You are on page 1of 17

Analysis of Variable-Rate

Well-Test Pressure Data


Using DuhamePs Principle
L.G. Thompson, * U. of Tulsa
A.C. Reynolds, SPE, U. of Tulsa

<, .-
JP& 130%0
Srmmary. ‘Ilk work considers the use of Dnhamel’s principle to analyze pressure drawdown and buildup
data when bnth bottomhole pressures (BHP’s) and sandface flow rates are available. This analysis procedure
uses Duhamel’s principle to convert pressure data obtained when the sandface rate is variable to the equivalent
pressure data that would be obtsined for a constant ssndface rate of production. The equivalent constant-rate
messure data then can be analvzed with standard mocedures (senriloc analvsis or twse-curve matcbiru z).

Introduction
Duharnel’s principle was introduced to the petroleum en- For example, Method 3 is preferable to other methods
gineering literature in the classic paper of van Everdm- when reservoir limit testing is conducted under variable-
gen and Hnrst. 1 They wed Dnharrrel’s principle to obtain flow-rate conditions.
the dimensionless wellbnre pressure-drop solution fnr a McEdwsrds9 hsa also presented an analysis procedure
continuously varying flow rate. Tlk4r solution is pre3entcd bsaed on theoretical concepts similar to those used in Refa.
in terms of a convolution integral. The well-larown Odeh- 1 through 8 snd in this work. However, hk procedure
Jones nrethodz arrd the more recent methnda of S01iman3 is radkally different because he used the convolution
and Stewart et al. 4 can be derived directly from the van integral 1 to calculate the theoretical pressure drops that
Everdmgen mrd Hurst solution by appropriate numerical shonld occur accordkg to the obsewed flow rate bistn-
integration procedures. The methods of Refs. 2, 3, and ry. This calcnlarion requires xrr initial estimate of reser-
4 nre restricted because their theoretical basis rests on the voir properties. The cwirnates of reservoir propcrdes are
aammption that the dimensionless prsssure drop term that then refmsd by arr iterative procedure that minimizes the
appears in the convolution integral is given by the semi- sum of the squares of the relative dtierences between the
log equationz.3 or the exponential integr3.14-i.e., the observed xnd calculated pressure changes. As in Refs. 2
methcds assnrne that if production were at a conxtarrtsand- through 4, McEdwsrds states that the pressurs-drop term
face rate, the dirnensiordess wellbore prsssure drop would that appears in the. convolution integral is given by the
be giveri by the sum of the line source sOhrtiDnand the line-source solution.
skin factor. Thus, at least theoretically, their metJroda do The derivations of Methods 1 and 2 and some numeri-
not apply to frachrrcd-well problems or to heterogene- cal analysis considerations perdnent to their application
ous problems such as naturally fractured reservoirs. The have been presented previously.’0 Brief details on
methods we consider are general and are not restricted Method 3, are given in Appendix B. Fnrther thcnreticxl
to plane radial flow problems. detaifs on all three methods can bc found in Ref. 11.
The methods of Refs. 2, 3, and 4 rely on redefining This paper (1) illustrates the applicability of our tech-
the dnre scale. The methods used in this work rely on Du- niques to a wide vsriety of problems, (2) presents a&-
hamel’s principle to convert variable-sandface-rate prca- tailed and carefid comparison between our s.rwlysis
sure data to the equivalent pressure data that would have teci-niques and the rate-nomrslized procedure, 12-14 (3)
bar obtained if production had been at a constsm sand- presents a new proccchrre (Method 3) snd illustrates its
face rate. Our methods are more general than those applicabiihy to reservoir limit testing, and (4) dkcusses
presented by others but the basic idea is not novel and how to apply our method to actual daci (Appendices A
has been considered previously in Refs. 5 through 8. In and B) and considers the srralysis of a field test. —.
fact, our Method 1 (discussed in Appendix A) includes In thk work, we generally use well-known mralytical
as a special case the metkod of Kucuk and Ayestars.n,8 solutions for the theoretical problem considered when
and our Methnd 2 (also discusssd in Appendix A) includes thsse analytical solutions are available. The analytical so-
ax a special case the method of Bostic et al. 6 To the best lutions are avsilable in Ls lace space and am irrvertcdwith
of our knowledge, Methnd 3 (discnssed in Appendix B) the Stehfest aknrithnr. 12 For unfractured wells rrrnduc-
has not been considered previously in any form. Method, ing kiyercd reservoirs, we gene.mte the necessti sOlu-
3 baa advantages over the other two methods when the tions using a finite-difference model that has been
late-time pressure &ta arc influenced by boundary effects. &c”ssd elxewhere. 16.17For fractured wells producing
cmrminglcd layered reservoirs, we generate the patinent
. Nw, ‘NM Rc@moResearti!nsl. solutions using a finite-difference model that is discussed
Copyright1sss S=xW ofPwole.n! Enginem in Refs; 18 and 19.

SPE Formation Evaluation, October 1986 4s3


Definitions For the mtumliy fractured reservoir problem conaid-
eredlatr.r, khin Eq. 1 should bereplaced bykflj,whe~
Many different problems are considered in tlrk paper, and
kpfi denotes the flow capacity oftbefracmre system.
tbe specific definitions of dimensionless pressure and
Moreover, in Eqs. 3rhrough 5,+, cf, andkshouldbe
dimensiordess time vary from problem t6 problem. All
replaced respectively by+f, c~, mdkf, where the sub-
definitions given here are in oilfield units, and all defini-
script f denotes the fracture system. Fer example, kf
tionsaasume single-phase flow of aslightiy compressi-
denotes the penneabtity (acmal, not bulk) of the fracture
ble liquid of constant viscosity. The analogous definitions
system, see Ref. 23.
for gas wells are well known. 20-22In thk paper, p WDal-
For tie two-layer, commingled reservoir (unfracmred-
ways denotes the dimensionless wellbore pressure-drop
weu) prob~m considered, k and 41ct sho~d be wlacx
solution of the variable-rate problem mtder consideration
by k_md @c,, respectively, in Eqs. 1, 3, and 4. Here k
andp~~ denotes the dnensionless wellbore pressure-
and g5c,denote tbickneas-averaged values detined reapec-
drop solution that would be obtained for the same prob-
tk’ely by
lemunder constant-producing-rate condltiom. All rates
used in the equations are sandface rates. In terms of
dlmemionless variables, the objective of the methods ~=kLh1+k2h2
. . . . . . ... . . . . . . . . . . . . . . ...(8) ‘-”
presented is to compute p ~CD given p WD and the dimen- h
sionless sandface rates.
and
Pressure Drawdown Proble~. For radial-flow prob-
lem: in single-layer reservoirs, the dimensionless well-
@lc,lhl+@zcnhz
bore pressure drop and dimensionless sandface rate are 6,= . ................ (9)
defined respectively by h

kh(p; –pwf) where throughout, the subscript 1 refers to Layer 1 and


pwD= . . . . . . . . . . . . .: . . . . . . . ...(1) the subscript 2 refers to Layer 2.
141.2qrp The deftiticms given by F@. 10 through 15 pertain to
and only the two-layer fractured-well probIem considered in
this paper. For this problem, the concept of reservok con-
qD=q(t)/qr, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2) d“ctivi~ i; fiP*t. MW The layer reservoir cmrduc-
tivitiea are defined by
where q, is any reference rate, iu reservoir bamek per
day. [Inthis paper, q(t) always represent tbesmdface
rate in reservoir barrels per day.] The dnensionless time
based on welfbore radius and the dimensionless tine based
on drainage area are defined respectively by
ROD=: J v
kj~jc~
k.$c,
. . . . . . . . . . . . . . . . . . . . . (lo)

for j= 1,2, where j is the layer index and ~ and @&tsre


given by Eqs. 8 and 9, respectively. The dimensionless
2.637 x10-4kt time based on effective fracmre half-length is defined by
~D= . . . . . . . . . . . . . . . . . . . ...(3)
cjctpr~
and —2.637 X 10 “~t . . . . . . . . . . . . . . . . . . . (11)
‘% – ~,~:,
2.637 x10-4kt
2DA= . . . .. . . . . . . . . . . . . . . . ...(4)
$c,pA Z., is the effective fracture half-leWL given by the fol-
lowing equation
The dimensionless wellbore storage constant is defined by
Lx, =R,DILxfl,+R,mL,n, . . . . . . . . . . . . . . . ..(12)
5.615C
cD=— . . . . . ..... . ......... .. (5) where L. , j= 1,2, is the fracture half-length in Layer
Zmjcthr; “ j. Sbnil@, we &fine the dimensionless times and dinten-
siorrless fracture conductivities based on individual layer
For a vertically fractured well producing a single-layer properties by
reservoir, the dimensionless time based on fracture half-
length and the dimensionless storage constant are defined
—2.637 x10-4kjt ................ (13)
respectively by
‘% – $jcti&j

