Lee
The fourth method developed was a rigorous identity which
equates the kxtndary dominated solutions for constant rate and
constant pressure production. The resulting two-parameter
relation @q. B-1) may bc used for dimensionless solutions or
field data applications. When the m and b parameters are
determined using the methods developed in ref, 7, data scatter
should have Iittlc effect on the VSIUCS
of the parametersbecause a
best fit trend is established. These characteristics make the
boundary dominated flow method the most useful product of this
work.
Applications:
Wc recommend using the methods prcscntcd in this work thr
the type curve analysis of variable-rate/variable pressure drop
production data, The methcd is relatively simple and should be
applicable to a wide range of WCI1test problems, including the
analysisof gas well test data demonstratedin this work.
Conclusions:
1, The recursion fonnulae discussed in this work should not be
applied in practice due to problems associatedwith the erratic
nature of field data, which could cause poor results.
2. The boundarydominatedflow method is the methodof ckaicc
to transform the constant rate analog time function ir,to a
constant pressure analog time function, This mcthid is
consistent, easy to apply, and should give accuratercsuits for
a wide range of problcmtypes.
3. The boundary dominated transform method can be used co
model constit wellborepressure productionbehaviorcxaciy
during boundary dominated flow and should give accurate
results during transient flow.
NO a
Dimensionless Variables
bD = dimensionlessconstant defined by Eq, B-4
CA = dirrmsionless shape factor
= dimensionlessconstant defined by FA.B-3
% = dimensionlesspressure (conslantmtc case)
f@ = dimensionlessslltc(constantpressurecase)
Y = Eulersconstant (0.577216 .,, )
Field Variables(Formationand Fluid ProDerliexJ
fo~ation volume factor, RB/$IT3
constant as defined by Eq, A-4 (liquid) and Eq, 3 (gas)
to:sI compressibility,psiai
gas-in-place,MSCF (or 13scfas in the fieldexample)
total formationthickness,ft
formationpcrmeabllity,m?
constant as defined by Eq. A-3(liquid) and F.q,2 (gas)
oil-in-place,ST13
initial reservoirpsessurc.psi3
flowing bottomholepressure,psia
flowrate,STB/Dor MSCWDfor the fieldexample
cumulativeproduction,STB
wcllbom radius, ft
reservoir chainagcradius, ft
re/rw, dimensionlessrescmoir drainageradius
porosity, fraction
viscosity, cp
Subscripts
a = adjustedvariable for gas well test analysis. Usc of
these variables in gas well test analysis yields an
cquivalcrstliquidsys~cm.
&. constantrate or conwmt rate analog
5
Cp = constantpressureor constantpressureanalog
D= dirncnsionlcssvariable
mp = matchpoint on a typecurve
Wc gratefully acknowledge the assistance of Elizatwth
BarbozattndJemniferJohnston for their hcIp in the preparationof
dds manuscript.
1. Fetkovich, M.J.: Dcclinc Curve Analysis Using Type
Curves,JPT (June 1980) 1065-77,
2. Fetkovich,M.J., et ULDecline-CurveAnalysisUsing Type
Curves.-Case Histories,SPEFE (Dec. 1987) 637-56.
3. Frairn, M.L. and Wattcnbmgcr, R,A.: Gas Reservoir
Decline-CusvcAnalysis Using Type Curves With Real Gas
Pseudopressureand NormalizedTime, WEFE (&c, 1987)
671-82,
4. Ehlig-Economides, C.A. and Ramcy, H,J., Jr.: Transient
Rate Dcclinc Analysis for Wells Produced at Constant
prCSSUR, :PEJ (Fcbo1981)98-104,
5. Ehlig-Economides, C.A. and Ramey, H,J,, Jr,: pressure
Buildup for Wells produced at a Constant pressure, $PEJ
(FcIJ. 1981) 105-114.
6. Blasingamc, T,A. and Lee, W.J.: Properties of
Homogeneous Rcscrvoi.m,Naturally Fractured Reservoirs,
and HydraulicallyFracturedReservoirsfrom Decline Curve
Analysis, paper SPE 15018 presented at the 1986 SPE
PcrrnianBasin Oil and Gas RecoveryConfcrencc,Midland,
TX, March 13-14, 1986.
7. INasingamc,T.A. and Lee, W.J.: Variable-RateReservoir
Limits Testing, paper SPE 1.5028presented at the 1986
SW? Permian Basin Oil and Gas Recovery Conference,
Midland,TX, March 13-14, 1986.
8. Blasingamc: T.A. and Lee, W.J.: The Variable-Rate
Reservoir Limits Testing of Gas Wcllst paper SPE 17708
prcscntcd at the 1988 SPE Gas Technology Symposium,
Dallas, TX, June 13-15, 1988,
9. Camacho-V,, R,G,: Well Pe@ormance Under Solution Gas
Drive, Ph,D, Dissertation, University of Tulsa, Tulsa, OK
(1987),
10, Iviccrav.T,L,: Reservoir AnalvsisUsing Productwn Decline
Data &d Adjusted Time, M,S, Tkis, Texas A&M
University,College Station, TX (1990),
Appendix A: ~ ,&!&QWtm
of v~ .
We will start with a relationwhich allows for the constantrate
analysis of vsiriablc-ratekwiablc-pressure drop test data during
boundary dominated f!ow, This relation has been verified for
liquid flow by Bhsingamc and Lec;~ro?d~~machog, This relation
has also been verified for gas flow by IMsingarnc and IA@, The
constantrate analog,wlationis given as
 6 DeclineCurveAnalysisfor VariablepressureI)rop/VaricblcF!mvratesystems SPE21513

4(0
=mtcr+b We need to prove the right-h~.,ld-side(RHS) of ~., A-8. This is
N) (A-1) done by integrating Eq. A-7 so it is the same form as the RHS of
where
t W,. A-8. This gives

= --1-
Cr f?(r) *
J
q(t)dr =
(t) (time in Days)
q(t)
(A-2)
or
m=5.6l5--@---
@hcfi (A-3)

b=70.6#n~
() eTCA r~2 (A-4) (A-11)
and CombiningEqns. A-7 and A-11 gives
,
rw = rw e-s
(A-5)
We will also need the general solution for a well producing at a
constant pressureduring boundarydominatedflow. Thk solution (A-12)
is given by Ehlig-Economides and Ramey~,Sand later by
Blasingameand Lec.c This so!utionis Notice that the right-hand-sides of Eqs. A-10 and A-12 arc
identical, This result proves that Eq. A-8 is exact for boundary
&)cp
row) (A-6)
dominatedflow,

Our objecciveis to develop a general time function that allows us AppendixB: ~J%!@Jl

to use Eq. A-6 to model a variable-rate/vanable-pressur~ drop
process, This generalrelation is arv 130ml~
The objectiveof this swion is to developa methodto computethe
rcpfunction. A relatively simple relation is obtained by quating
Eqs. A-1 and A-7 and solving for tcp.This gives
(A-7)
~ Exp(~@), *
where tcp is the time at which the constant pressure solution is
valid for a general variable-rate/variable-pressuredrop response. or
In this sense, ICPis an unknown which must lx dctcrmincri.
%/2 =$ql +fpcr) (B-1)
McCray10proposed [he following rcl.~tionas a defining refwion or in termsof dimcnsimless variables
Jmq..
for tcp
tp
where
c/)D =mD td
( ) (R-2)
Q(t)=
. &l ~T
mp=L. tirW2
M(t) o AP(o r@2 A (B-3)
1 (A-8)
McCray developed Eq, A-8 via induction from the constant
pressure solution where ~(?) would be constant and could be
factored out, (B-4)
whew
Although McCray proved Eq. A-8 cmpincal;y using simulation,
wc sought a more rigorous justification. In the fcJlowing
dcnvtitions we provide a rigorous proof of Eq, A-8 for boundary
dominate ! flow. Although Eqs. B-1 and B-2 are strictly valid for boundary
d~ot~a$d flow, we find that, for short times, these relations
Frst we rquire the coustant rate analogresult, Eq, A-1, and
Q. A-2, CombiningEqs. A-1 and A-2 gives tcp Z tcr (K3-5a:
9#1.mQ~+h and, in dimcnsiortlcssfomr
(A-9) tq)J)
= t@ (}1-5b
Solving Eq. A-9 for Q(t)/dp(t) gives Although wed? not present a rigorous proof, J@ B-5a and B:51
do suggest thut Khctcpfunction can M approximated by the tc
Q(O .1 ,- b-@- function during transient flow: General applicationof I@ B-5
AP(t) m [1 Ap(t) (A-10) and B-5b requiresfurlhcr invcstigaticm,
Eq, A-10 provides us with the Icft-:and.sidc(I,HS) of Eq, A-l?.
 SPE21513 T,A. T,L,
131asingamc, McCray and W,J, lee 7