—2.637x 10 “kr ... . . . . . . . . . . . . . . . . . ...(6) and


‘D’f– .+c#Lij
, kfibj
— ... . . . . . . . . . . . . . . . . . . . . . . . ...(14)
and cfDj = k ,L
J .%
5.615C
—— . . . . . . . . . . . . . . . . . . . . . . . 69 for j= 1,2, where kfibj denotes the conductivity of Frac-
CDXf– 2m$c1L%fh ture Layer j. Here Fracture Layer j denotes the part of

454 SPE Formation Evaluation, October 1986 .


2
, ; 26
0.0s-

!3
0.06- ‘3
$ 20 -
&
~ 18-
z
0.04- Y 16-
S.10
~
~ 14-
$ ,2 1 ! 1 ! !
0.02. 1 ! , ,
, ~7 z I(Y 103 10* 105 Iti ,,37 ye
1(7 1(Y I& Ios ]&
DIMENSIONLESS TIME, to DIMENSIONLESS TIME, tD

Fig. l—Rate resp.mme for constant-pressure production. ~9. Z+nstant-rste 201UN0nfrOnI Methods 1,2, ands.

the vertical fracture that is adjacent to Reservoir Layer and


j. For this problem, there is no crossflow between fr&-
ture layers or betwsen reservoir layers. In a practical kh(pw$-p.fi)
sense, this means tbst the two layers of the reservoir me PDs = . .................... (18)
separated by m impermeable zone and have been stimu- 141.2q,#
lated individually. Finally, we define the equivalent
dimensionless frscture conductivity by rw=tiv+, where P., denotes shut-in pressure and p@
denotes the flowing wellbore pressure at the instant of
2 2 shut~n. For layered reservoir problems, kh is repIa@d
(?J kfibjhjR=jD by kk in f?+. 17 and 18. The dimensimdms ch~ge in
.=1 ) sandfsce rate during shut-in is denoted by AqD snd is de-
cm= . . . . . ..’ . . . . ..(15)
m%, fined by

The motivations for the definitions given by Eqs. 10 q(At=O) –q(At)


through 15 sre discussed in detail in Ref. 24. AqD = . . . . . . . . . . . . . . .. . . ..(19)
q, . . . .
For all drawdown problems considered, the rate-
nomralized dimensionless pressure drop is denoted by
The rate-normalized dimensionless buildup pressure is
pWD/qD and is given by
given by pD,/AqD. From E+. 18 and 19, it follows thst

pwD ~(Pf– PI@)


—. . ...... . . . . . . . . . . ..(16) kk(pw, –p~)
~Ds
—=
qD 141.2q(t)p . . . . . . . . ..(20)
AqD 141.2[q(At=O) –q(At)]p
(See E@, 1 and ~.) For layered reservoir problems, k must
be replaced by k in Eq. 16. Note that Eq. 16 represents (Rx layered reservoir problems, replace k by ~ in
the dimensionless form of the actual pressure drop Ea. 20.)
(P: –PM) divided by the ssndface flow rate, q(t). he &mensiordess pressnres defined in this section
(I@. 1, 17, and 18) represent the dimensionless solution
BrdIdup Problem. For buildup problems, tp &notes the to the varisble sandfacerate problem under consideration,
producing time before shut-in nnd & denotes the shut-in snd p ~ represents the dimensionless constsnt-sandface-
time. The various dimensionless producing times are then ratc solution for the corresponding problem. Given
given by Eqs. 3,4,6 (or 11), and 13 with t replaced by PWD(PDS for buildup) ~d qD(AqD for b~dup), PWD
tp and are denoted by ,tPD, tPDA,t D. , and tPD~~. S~- is obtsined by the procedures discussed in Appendices A
gf. .
lady, the various dimensionless s ut-m tunes sre given snd B. Regardless of whether p.@ is generstsd from
by Eqs. 3,4,6 (or 11), and 13 with t replaced by At and buildup &tam- drawdown &ta, pWcD afways represents —
sre denoted, by AtD, AtD~, AtDXL,.~d AtD= . The per- ~ Wlvdent cOm@t-mW drawdown solution.
meability and perosity/compresslbfity prodfuct used in
these definitions must be adjusted according to the prob- Results
lem under consideration as in the drawdown case. Here, we show thst given the dimensionless drawdown
For buildup, we define the dimensionless shut-in pres- pressure response pWD(pD, for buildup) resulting from
SURS, PD, and ~~, , by the known variable sandface rate qD, we can use the
metbnds of Appendices A and B to compute pW@, where
kh(p~-pw.) P WCDrepresents the e@vnlent dimensionless pressure
pDs = . . . . . . . . ., . . . . . . . . . . . . . (1?) response thst wordd have been obtained for a constant
141.2q,.u
ssndface rste of production. Throughout this section,
SPEFomtio. Evaluation,October19s6 455
-S
g J?,o., _-
1
0
!d g
j . ~g
g
o PRESSURE DROP
~!5 ❑ SANDFACE RATE
g<
~ g 10-3 -
z~
PRESSURE DROP ●
am
Mam
WEUJ jow
go~ 25 _
. g;
gE * m g 105 -
Ij$ ~g DRAWDOWN CD = 105
. J
=W BUILDUP CD= 102
~ 521 - 02
?5-2 S=IO
Zz ?6Q
WE z
Xa w
z ,.-7 1 1 1 1 !
E ,.2 ,.4 ,06

DIMENSIONLESS TIME; to X JO-7 DIMENSIONLESS TIME, tD

F@. 3—Rate-normalized solution and sOlutiOn frOrtI Fig. 4—Pressure and rate responses for a DST, log-log
Methods 1, 2, and 3. plot.

Method 1 refers to the dimensionless version of Eq. A4 snd at late times (tD >~x 106 in Fig. 1), the rate
when drawdown data are considered, and to the dimen- response (qD/p WD) is governed by exponential decline.
sionless version of Eq. A-9 when buildup data are con- Throughout fis work, T denotes Euler’s constant,
sidered. Method 2 is given by the dimensionless T= O.57722. Fig. 2 presents a SefilOg plOt Ofp WD VS.
eqnivdents of Eqs. A-12 and A-15 for dtawdown and tD where p ~ baa been computed in three diffetent ways
buildup, respectively. Method 3 is given by the dimem (fvfethnda 1, 2, and 3). The solid curve represents the
sionfess version, of Eq. B-1. The buildup version of equivalent mnstant-sandface-rme solution and we note that
Method 3 can be found in Ref. 11. See Appendices A and pti matches this solution at all times regardless of the
B for a discussion of when and how to apply the various method used to generate p ~. For transient flow, we
methods. For all problems considered, the applicab~lty shmdd obtain
of the rate-normtilzed analysis procedure is con-
sidered. 12-14 PWCD=l.151 10g[(4tD)/ey]+S. . . . . . . . . . . . . .(23)

Unfractstced WefIa. AU problems considered in this sec- Fig. 2 indicates that all methods work well. By using
tion assume a wefl located in the center of a cylindrical regression analysis to fit a semilo straight line tO the
reservoir. i
PWD W. tD dat8 for 102< tD <10 , we determined the
It is now popular to coaaider well-testing problems following values. For Metbnds 1 and 2, dope= 1.148 and
where production is at a constant BHP—e. g., see Refs. S= 10.O6; and for Method 3, slope= 1.149 ands= 10.09.
25 tbrouzh 27. The first mobIem considered assnmes Thus dl three procedures yield exccUent approximations
constant-~ressure productk% at a well lncated in a bound- to the correct values, slope= 1.151 and s= 10.0.
cd cylindrical reservoir-i. e., the same problem consid- Fig. 3 depicts a Cattesian plot ofp WCDvs. tD and
ered in Refs. 25 through 27. It is importrmt to note that p ~tJ4D W.. tD. Tbk is a dimensionless version of the
the right side of the rate-normalized equation (Eq. 16) is type of plot used to estimate reservoir PV in teservoir
referred to m l/qD in Refs. 25 through 27. limit wting.zs,zg During pseudosteady-state flow, the
Fig. 1 t’C3p1&3ntSa Setibg plot of qD/p@ vs. ?D fOr constant-sand face-production solution is given by
the constant-pressure production problem wheres= 10 and
reD ‘4,000. Here, s representa the skin factnr and rcD PWCD ‘2mtDA +h(reD)-o.75+S
denotes the dimensiotdess drninage mdi~ and is defined
by
=>+ln(r~)-0.75+s. . . . . . . . . . . . ..(24)
. . . . . . . . . . . . . . . . . . . . . . . . . . . ..(21) r eD
r~ =relrw.
mere the drainage area ia given by A=rr~D and thus
As is well known, ‘-27 the trmsient portion of the data tD.4 ‘rDh~~); s= f%s. 3 ~d 4.1 AS is we~ ~Own, 29
approxnnately satisfies the semifog equation—i. e., pseudosteady-sfsde flow @q. 24) begins at tD~ =0.1,
which corresponds to tD =5 X 106 fnr the problem con-
hD@D=l.151 10g[(4tD)/e7]+S, . . . . . . . . . . .(22) sidered.

456 SPE Formation Evaluation, October 1986


1
on 0.0 <“ METHOD
s ,&—-——=——
——
~ 17- o,l, e.1
!g= . m2, e.1
$?
s % . Rm N0Rf4ALEE0
-1 ZW PRESSURE DROP
g ?!j ,5 -
& ~
. . &!
2$.
g= 0 DRAWDOWN
CC,.105 ,“
Wul,s -
22 B
%f s
E~
=.om 1.2
0
, , I I
= II
00 Wxlo+ LOxl& I,sxlos 2.OX1O5 10 ,.2 I& I@
DIMENSIONLESS
TIME, to DIMEN:%NLESS %E, tD
-.. .
‘lg. 5—Pressura and rate responses fOr a DST, GxIesian Fig. 6—Methods 1 and 2 rata-normalized solufions DST
Olot data.

when regression an~vsis is used to fit a $might lime of tbe flowing -. ncriud and tn =106 at the end of the buifd-
of the form given by E.q. 24 tfuougb the p .,D or p .D/qD up test. In Eq. 25, p ~ represents the wellbore
late-dme &ta of Fig. 3, snd rm is cslctdatod from the pressure-i.e., the flowing wellbure pressure for draw-
slope of the line (s10pe=2/r2~), the fol.lowiug results are duwn and the shut-in wellbore pressure fur buildup. Simi-
obtained. Using &ta fmm tDA= 0.1 to tDA= 0.5 yields lsrly, q(t) in Eq. 26 represents the sandface flow rate in
reD =3985 for Method 1 or Method 2 pWCD &ta, reservoir bsrrels per&y butb duriug rkawdowu and build-
reD ‘3993 fOr Method 3 P WCDcL@ snd uP. Fur the res~~ sf30wIIin Fig. 4,s= 10 and the dirnen-
reD=3813 for rate-nor’maliied da@.. siordess weUbore storage constsrrt is CD= 105 for
Using all data for tD~ >0.1 yields drawdown and CD= 102 for buildup; During the draw-
reD ‘3975 fi0rr3 Nfethorf 1 m’ Method 2 p WCDdats, down pctiod, p ~D decreases (PW incr~scs) as a result.
ro ‘3997 for Method 3 Pd data, and uf the increase in the fluid head, but this behsvior is not
reD = 3354 for rate-nortnaliied dsts. clearly ideudtiable on Fig. 4 because of the mture of log-
From these results and additional resufts given in Ref. log plots. The decrease inp ~D during drawdown can be
11, we conclude that Method 3 provides the best unalysis seen inure cleatly in Fig. 5, which represents a Cartesi-
prucerfure when,pressuze data are influenced by reservoir ~PIOt OfP@ ~d qn W-. t~ for 0st~s2X105. The
boundaries. Mormver, rate normalization is not an ef- sandface rate q~ decreams throughout the total test.
fective procedure for the nmlysis of &ta strongly influ- Fig. 6 is a semilog plot ofp ~CD vs. tD, where pW@
enced by reservoir boundaries. is generated from the pressurekste data of Fig. 4 witfr
We now cmtsider a drilfstcm rest (DST) problem. 3@32 the dimensionless equivalents of Method 1 (&L A4) and
We consider only one flow period fo310wedby one sfmt- Method 2 @q. A-12). Note that, in using rfrese methods,
~ ~ri~. The flowing psriud reprcsenta the slug test pru~ we have used txztbthe drawdown and buildup data uf Fig.
lem considered in Refs. 31 and 32. To conform to stsn- 6 to obtain tJzeequivalent constant-sandface-rate dimen-
dard notatiun, we,bave defined the dimensionless pressure sionless pressure drop for a total flowing time of
change by tD = 106; scc Ap~ndlx A. The solid line in Fig. 6 rcp-
resenrs the constant-rate solution we should obtsin from
our methuda and is given by Eq. 23. As shown in Fig.
Pi–Puf
pwD=— . ... . . . . . . . . . . . . . . . . . . . . . . . . (25) 6, Methods 1 and 2 generate the correct p ~cD solution
Pi -Po very accurately. Using regression airslysis to fit a semi-
log straight line to the generated p .,D vs. tD datn of Fig.
where pi denotes the initial reservoir pressure and PO 6 yields the following results.
denotes the welfbore pressure at time zeru—i.e., p. is For Method 1, slope= l.165 and s=1O.O1, and for
the pressure iu the DST tucdjust before the beginning of Method 2, alopc= 1.158 ands= 10.03. These values are
the flow period. Simifnrly, we define the dimensionless in goud agreement with the correct values, slope=1.151
sandface rate by snd S= 10.0. Note that we have not applied Method 3 be-
cause Method 3 wifl encounter diftkxdties when sharp —.
141.2q(t)Jl changes in rate or pressure occur (see Ref. 11), and for
qD=
. . . . . .. . . . . . . . . . ... . . . ..(26) this prublem, the rate snd pressure charge abruptly at
kh(p~ -p~) JD ‘tpD = @ (~= W 4). Fig. 6 ~SO shOwstie dimen-
sionless rate-normalized pressure drop p .D /qD for
In this paper, the deftitions given by F@. 25 and 26 arc tD < tpD. ‘fhe mtC nu2MdiZ.StiOn~Ot t% Wntirtutd intD
used ordy for the DST problem. the buildup-i.e., we cannot view the buildup as a con-
Fig. 4 presents a log-log plot of p ~D (Eq. 25) and qD dnustion of a variable-raie drawdown test when using rate
~q- 26] VS. ;D. ‘fbe results of Fig. 4 represent a flow- nunnalization. Ir3fact, v@ues of pWD(fD)/gD (fD) we uff
ing period (producing time, tPD= 105) fullowed by a the acsfe @eater tbrm 18) uf Fig. 6 for tD >1 x 105.
buifdup period of duration AtD =9 x 10s. Here CDrep Ws is not surprising in view of the pressurekate data
resents the total time of the test su fD = 105 at the end abuwn in Fig. 4.) The rate-normalized dmwdown &ta dis-
SPE Fomtim Bvaluarim,October1986 4s7
.
$ I I I I 1

g $ =,0
2- n
~ w“ 3,5
!5! ~~ b=O.4
2
al 07
o? SLOPE .1.303 O-Jw
!s0
%6 - ~~
0. n [0-’ -
z ~~
7 0 (,7
m!xm
5 -JOm
g 10 -
$:
al
g g Ki3 -
g ~al
AZ
‘z *W
g 14 - 02 0 PRESSURE DROP, pwo _
CD 50 ❑ SANDFACE RATE, q.
z 5
% z
165
E E ,.3 5
I I r u 10 107
18
DIMENSIONLESS T~iE, to
1 KY
l+O&R TIME !:TIO, (tp+4t) /At
Fig. 8—Pressure and ratw partially penetrating well with
Fig. 7—Homer analysis fir a DST. wellbore storage.

play a semilog straight line with slope= 1.157 and yield and
s= 10.06. As in previous cases, these slopes srrd skin-
factor vslucs were obtained with res!ression analvsis. fir
Fig. 7 presents the ststrdard dimen~onless Hom~r plot fzWD=~@ . . . . . .. . . . . . .. .. .. . . . . . . . . (28)
of the buildup data of Fig. 4. In a nomrmlDST, the dura- ~w z
tion of the buildup period will be no greater tbaa twice
the duration of tie preceding flowing period. Thus, in our Here,
analysis of the data of Fig. 7, we use only dats corre- h. = Iengtb of the open interd (perforated
SpOn&IIg to sfmt-in dares such dtst AtDs 2tPD=2 X 105. interval) of the wellbore,
Thk msam thst the minimum vrdue of the Homer time k = horizontal perrnsabfity, and”
rstio med in the anslysis is (tP +At)/At= 1.5. BY ~sing k= = vertical pernreabJIty.
r%wssion analysis, we found fhat the data of Fig. 7 that
currespard to the intsrval 1.5< (tP +At)/Af< 5 can bc tit
Fig. 8 presents log-log graphs of the dimensionless
with a semilog straight line. However, the slope of this
drawdown-pressure drop snd dimensionless rate obtabrsd
semilog straight line is 1.303 snd the skin fsctor estimat-
for a partisl penetration problem. The open interval of
ed from this semilog line is 8.97. These vafucs differ by
the weUbore is adjacent to the top hW feet of the forrrra-
more thsn 10 % from the correct values, slope=1.151 and
s= 10.O. ThIS indicates that the conventional Sndysis Of tion, and CD=105, hWD=103, s=5, snd b=O.4. Note
that all pressure data shown are influenced by weUbore
DST buildup dats can yield erroneous estimates of the
storage effects.
flow capacity and skin factor if buildup data are strongly
Fig. 9 presents a senrilog plot of the dimensionless rate-
influenced by wellbore storage effects. The resalts ob-
nornrdized pressure &ta, arrd pWcD generatsd by Method
tained would be even less accurate if the drrrstion of the
1. The solid curve in Fig. 9 is the constant-sandface-rats
shut-in were decreased.
(CLI=0) dinrensionkss pressure solntion for the same
Difficulties in Homer snalysis sire of course not re-
partial penetration prublem. For the case where CD =0,
stricted to DST data and can occur in conventional tssts
it is well !ctrovmw that at early times the fOUOwing5cmi-
if the buildup test is not of sufficient duration to obtsin
log equstion boldx.
buildup data that are not influenced by wellbore storsge
effects.
We next consider a partially penetmtin well located
f
in the center of a cylindrical reservoir. 17,3’35This is the
only prublem considered in this paper where the opsn irr-
-=(?)d%)+’~,v’) -
terwd of the wellbore is not equal to the reservoir drick-
snd at Iste times (pseudoradisl flow period), P IWD is
ness.’ The penetration ratio, b, and the dimensionless
given by
weflbere length sre defined respectively by

4t~

b=hwlh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..(27)
pWCD=l-151 log
() ~ +Sib+S~, . . . . . . . . . (30)

45s SPEFomtion Evaluation,October1986


!6B , I t 1 I ( 1 , I , (

. ANALYTICAL sOLUTION, CD
=~ - 0 KY
~ ~~ —0
d-
3!3 SLOPE .0.59

0
a o
s ,,,~ ,.7
0
I&
! f
lCP
( ,.0 I
I@
I
1o11 ,~,.