I And combining Eqns. B-8, B-9, and B-11, and solving for AtCP,i
gives
Integral hiethod
McCmylo proposed to compute the tc functionby approximation . ~ @(ti) Q(ti.$ 4Q(ti-1)+ ~
of the integral in Eq. A-8 using ti c trapezoidal rule, This
essentiallyresults in a recursion formula where the tcpfunciion is
repl %(ti) [Ap(ri.2) @(ti-1) AP(ti)1 (B-13)
computeda:
Appendix C: ~~
Cp= f At.p,i
i= 1
[
The AtCP,i
terms are computedusing individualtrapumid panelsof
In the calculations, the constant pressure dimensionless rate
*C q(1) solution is definedas
function. For an individualtrapezoidwc have qcpD = ~
APO) (c-1)
Theequivalentdimensionlesstime is
Ii .%!?L * + ~ti-l) tcpD = i~l AtcpD,i I
2 [@(ti) &(ti. 1) 1 where the ZMCpDj
are given for the integralmethod as
(c-2)
also
- Q(tJ .
/i .. Q(ti.])

[AP(ri) Wi-1) 1 =2[%%+1

AtcPD,i
~ ~ ~-
for a given pane!. [PD,i FD,i-l1 (c-3)
and fol dxivative method 1 as
Combiningand solving for AtcP,igives tD,i pfi,i
AtcPD,i = tD,i -
@2iL. aLIL PD,i-1 (c-4)
[ ~(ti) A~ti- J and for derivativemethod2 as
t~p,i =
,- pll,i tD,i-2 . 4tD,i- 1 &
&,. 4U.L
[@(ti) AP(~i.] ) 1 (B-7)
AlcpD,I 2 [PD,i-2 PD,i-1 PE,; 1 (c-5)

DerivativeMetlwds We can also u~eihc Imund.uydominatedflow method to compute

Other methods, which are based on the dcnvativc of Eq. A.7, can the tcpDfunct]on, In this case, tcpD is obtained using E+ R-2, I
be developedto compute the tcpfunctionusing Eq. B-6.Differen- B-3 and B-4.
tiationof Eq. A-7 with respect to rCPyields
q(t) - d Q(O I
dtcp ()Ap(i) (B-8)
Ret%) .,ICgeneral flnitc diffcrcncc cxp~nsion for the first
dcrivati.
f(x) s k
dxi (B-9)
whwc
~(x) = fi~t derivative
Af = differenceexpansion
a = constant for a given differenceexpansion
xi = panel width for the differenceexpansion.
in our case we will use the first and second backwardsdiffcrcncc I
Expansions,These arc
CaSCl:&=$i -fi.l, U=l m-lo)

case 2 Zw= fi.z - 4fi.1 + 3fi , CS~2 (B-11) I

Combining I@. B-8, B-9 and I3-10and solving for Mclj,igives

Atcp,l
.= MQ.L?U. M!UL
~(ti)[A/l(li)*(ti.1)1 (I-.12)
 o
0 o Y
u
11,,
,, # 1 Lfitll I 1 1 1,,,,
,, I 1,,1,
,, I
1 t
9
j

. .. . . . .. . . . .. ,, . . ...,,,... ... . . . . ,......-.=

--

.............. .!..,,,..,,. . . . . . . .

.,,
.,,.,,.
//

-!
o ,...,,,,,, .

4
v
o

1
2
,,,, ~ ............