DIMENSIONLESSTIME, tD DIMENSIONLESS mME. ID


,.
Fig. 9-Rate-normalized and Method 1 solutionq partial. Fig. 10—Prasaure responaq naturally fracturad resermir
[y penetrating well. with wellbore storage.

where Sb is the pseudoskin factor resulting from partial In recent years, an intrnse research effort has been
~ne~ation, 17,33.35The pWcDvslues (open, circular&m devoted to the study of naturally fractured reservoirs. It
points in Fig. 9) constructed with Method 1 correlate well haa been shown that in an ~nits nsrordly fractured reser-
with the constant-sandface-rste solution (solid curve). If voir, a semifog plot of the wellbore pressure res nse vs.
we fit these p .,D data with a semilog straight line for time may exhiih three semilog straight lines.’, P7.38Ref.
102s tDs 2 X 103 by regression analysis, the semilog 39 examines the conditions under which afl these senri-
straight line obtsim?dhas alops=2.M7 and yields s=5.06. Iog straight limes will exist when the reservoir is bound-
The correct vslues are slope= 2.878 (see Eq. 29) and ed. Refa. 38 ad 40 consider the effect of wellbore sterage
s=5.o. Fhting the kte-time (3x I06 StDs 107) p.,= and skin on the pressure response but do not consider the
values with a senrilog straight lime usirrg regression soal- effect of the outer reservoir boundmy. By using the ana-
ysis yiehla dope= 1.151 and (s/b) +Sb =22.09. Using the lytical techniques of these studies, or by combting the
value s=5.06 obtainrd above and b=O.4 then gives sb = known solutions in an appropriate way, we found that it
9.44. The correct values are slope= 1.151, (s/b) +-sb is easy to construct the dimensionless wellbore pressure
=22.04, ad Sb =9.54; see Eq. 30 and Refs. 17, 34, response in a bounded, naturally fractured resexvoir when
and 35. early-time data are influenced by wellbore storage effects.
Nme shat we have not applied Meihod 2 or 3 to the data The solution to this problem can aho be obtained with
of Fig. 8. Method 2 has stability problems when qD is a tinite-difference model. 41
an increasing function of tD (see Appendix A and Ref. Fig. 10 presents a semilog plot of the dimensionless
10). Method 3 can be applied and does yield accurate re- welftmre pre.ssum drop (drawdown) vs. Slowing time, tD.
sults but offers no advantages over Method 1 for tmn- The solid curve repre.ventsthe solution for CD =0 and the
sient data (SW Appendix B). circulsr tits pointa represent the solution for CD= 108.
From the results of Fig. 9, we see that the rate- Here, the naturally fractured reservciii model of Ref. 23
norumliied dimensionless pressure drop falls below the is’’used. As discussed in the definition section, p ~D, tD,
correct constant-rate solution and canuot be used to ob- and CD are given in terms of fracture properties; for
tain bigbly accurate results. Thk behavior is typical for emmple,
wellbore-storage-dominsted problems. 10.11.36
The results of Fig. 9 offer m intereatiog poasibtify from 5.615C
the viewpoint of analyzing tield data. If wellbore storage CD= ... . . ... . . . . . . . . . . . . (31)
2~4~fhfir; ‘
effects are present, but sandface rate &m are obtained,
our remdta indicate tlmt Methcds 1 and 3 will yield equiva- where $ ~ denotes the perosity/compressibility product
lent ccmatant-samiface dats that can be analyzed by senri- for the 1 acture system snd hfi is the total thickness of
Iog methods. In particular, two semifog straight lines may the fracture system. Because CD is based on fracture
bs obtained. The flow capacity adjacent to the open in- propwtisa, CD= 108 is a reasonable value for the dimen-
tcrwd (kh~ for the problem considered here), k, and the sionless wellbore storage constant. For this problem, the
true skir3factor can be estimatsd ~m the early-time semi- dnensiodess fracture transfer coefficient, h‘, and the
log data. Using the late-time semdog data, the total flow dimensionless matrix storitivity, co’, are defined respec-
capacity (kk for the problem considered here) and the to- tively by
tsl skin factor SI [s, =Sb +(s/b)] can be estimated. The
value of Sb can then be estimatsd from the equation
$6 ‘s, – (sb). The penetration ratio b can bc obtained by
dividing kh. by kh. Because Sb isa unique function of
A’=12~
m ()
W
kfhf
.: . . . . . . . . . . . . . . . . . ..(32)

h .D and b, h. can then be estimated from Brons- snd


Martingcharts,8 aud then .WkZ and kz can be estimated
with the obvious rearrangement of Eq. 28. T&S type of ~,_ Omcmhm
analysis procedure can be extended to layered-reservoir (33)
problems.’7 $~fhf ’”.””..’..”””.””””’””””’””’”

SPE Fomtion Evaluation, October 19S6 459


*: 60 , ,,;,,,, I , t 1 -
<o‘6 , I I 1 k! kl/k2s0.1. (~c,l M+2C,21. 1, h,/b2 .1
. METHOD
[ ,$ 40 -,,.-, ,2.,0
. RATENORMALIZED
SRES3URE
DROP
2 Pit. soeo. P12= 3m0.mi
: 20 -
reo. 4000’
z
j. 0 -“.

%
m -20 -.
~

., .200 g
3
s g .So 1 1 r , 1
0
* K? W -
.
1$ I* I@ 105 IW 107 @
‘%lmfislew’%w, ,:” DIMENsIONLESSTIME, tD

Fig. 11 —Rate-normalized and Method 1 SolutlOnS, natUr- F@ 1S!-Pressure responsq two-layer, commingled
ally fractursd resewolr. reservoir.

whers the srrbscript~refers to the fracture system and the late-time semilog stmight lines by the proccdnre discussed
subscript m refers to the matrix system. See Ref, 23 for in Ref. 23, we obtain @‘=71 with the rate-norrnaliied
a more detniled discussion of these parameters and their data, whereas the s2me procedure applisd to the p..D
physical interpremtion. dats of Fig. 11 yields o’ = 196. Note that the comcct value
For the solution shown in Fig. 10, reD is sufficiently is u’ =2@3. Method 3 can also be applied to the CD= 108
large that the outer reservoir bormdary dces not affect the solution and yields reliible resrdfs, but Method 2 is not
wellbore respnnse during the time interval shown. Note apprOpriatS for wellbore storage problems.
that k’ =10–9, o’ =2(X), and s=O for the results shown The last problem considered here is a two-layer com-
irr Fig. 10. Also notethat dre C~=Osolution exhibitaall mingled rese~oir system (no crossflow between
three semilog strnightlines idenfitiedirr Refs. 23,37, and layera42,43).Fig.. 12 presents a semilog plot of the dimens-
38. In both Figs. 10 and 11, a cross is usexl’to derrotc the ionless drawdown prcssnre-dmp solution for a two-layer
begimdrrg and end of the various semilog straight lines. commingled reservoir problem. Here p WD and tD are
Note that the CD= 108 solution does not exlibb the based on average properties. Individual layer properties
early-time semilog straight line with slope= 1.15 and the or their ratios are given in Fig. 12, where the subscript
intermediate-time semilog straight line with slo =0.59 1 refers to Layer 1 and rhe subscripr 2 refers to Layer .—.,..
is obscured. (If CD were increased to CD= 10 r. for dns 2. The initial pressure in Layer 1 is 6,000 psi [41.4 MF?a]
problem, the intmmediate-dme semilog straight line would and tbe initial prcsaure in Layer 2 is 3,000 psi [20.7 MPa],
be totally obsmwed). and the indkidrral layer skin factors are s ~= –2 arrd
Fig. 11 prexmts the p .m fiction generated with our S2 = Ift thus, this problem is an exmeme case and the
Methnd 1 using the CD = 108 solution of Fig. 10 and the analogous field data wmdd be extremely difficrdt to arra-
corresponding dimensionless sandface rate, qD (not lyr.c by conventional methnda.43 The initial pressure used
shown). Fig. 11 also pmsenta the rate-normalized pres- in the definition ofp WD is
sure drop for the same problem. The solid curve of Fig.
11 represents the CD =0 solution of Fig. 10, which is the
dimensionless pressrrre drop solution obtained when the
shrrdfacc rate is constant. Note that the p ~cD solution . (34)
generated wifb Method 1 is in excellent agreement with
this constant-sandface-rate solution. By using regression
srralyais as in our previous examples, we found that the
PWCD function Of ‘lg. 11 disP1aYs~ e~IY-fie se~Og as auggestcd by Larsen, 43 brrt the definition of pi has no
stm@rt line with slops= 1.14, an irrtermedinte-dnre smrri- bearing on the semilog slope. For the data considered
Iog straight Iine with dope =0.59, md a late-rime semi- here, Eq. 34 yields pi =3,273 psi [22.6 MPa].
log straight line with slope= 1.15. The 12st,tw0 slopes are From the results of Fig. 12, we see that no semilog
the desired cor’rcct values but the esrly-dme slope (1.14) straight line with slope ?1.151 exists. frr fact, the mirri-
is slightly lower than the corrwt value of 1.15. mum semilog slope that can be obtained is approximate- ..
The norrrmliied pressure drop shown irr Fig. 11 falls Iy 2.8 (approximately 2.5 times the desircc! correct slope
below the constnrrt-rate solution fortDS5X1010. Any of 1.151), which is not surprising in view of Larsen’s43
reasorrsbly chosen irrterrncdiate-dme scmilog straight line results. The rcsuft.sof Fig. 12 are for a constant total sand-
will have a slope significantly greater than 0.59. By using face rate of production, and reservoir bomrdmies irrtlu-
re ression anslysis on the, rate-norrnslized dat2 for ence dre wellbore response at tD = 5 x 107. .For this
10% stD<Io5, we detemrined an early-time semilog problq, fluid is actually injected intDLayer 2 during vir-
atmight line with sIOps= 1. I 1. Aftbougb this slope is close tually the entire duration of the drawdown test. This is
tn the correct value (1.15), the line falls below the cor- a rmult of the low initial pressure of Layer 2.
rect early-time semilog strsight line, as shown imFlg. 10. Fig. 13 presents a semilog plot of the dimensionless
Estimating d‘ from the difference between the cdy- and pressure dropp .D1 vs. dimensionless time t~~. The rate
+50 SPE Formation Evaluation, Ocrokr 1986.
[B!’
Fig. 13—Pres8ure
properties.
‘lMENslONE=

and rate responses


I@
TIME,
’01
I@

basad on Layer 1
a

qD1 is ~~ shown. Here PwDI , tD1 and q~, are based transient flow. (The p ~ solutions also correlate well.
on Layer 1 properties are defined respectively by with the sin le-layer solution during pseudosteady flow,
tD >5 x 10 $’.) In particular, the pW=D and p .D1 /qDl vs.
k,h, (pi~ –pwf) tDI PIO~ ~ disP1aY a semilog stmight line with
PwDI = , . . . . . . . . . . . . . . . . .. (35) slope= 1.151, and anslysis of data yields an excellent ap-
141.2qr/t proximation for the skin factor of Layer 1, s, = –2.
(Data obtained for Layer 2 can also be &alyzcd by simi-
2.637 x10-4k1t lar procsdurei. 11)
t~, = , . . . . . . . . . . . . . . . . . ..(36) The results for the two-layer problem indicate that, if
@,c,, JLr; individual layer ssndface rate data can be obtained, then
individual layer properties can be obtained with a con-
ventional semilog analysis by Method 1, 2, or 3 or the
rate-normalized snalysis procedure. Fetkovich” clem-
ly indicates that it zhould be pmsible to estimate individual
ql ($
qDl=~, . . . . . . . . . . . . . . . . . . . . . . . . . ..(37) layer propemies by the rate-normalized amdysis proce-
dure when individual layer sandface rate data are availa-
ble. Thus, the preceding re.suhaon rate-normalization may
where q ~(t) k the sandface rate of prcducdon from Layer b viewed as a theoretical confmrion of hia hypothesis.
1 in reservoir bards per day. Fig. 14 presents a semilog -..
plot of the p ~cD solutions vs. tDl, where the p ~cD solu- Fractured Weffs. We consider the application of our
tions are generated by Methnda 1,2, and 3 using the pres- methods and the rate-norrmdizcd procedure to vertically
sure/rate data of Fig. 13. The rate-normalized fracmrcd well problems. We discuss uniform-flux and
dimensionless pressure drop, p ~Dl /qDl, is also shown infinite-conductivity fractures45 as wefl as tinite-
in Fig. 14. The solid curve is the Layer 1 pressure-drop conductivi~ fractures. 18.19,46.47For planar fractured
solution that is obtained when Layer 1 is produced at a wells (uniform flux or infnite conductivity) where well-
constant sandface rate. Note that the p ~CD solutions bore storage effects exist, the dimensionless wellbore
generated smdthe rate-normalized pressure-drop solution pressuredrop solution can be genemted with well-known
all correlate well with this single-layer solution during mPerPnsitiOn prOc~u~s—e. g., see Ref. 48. The finite-

Fig. 16—Rat~norrnaIized and Method 1 solution% infinite


conductivity fracture.
Fig. 15—Pressure and ratq infinite-conductivity fracture
with wellbore storage.

SPE FonnaticIn Evaluation, October 1986 ‘al


“\.. ,(’--
.s[0 1 I T
< E — G?NSTbNTRATESOLUTION,WUFORM FLUX
: . NoRMALIZEDPRESSURE,CD% -0.1 kl/kz=O.O1, hllh2 .2,
* z

k
&- p
0 (IDX,]U.0,07Z, (PW.)u-1.10 CfDj/QM50. Lxf,/Lxfz.4
10 “!
s 2:
%/Rcoz .2 ,CmP 1.6
~ %a
%, :u-
~g ,
k?
#1
E 0:
.
ZE
: $- ,.-l +1%(+2W =no
; x
z ii kfl b#kf2b2 .2
E
z ,.-1 1 ! LLL 162
,@ ,0.5 10
10-) I 10
DIMENSIO’8LESS sHUT-IN%’ME; At%
‘iMENslONLEss TIME, ‘Dxf

Fig. 17—Rate.normalized inf inlte-ccmductivify solution Fig. 1S—Buildup pressure response; Sinite. conductivity
matched with uniform-flux solution. Iractur* two-layer system.

conductivim fractured well with wellbore stors~e and skis iafinhe-conductivity scsfes, respectively; that is, the type-
problem his recently been solved aualydcaily ~y Lee and curve match sligns the point ZDX=0.072., p WD= 1.1 on
Brockenbrough. 49 Tecfilcally, their solution is an ap- the uniform-flux scale with tLe point tD1; = 10-1,
proximate solntion but correlates extremely well (for P@= 1.00 on the inffite-conductivity scale.
tD.rJ < 0~5) wifi *C rigOrOus sOhJtiOn Ob~nable by We now consider a vertically fractured wefl draining
f%ute-difference methuds. The results of Refs. 45 through a cnnuuingled two-layer resemoir. The fracture is of finite
49 sssume a single-layer re~rvoir. Refs. 18, 19, and 24 conductivity and there is no crossflow between the frac-
considered the performance of finite-conductivi~ verti- ture layers: no welIbore storage exists, fracture storage
cally fractured wells draiuing mrdtilayered reservoirs. effects sri negligible, snd the initial pressure is uniform
Fig. 15 presents a log-log plot of the dimensionless wefl- throughout the system. We discuss only the buildup
bore pressure drup snd dimensionless ssndface rate vs. response subsequent to a long producing period
dimensiouIess time, tDx,, for an infinite-conductivity (tPDx =2.5x10”) at a constant totsl sandface rate.
fractured well producing a single-layer reservoir. Well- Fig. {8 presents a log-log plot of the dimensionless build-
bore storage effects ~e present with CDzf=0.1 and the up response, PD,, vs. dimensionless shut-in time, AtDx,.
surfsc.e rate is constant. (See the definition section for the Here PD, snd AtLMrare based on aversge properties snd
definitions of the variables used here.) As shown in F,g. QrDl is based on Ore effective frscture half-length, ...
15, qD = 1 for fDzf >1. Here qD = 1 implies that the L.f. tslues of the pertinent variables are given in Figs.
surface and ssndface rates are equal. 18 and 19. As shown in Fig. 18, the dimensionless
Fig. 16 pressnts a log-log plot of the rate-nonnsked builduppressure. response (circulsr data poinrs) $urrelates
dinreusiordess pressure drop arid P.@ data obtained from extremely well with the dimensiordess drawdoyn-pressure
Method 1 with the pressure and rate data of Fig. 15. response (solid curve). This drawdown response is for
(Method 3 can afso be used to obtain p WCDbut Method an equivalent singfe-layer problem that is discussed in de-
2 is not appropriate.) The solid curve in Fig. 16 is the tail in Ref. 24. This aspect is not pertinent to our main
constsnt-smrdface-rate infmits-conductivity solutions theme that individual layer properties cm be determined
(CD=,=0). The resufts of Fig. 16 indicate that the PWD if individual layer sandface rates are known.
solution generated by Method 1 correlates extremely well In Fig. 20, PD,I and AqDl sre plotted vs. dimension-
with the constant-ssndfacc-rate solution (solid curve), less shut-in time, AtDxf,~pD.1, Aqm ad Atmo are
wheress the rate-normalized dlmensiordess pressure fails bssed on Layer 1 propermes. In particular, AqDl 1s de-
below the constant-sandface-rate solution for t~z,: 1. fined as
We were surprised to find that the rate-nornmhzed
dimensiordess pressure of Flg. 16 cmr be type-curve AqD, = ql(Ar=o)–ql (AO
mstched reasonably well with the uniform-flux fracture . . . . . . . . . . . . . . . . . (38)
solution during the time rauge 10’2s tDT~s 4. Thk is q,
illustrnfcd in Fig. 17, where the solid curve N the constsnt-
ssadface-rste wellbore pressure drop for a uniform-flux where q ~ (in reservoir barrels per dAy) represents the
fracture. The sofid, circubr dats points are the rate- sandface flow raft from Fracture Layer 1 during the build-
nornrti]zed pressure drop solution of Fig. 16 type-curve up period.
matched with the uniform-flux solution. Note that these Fig. 19 presents the rste-nommfizcd dimensionless prcs-
data points represent the rate-normalized pressure drop Sure, p~,~ /AqD1, and the dirIWXISiOIdSSS pressure p WCD