E I

(3

...
, !,, ,
? . . . ...<.
................ I

II
g
/ t%
l+---
H L
.$-!
! &i
!

1 1 1
1
l.+
t-)
o 9
m
.W-)

cd

,,,.,,,,,.,,,,,,,..,,,,.

,,.,,,,,,,,,,
,,,,,,,,,,.,,,.,.,,
k-l

c+

1 nl I I # o

suoyxq %
 .. /
o
,-cu~
d

&n /
~twl

r : .
CO
a~a;
cl!
-i+
p>

CQ

(
,.1

o ..$J ~ ,................&..d..,d,,
............... ,, ..,,.,,,,..,,, ,,, ,,, ,,, ,,, ,,, .,,,,
%., ~a
Wo
yc& -/ CQ
g :-
k
.
w
-z T.
c%
Lg =$&3
%
: 5?
8g
,,,
(Q
..-!

V-i G
/If
1~ u
(
,,.,,
\

..,.,,,.,,,,.
,, .,.,,,.,.,,..
IIUQ
$
CA
.,,,,4,,,,,,, %
$
~
TY
.-
.-
-
%

2 m
.,-!
\; - -
% - $
%g u
IL w J
d
a ....,. ,,, ,,, ,, .,.. ......... ....,, ..,,,, ...,, ,.,,,,,,,,,.,,..,...,..,.,,,,,................., ..................... -o
m cd
; ..- /
0 g ~~
/- ,
-
ac

83
8%

qg %g
1(. w
r
a ml
.- &
I

o -.,.. ..... ..........%....>.... E..,,, ,,.,,, ,,, ,,, ,,, ,, ,,, ,,, ,,, ,,, ,., , ,, .,,,,,,,,,,,
%

z/
T j
1!
* !7/
a/ h
.
7

,. t
o
(

CT/am i)
m

L-9 G
0
1- c)
t- r-l

1....... H :
%

-3
;:

~-% E
/
,.,
,,.
I-.4
WJ

..... / T1
.-

A)
%
,,! ,,, ,,4
! . ...4.... . ... ,, ! ...,,,,.,,, ,,, ,,, ,, ,,, ,,, .,

.ij-
.
t-a
> ~- -

./ J /,:
&
; g~
\

::/

W-io
y~ fiq .-8
~
c1. I 2
C&

.,
.,,!.,,. . . . . . . . . . . . .,.!...., !! .!,,,.. .!......!!, ,., ,,, ,, ,,, ,,, ,,, ,,,
.,, ,, . ...... ......... !.,, ,!, !,,.,,

,r:
.
(u
LA
.?-!
VI
cb@
Z
~ ~~
/ . m
0?
8$ ,!-!

8g
m ~
IL w
%g
II w
Y
;i$
8

J
g
K.

.,,
,,,
,.,
,.
,,,
,,,
,.,.,,
,,
&

I
CA
V*
... .......,,,,,. .,,.,,,,,,,!,
,,,
,,
..,,.,.
,.,,,,,.,,,,,
,, ...,.,,,,.,,,,,,,,,,, .n.2,,.,,,,,,
d)
,-m $
- 2 <T / P.

I
g; /#---
~ /
%@
1(+w

. ,.,,,,,,,,,
,,, d

: -1:
!,,,,,,,,,
,,,
,,,
,,,
,,,
,.,
,
o
1
!,,,,!, ,,, .,,,,,, .,, ,,, ,,, ,,, ,,, ,,, ,,, ,

.
Q
~n
%.4
c)

~mlll
l-
4 Q&
QCJ
W)G
II w
0
-4
Oawq (-4 Iww)w (4 +U2 -$
q
% q)
CJ ~ ~

!sd/u/tiJs dV/[)
. 0 ~(jo 1000 1500 2000 2500 3000
120 I 1 1 I I
~ 120
~ p=+()!ltl psia ~

j)o-. .. ..-..--- .-.-.-_______

............. ........ ............... ..... ....... ......... .................. ........................... . .. .....
.. .... ......_.-...-.__

~ 80- .---.
-+..._-._-
..._.
-.....
-....-.-+
....................................+.............................