obtained from the infinite-conductivi~ fracture solution generated from the data of Fig. 20. Here Method 1 (Eq.
with wellbore storage effects, CDX,= 0.1. The scales in A-9) was used to generate p ~cDl und the subscript 1
Figs. 16 and 17 am the sarnc-i.e., both represent the denotes Layer 1 properties. The solid curve represents
scales uacd in the infiuite-conductivity sohrtion. The point the dimensionless drawdown wellbore pressure drop for
Iabcled with a cross represents the rustch point, with the the correspendmg single-layer problem. Both pwl sad
subscripts u sud i representing the. uniform-flux sad ~Lkl/AqIX corda~e we~ Withthis drawdown solution for
462 SPE Forrrdion EvahMion, Octob$r 19a6
m
10
+:=

cm?.%,
R,.o.cazmd.+,c. ‘3x10’P3i-’ .O
Z& ,
ma. .
yu-::
?g::: ,..,
:!
.= -+
:s O PRESSURE,E&,
o RATE, Aq,,
ii=
g,= 0 16;0+
DlMENS!&ES2 WJT+N TIME, At
10- IW3 ,& “Jr, 10
~fl DIMENSIONLESSsHuT-IN TIME, AtDx~,

Fig. 19-Correlation of rate-normalized and Method 1 so- ~9: 20—shut-in PreSSUre and rate based on Layer 1 prop-
lutions with Lsyer 1 solution. erties.

..
AjDXf<1, although Method 1 yields slightly better re- Here, f@) represents the pressure drop in pounds per
sufts. Buildup data for AtDx >1 fall below the drsw- square inch that would be obtained if production were at
down solution as a result o t producing-time effects. 50 a constant sandface rate q,. Eq. A-9 is used to generate
Results similar to those shown in Fig. 19 cm SISObe F’,rm %sa function of shut-in time so achmf knowledge
generated from Layer 2. data. The resufts of Fig. 19 il- of @(tj) aod q, is not necessmy. As dkcusse~ in AP-
lustrate that individual layer properties (k, L x{, ~d Cf) pendm A (also see Refs. 6, 10, and 11), ~~dtj) ~P~-
can b-sobked for simifsr field cases provided mdividual- sents the equivalent (constant-sandface-rate) normalized
layer sandface flow rate data are ms&urcd.This conclu- drawdown ‘pressure ‘change; thus, standard drawdown
sion applies both to drawdown and buildup data. 11 analysis procedures (type curves or semilog analyais) can
In general, we have found that when these methods are he used to anslyze the PFCN vs. & data.
used to convert (variable-ssndface-rate) pressure build- For the buildup test we consider, fhe rate-normafised
up data to equivalent constant-sandface-rate drawdown- buildup pressure is denoted by Ap/Aq and is given by
pressure data, the equivalent drawdown data generated
will match the desired constant-sandface-rate drawdown Ap _ Pws-PWf,
. (40)
solution provided that Ats O.Itp, where tp is the produ- ~–q(At=O)–q(At)’ ““””””’””””’”””’”
cingtime. This condition is only a rule of thumb. A more
detailed discussion of producing-time effects is given in where p ~$ is the shut-in pressure and p ~$ represents tbe
Ref. 11. , final flowing welfbore pressure-i. e., the pressure at the
instant of shut-in.
Field Example
Brownscombe Csse 5. The buildup test considered is
Here, Method 1 is applied to analyze an actual buildup
Brownscombe’s Case 5.’1 This is a mukipbase-flow
test where the buildup data are influenced by sftertlow
problem and hence the total sandface rate (q,), must be
(wellbore storage effects). We also discuss the rate nor-
malization procedure for this problem. used in the analysis. 14.m-54Fig. 21 presents the conven-
tional log-log plot of Ap vs. At and the log-log plot of
For buildup data analysis, the red variable version of
the chsnge in sandface rate, Aq,, vs. shut-in time, At.
Method 1 is given by Eq. A-9. Given actusl buifdup pres-
Here
sure snd sandface rate data, Eq. A-9 can be.used to gener-
ate the equivalent normalized drswdown response, pFm, ,AP=Pws-P@ . . . . . . . . . . . . . . . . . . . . . . . . . ..(41)
where

PFw(;j)=x!l
qr
.........................(39) ‘dAq, =q,(At=O)=q,(At), . . . . . . . . . . . . . . . . . ..(42)

> 10

.:$

D
MAT.H
MINT
q L? PWO.l .16 + .!.0 +;
—m
,%-. E lhf . 2,25Z10-%!.IOO”; ”U!M
og~, \
& ~
~: ~ ~f::r+p ‘

$$$ ,,2 SLOW L,NE

.-. . .
% ,0-’,
10 10 I@
SHUT-lNTIM&: At, MINUTES

Fig. 22—Match of rate-nonnaliid response with uniform.


flux type Cuwe.

Fig. 21 —Shut-in pressure and mtq field example.

SPE Formation Evahxion, October 1986 463


1 1 1 1 ! 10

MATCH
MINT

“m
gz %..1.0!, P.c.. 1%5

“)
0.6 - “o
.> fox,
.2.4
.10-2,
A+.100.1””!8,
?& ~-+---
1
m.
=?.
:;
& ho-al
0.4 - J +.,,.01
00
10”1 0 * ,.,
10 10*
sHUT-IN TIM!: At, MINUTES

’19. 24-Match of Method 1 solution with infinite.


conductivity type curve.
0.2 -

0 By using Eq. A-9, we constructed the PFm function -


from the pressure/rate data of Fig. 21. We were umbIe
to obtain a eood match of the P nC,Vvs. At data using the
mriform-flu~ type curv~ howe~~~, we found that ~ese
dsta could be reasonably matched with the infinite-
SQUARE ROOT OF SHUT-IN TIME, conductivity type curve for. fractured wells. Our match
@~ is shown in Fig. 24. Fmm the match-point values rccordcd
on Flg. 24, we obtained (khlp) ~= 142.6 md-ftlcp
Fig. 23—Square root of time plot of rats-normalized [43.5x103 red-m/Pa. s] and Lti=87:4 ft [26.6 m].
These estimates am roughly 10% less than the correspmd-
ing estimates obtained by matching the rate-normalized”
data with the uniform-flux type curve. From these rcsnhs,
where 4, remcsents the total sandface flow rate (oil. ms, we found r; =LX /2=43.7 ft [13.3 m] ands= —ln(r; /
and wa;~r) ‘k reservoir ba~els per day. Because ~er~ r~)= ‘5.3; thus ~e values of the r; ands obtained are
tlow (wellbore storage) is present, the log-log plot of Ap close to the values obtained using the rate-normalized
vs. At displays a unit-slope fiie at early time$ that is the procedure.
characteristic behavior of wellbore-storage-dominated It seems possible that the difference between the rate-
data. (The well-known 1.5-log-cycle rule indicates that normsliied analvais results and the results obtained from
none of the data lie on the conventional semilog straight the P,FcN am?tys;s &dd be the resutt of the behavior il-
line.) lustrated in Figs. 16 and 17, which showed ‘lhatthe dimen-
The log-log plot of tfre rate-normalized dats display a sionless rate-nomrsliied weUbore pressure-drop solution
baff-slope line at early times that is characteristic of a frac- for an infinite-conductivity vertical fracture with wellbore
mrcd welf. ‘II@ can be seen in Ilg. 22, which shows the storage effects wifl correlate well with the uniform-flux
type-curve match of the rate-normalized data obtained solution for 10’2s tDx, s 4. On the basis of field expe-
with the uniform-flux type curve for vertically fractured rience, however, most people believe that the itiltdte-
wells. Using the match points recorded in Fig. 22, we conductivity type curve should be applied only when. a
obtain (k/r/p)r = 163.8 md-ftlcp [49.9x 103 md. dPa. s] propped hydraulic fracture exists, and Ref. 51 gives no
{(~#),=455 md/cp [455x 10’ md/Pa.s]} and L.ti= indkation that this is the case.
96.8 ft [29.5 m]. (These values arc withh 2% of the
values obtained by Fetkovich and Vtenot, 14 who conside- Conclusions
red the ssme problem.) From these results, we find that We have demonstrated the applicability of annlysis
the effective wellbore radius r~ and skin factor are methods that are based on Dubamel’s principle. These
rL=Lx /e=35 .6” ft [10.9 m] and s= –ln(r~/rW)= methods csn be used to convert pressure data obtsined
–5. 1. {n our results, (klp), represents the total mobMy when the sandface rate is variable to the equivalent pres-
because the total sandface rate was used in the analysis. sure dsta that would have been obtained for a constnnt
(Both the type-curve match and the double-Ap rufe indi- sandface rate of production. The methods can be applied
cate thnt semilog analysis ia not applicable.) to both drawdown mrd buildup data. For the drawdown
The enrly-time data of Fig. 22 are flat and approach” case, the methods can of course also be applied with sur-
the haff-slope line from above. This behavior could be face rates provided that the surface rate is an accurate
a rcsuft of the existence of damage on the fracture face; retlcction of the sandface rate. It has been shown that
however, the Cartesian plot of AplAqZ vs. ~t shown Methods 1,2, and 3 can be applied to a wide variety of
in Fig. 23 exhibits a straight line during the linear flow problems and sre particularly advantageous for analyz-
period (lmlf-slope line on the log-log plot) that extmpo- ing well test &ta obtained from commingled layered reser-
latcs to zero at Ar=O. If damage were present, this stight voirs. We have imoduced a new procedure, Method 3
line should extrapolate to a positive vafue of (Ap/Aq,). of Appendw B, which is ptiicularly useful for analyz-
Thus we conjecture that the flat portion of the data may ing well test &ta influenced by reservoir boundary ef-
be csuscd by smalf errors in the measurements of pres- fects. The three methods based on Duhamel’s principle
sures or rntes. The small oscillations in Ap/Aqt (see Figs. have been compsred to tie rate-nommdiied analysis proc-
22 snd 23) seem to indicate this possibility. edure.It has been shown that when data are strongIy in-
464 SPE Formmion Evaluation, OclOba 1986
fluenccd by wellbnre stcmige effects, the rate-normalized Cm = dimensionless fracture conductivi~
pressure solution will be displaced below the correct h = total formation thickness, ft [m]
constant-sand-face-rate solution. We do not view the lat- hfi = total fracture thickness (natrrrully fruchrrcd
ter statement as a castigation of the rate-normalization reservoir), ft [m]
prncedure because for many problems this displacement h. = matrix thickness (natnrdly fractured
will not bc large. Morcnver, rate-normalized analysis is reservoir), ft [m]
simpler to use than Methods 1, 2, arrd 3, and dues give
h. = thickness of perforated interval, ft [m]
a guod irrdication of which data points may be erroncmrs.
h .D = dirneusiordess wellbnre length defined by
Of the three metfmds based on Duharnel’s principle,
Method 1 is rdways applicable. We recommend the use Eq. 2g
of Metbud 2 ordy when the ssndface rate (change in sand- ~ = “prmeabfi~, md
face rate for buiIdup) decreases throughout the test. Be- k = thjckrress-averaged pcnneabiity, nrd
cause of ita murrputzdiond complexity, we recommend the kf = fracture perrneabdity, md
use of Method 3 only when it is impnrtarrt to generate k, = vertical permeability, md
aczumte equivalent constarrt-aarrdface-rate.data during the LX, = fracture half-length, ft [m]
drrreperiud brtluenccd by boundary effects-c.g., irrreser- L = effeqive fracture half:length defined by
voir liit tcstirrg under vsriablc-flow-rate rcmditions. Tbe Xf
E’+ 12, ft [m]
obvious firrritation of Methods 1, 2, mrd 3 and the ratc- n = index’
norrmdization arrslysis is that sandface flow rates must be
p = pressure, psi [kPa]
available. However, because of the importaut analysis ad-
vmrtage these methuds offer, particularly for heteroge- Ap = p~ssure drop, psi ma]
neous formations, we expect that intcrrse efforts will be PD. = henSiOIrleSS shut-in pressure defined by
made to iruprove our abdity to measure sandface flow Eq. 17
rates accurately. Currently, sandface.tlow rates are meas- PD. = dh;~ shut-in pressure defined by
ured by production logging, aud this requirss that we con-
vert spinner rates in revolutions per second to sandface qi = initial reservoir pressnre, psi @a]
flow rata irr reservoir barrels per &y. 55 For buildup data @(rj) = pressure drop if production is a cOnStant-
obtained from pumping wells, the sarrdface rate during sandface-rate of q,
buildup can be obtairrcd by measuring the rising liquid
P WCD= qtivdent constant-sandface-rate
level using acouatic devices. 51
dnensionless wellbore ‘pressure drop
On the basis of Ork study, the foflowing conclusions
PWD = dimensionless wellbore pressure drop ob-
are warranted.
1. Methods based on Duharnef’s principle can be used tahd under variable-sandface-rate pro-
to adyzc well rest data when sandface flow rates ure duction
available. PW = flowing bottorrrhole pressure, psi ~a]
2. If individual layer ssndface flow rates are avsilable, pm = flowing wellbore pressure at shut-in, psi
then the methods can be used to eatinrate individual layer &Pa]
properties. PW. = dtut-in wellbore pressure psi @a]
3. The three methods can be used to estimate reservoir PO = bOttOfiole opening pressure (DST), psi
PV when reservoir fimit testing is conducted under ma]
variable-flow-rate conditions (the rate-normalized pr.xc- PFCN =’ norrnaliied pressure drop for constarrt-
dure should not bc used under these circumstances).
sandface-rate production, psi/RB-D
4. Although the rate-norrnaliied analysis prucedure m2y
yield suftlcieritly accurate results for practical purpnses, walm3. d]
the rate-normaliicd pressure drop is displaced below the q = s~dface flow rate, RB/D [~ m3/d]
equivalent corrstarrt-samlface-rate solution during the time q‘ = derivative of the sandface flow rate
periud dominated by wellbore storage effects. (tn+l ‘tj)
CID = diMenSiOIdeSs saudface rate defined by
Nomenclature Eq.2
,4 = drainage area, fiz [m2] qr = reference sandface rate, Rf3/D [reS ~3/dl
b = penetration rstio &fined by Eq. 27 r, = reservoir drainage radius, ft [m]
Cj = cc-et%cient defined by Eq. B-2 ro = dinrensiouless drainage radk
c. = coeftlcient defined by Eq. B-2 rw = we~bOre ~di”s, ft [m]
c1 = system compressibility, psi’1 [kPa-l] rw‘ = effective wellbore radius
ctf = toti fiaCtUm Compressibiity, psi -l RqD = dnerrsiordess reservoir conductivity of
[@a-l] Layer j, defined by Eq. 10
C = wellbore storage cmrstarrt, RB/psi s = skbr factor
[rcs m3/stock-tank m3] sb = “pseudoskin factor rcsuking from ptisI
CD = dimensionless wellbure storage constant for peuetrmion
mdiaf flow problems s, = total skin factor
C&, = dimensiordess weflbore stnrage constant for r = time, hours
vertically fractured wells fD = dimensionless ti3rreba.scd On ~e~~rc
Cf = fracture conductivity radius
SPE Fomtion Evahmian, Cktoter 1986 4s5
t n, = dimensionless time based on Lswer 1 nrou.
“.
6. Bostic, J, N,, Aganval, R.G. and Carter, R.D.: “Combined Anal-
efty defined by Eq. 35 “ “ ‘ ysis of Postfmctming Performance and Buildup Data for Evaluating
an MHF Gas Well,,, JPT (Oct. 1980)1711-19.
‘D,4 = ~~ensio~ess time bas~ 011 dr~nage area 7, Pascal, H. and Quillian, R. G.: ‘<New Method for Predicting
tDx, = dlMeIMiOUless time based on fracture half- Deliverability From Variable Rare Drawdown Data,,, paper SPE
lenglh 7932 presented at the 1979 SPE Syq-.xium on Low-Permeability
Gas Reswvoirs, Denver, May 20-22.
~j = point in the subinterd [tj ,tj+, ] defined in 8. Kucuk, F. and Ayesm’an, L.: ‘sAnalysis of 3imulianeo. sly Meas-
EG. A-7 ured Prmwre and Sandface Flow Rate i“ Transient Well Tesd”g,, Y
tp = prod~cing time before shut-in, hours JPT iFeb. 1985) 323-34,
9. McF.&wds, D,G.: ‘iMddwdl Variable-RateWell Test Analysis,,,
tPD = diienSiOfdess producing time before SPEI (Aug. 1981) 444-53.
shut-in 10. Tiwmpmn, L.G., Jones, J.R., a“d Reynolds, A.: ‘,Analysis of
At = shut-in time, hours Pressure Buildup Data 3nfluenced by Wellbore Phase Redistribu-
tion.” SPEFE (Oct. 1986)435-52.
AtD = dnensionless shut-in time based on well-
bore radius
ASDX,= dimensionless shut-in time based on . .
fracture half-length 12. Gladfelter, R.E., TIXY, G.W., and Wilsey, L.E.: ‘W&l@ Wells
Which Will Rq.md to Fmducdcm Stinmlatim Treatment,,> Drill,
~ = ~atch point and Prod. Pmt. , API (1955) 117-28.
y = Euler’s constant (0.57722) 13. Winestrxk, A.G. and Colpins, G.P.: ‘,Advances in Estimathg Gas
Well Ddiverabfify.z> J. C&. PeI. Tech, (ldy-Sem. 1965) 111-19.
O = weighting factor for a uniform weighting
14, Fetkovich. M,J. and Vienot. M. E,: %x. Normalizationof Build.~
for all ;. Presmre Using Afterflow Data,,, JPT (Des. 1984) 2211-24.
A’ = dimensiofdess ffacture transfer coefficient 15, Mehfest, H.: “NumericaJ Inversion of Laplace Trmsfonns,’,
Commu”icariom of rheACM (lam 1970) 13, No. 1,47-49,
defined by E@. 32
16 Prij.rnb-cdo, R., Raghavan, R., .md Reynolds, A.C.: ‘.Well Test
p = viscosity ~y$is f.r wells pr~.ciag tiYe~ R=ewOi~ wi~ CMSSflOW;’
_ q$ = po;os~y, fraction SPEJ(Jme 1985) 380-96.
#c, = tficfmess-averaged porosity/compressibility 17, Reynolds, A.C., Chen, J. C., and Raghavan, R,: <‘Pseudo-3kin
Factor Cawed by PardaJPenetmdm,!, JP7fOec. 1934)2197-2210.
product defined by Eq. 9, psi-1 18. Bmmtt, C. O., Reynolds, %A.C.,and Raghavam, R.: ‘.Analysis of
[kPa-i] Finite-Condwivicy Fractures Imercepti”g Multilayer Reservoir,”
SPEFE (June 1986)259-74.
$f = fracmre porosity, fraction 19. Be.neu, C.O.: ,’AmIysis of Fractured Wells,,, PhD dissertatim,
co’ = diiensiofless matrix storativity defined by U. of Tulsa, Tulsa, OK (1982).
Eq. 33 20, A1-Hussdny, R. , ~ty, H.J. Jr.. and Crawford, P. B.: “The Flow
of Real Gases Thro.& Porous Media.’, JPT(MaY 1966)624-36
Tin., AIME, 237;
SubWripts 21. A@v@, R.G. : .<Real Gas Pseudo-Thn+A New Function for
f = fracture system Pressure Buildup Analysis of MHF Gas Wells,,, paper WE S279
i = intinite-conducti~lty scale presmmd at tie 1979 SPE Ann”al Technical Conference and
Ex.libitiq Las Vegas, Sept. 21-24.
j = layer index 22. !-% W.J. and Holditch, S,A., ‘<Applicationof Ps?udotimem Build.
m = matrix system “p TM Analysis of Low.Pemeabilhy Gas Wells with Imnz-
~ = ~*ifO~.fl”X ~~~e Duradm W.ellb.x?SfomSeDismrtion,’, JPT(Dec. 1982) 2877-87.
23. Sem, K., Reynolds, A. C., and Raghavan, R.: ‘,New pressure
I= Layerl Tramient Analysis Melhcds for Naturally Fractured Reservoirs,,z
2 = Layer 2 JPT (Dec. 1983) 2271-S3.
24. Camacho-V., R.G., R@OJm, R., andReymlds, A.C. : ‘.REWW.SS
of FracmmdWdk Fm&.ing Layered Reservoirs,UnequalFracture
Length,’, paper SPE 12844 presented at !he 1984 SPE/DDE/GP3
Acknowledgments U.wnve.tionslGasRecoveIYSymposium, Pittsburgh,
May 13-15.
Portions of thk study were completed by L.G. Thomp- 25. Fefkovich, M.J. and Thrasher, T. S.: ‘aConsfant Well Pressure
Testing and Amdy$isin Law Permeability Reservoirs,>, pqer SPE
son whh a grant from Flopetml Johnston/Schlumberger 792Spresented at the 1979SPE Rccky MounminResional Meed”g,
as well as a U, of Tulsa research assistantship. We thank Denver, May 20-22.
these sources and the Dept. of Petroleum Engineering, 26. Uraiet, A.A, .md Raghavrm, R., ..tiniteady Flow to a Well
U. of Tulsa, for financial support. Prtiu.im at Constant Presmre.,, JPT (Oct. 19gO) 1803-12.
27, Ehlig-Ecmomides, C.A. and Ramey, H,]. Jr.: “Transient Rate
Dcclim Analysis for Wells Pmdwed m Constant pressure,,, SPEI
Feb. 1981) 98-lC@.
References 28. Jones, P.: ‘<ReservoirLindt Test,,, Oil& Gas J. (1.”. 18, 1956)
1. van Everdingen, A.S. and Hum, W.: The Application of Ow 13+96.
LaplaceTransformation
to FlowProblemsinResemoirs,’! Tram., 29. Earlougher, R.C. Jr.: Advances<. Well Test Anolyif, Mmograph
,4fME (1949) 186, 305-24. Series, SPE, Rjch?.rdsom,TX (1977) 5, 29-30.
2.Odeh,A.S. and Jones, L. G.: ‘,Pressure Drawdown Analysis, 30, vm Poolle”, H. K.: ‘Status of Drill-Stem Testing Teduiq.es md
VariableRate Case, ‘sJff (Aug. 1965)960-64; Tram., ASME,234 AlldySiS, XVJ.w (April 1961) 333-39.
3. Soliman, M. Y.: .<New Technique for Analysis of Variable Rate 31. Koldhaas, C.A.: .’A Medwd for Aru.lyzi”g Presswes Measwed
or SlnS Test,’7 paper SPE 1CCB3presented at fhe 1981SPE Annual During Driliwmn-Test F1OWPeriods,,, JPT (OCt. 1972)1278-82
Technkal Conference and Exhibition, %n Antonio, Oct. 5-7. Tram., ASME, 253.
4. Stevmn, G., Winmmn, M. J., and Meunier, D.: “Afferflow Meas- 32. Ramey,H.J. Jr., Asarwa&R.G., and Matin, I.:”...%alysis of Slug-
uremem and Deconvoludon in Well Test Analysis,” parer SPE Test or DST Flow Period Data,, sJ, cd.. Per. Tech.(July-Sepf.
12174presented at the SPE 1983 Annual Tezhnkal Conferenceand 1965>3742.
Exhibition, San Francisco, Oct. 5-8. 33, Gringariem A.C. and Ramey, H,J. Jr.: ‘CA. Approximate fntinite
5. hrgon, J,R. and van F’collen,H.K.: ‘Thdt RespmiseFu.cdon Fmm Conductivitysolution for a F%m&JfyPenmming Line Source Well,”
v~~g ~ Da%” JpT(Aus. 1965)963-6% TWLS..~E. ~. SPEJ (APril 1975) 14048.