2 40 ............. ....-
.-_.
.&______

q=50 S?B/l) ~?.~z~ p~a ! pWf=15U0 psia ~

@W;] ; (ficw 2) ~ (fl~v~ @ ;
i \ i 1 i
o 500 1000 1500 2000 2500 3000
L (=Q/q),days

F@ITe 5- ~afi~sja~ plot of Ap/q versus ta for the liquid simulation cases.

\
 ,,,,,,,,.,..,,,,,,,,
i

., !,,,,,,,,,,,,

~ ,, .,,,,,,,,,,,, w

l,.

o
?m7
to

%)

,0--
{
(J

--23 .,,,.,,
II..........................,, !!,.

cd
.

K
.

%
cl.

% . .,, ,.,,,,,,,,,,,,,. .,,,,.,,,,,,,,,,,, ............ ,,,,,,.,

! . . /
C2

- r-

-t)
F*

.-

0
I

.-

/
III
1
I Jl
Ca I
L

Ww .$
(-4
7
J?J*~ J
m
~
o ,
C3 b o
$5

c
-&)

!.-! 10

,,,,,,,,,,.,.............,,,,,,,,,....,,
t ,,
...,,,,,,, ... .. ....,,,,,,,,,,,,,,,. - o
~

m
s
t
00

l-f-l
QWw
1 1 I I
C-4
111
IWu) 1 r 1
w
1 I
et I ne
>Wv
1 i I I
C-4
a
o G
b
r+

cI/LnswsLl I)/dv
 sPE 21513
.

10 101 102 103 104

*
d---
6

1 n 1 1 11
q7J01

4
1

2
{
--------------------------------s,
----
-
"---""-"-*.-
..------------------------------------------------------------
-----------------------------
100
6 / 4
4
\
Bmmdaiiy Dominated Flow Solution
@APa=(l/ba) El:p[-(rnJbJtcP,.l \:
-). or 2
@Pa=[% 1.,,, + baJ-l
............ .... .............................+....
..... ................................ ................................................... ....................
..........
........
10-]
6 Results of Gas ~aterial J3alance Iterative
4- ~
solution ( 131asinmrne and Lee8~ f
ma =2.05536x10-3 Psi/MSCF~~
2- ba =1.3094 @./lvlSCF/D 2
G=2.6281 Bscf [tW,.= (bJmJ ~[1 + (m@.J&J I
~o-2
i 1 # I1#I#J
10-2
10 101 102 4
103 10
t.P,a, days

Figure 9- Plot of q/Apa versus tCP,afor d~.ta of Fetkovich, et aL2

t~P,afunction computed using IC,,
. and the equation shown.
 r
Ill 111
102 I I I 3 I 1 11 I a I I ! I I 1 I t f a 1 1 t I I I a I 9 I I I
u .
102

101
r=k= 1
.......................... ..8.
.r.......
1
10

4
10
JII 1 1
.. .... ........ ........ ..........
* 1 r~
: 102
... . ... .......
1 1

Mat~h #oint
1

(@A)m; = 0.78 MSCF/Dfi i

1 1:1
; 103
::
111 1 1 1 I

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
104
10]
...-................................. @

(qdD)mp ~ = 1-0
2
(~q,~w 630days : =
(td~):np ; = l.~

1......~~ 10
------rJrw : = 20.0
10
e
W
.................................+.....
...................................... ................................! 1 no
6j

4-

2-

.- 1 .. . ..... . .. ... . ... ..... . ... ... ...... .. .. ... ... ....... ........ ....... . .....
1
10-] .........................<6............ ..8.. .... ........ . .... ...... ...... ............ ... ...... . ....... . ..... ....?......
6
g
~ 4

z
; 2

5
m 2
~o-2 1 L -l?~ I # 8 I 1. t 1(
a n 1 4 13if 1 8 1 8 x JI I &I Iv1II
.Iu
.A
-loi ,1
io4 10- tm,a,days 10-L 10- 10u 1011 102

Figure 10- Type curve match of q/Apa versus tCP,,for data from Fetkovich, et al.2
Fetkovich type curve for radial flow.