464 SPE Formadon Evalmtion. October 1986


34. Bilhartz, H.L. Jr. and Ramey, H.]. Jr.: “The Cmnbincd Effccfs To apply the schemes to variable-sandface-rob draw-
of Slorage, Skin and Ptial PenetrationonWellTestAIMIysis,,, down data, partitions of the time interv~ [0, t] must be
paper SPE 6753 presented at tie 1977 SPE Annual Tecimicaf
Conference and Exhibition, Denver, CM. 9-12. used, and for buildup, partitions of the time interval [0,
35, Bmm. F. and Mating, V.W.: “TIE Effectofkmkted Fluid Enlry At] must be used. For both drswdown and buildup, these
on Well FTcducdvify,,, JPT (Feb. 1961) 172-74, Trans., AfMB,
.-.
psrdtions are denoted by {tj}jE~l, where for drawdown
Z22.
36. Ranmy, H.J. Jr.: .<Verikicmionof the Gladfelter-Tracy-Wdsey
Concept for Wellbore Storage Dombmted Transient Pressures O=to<f, < . . ..<tm<tn+. =t. . . . . . . . . . ..(A-l)
During Pmd.cfion,” J. C&. Pet, Tech. (April-June 1976) 34-85.
37. Soelt.wva, T.D.:‘.Well JWssure Be&vim of a Na@maJJy Fracmred
Reservoir;’SPEJ (Oat. 19s3) 769-SO. sod for buildup,
38. Cimo-L., H. and Samaniego-V., F.: <‘pressureTransient Analy-
sis Methcds for Nammlly Fractured Reservoirs,.+ papr SPE 11026
W.$en~ at fbe 1982 SPE Annual Tcchni.al COtiere”.e d O=tO<tI <... <tn<tn+l=Af. . . . . . . . . . ..(A-2)
Exhibition, NW Orleans, Sqt. 26-29.
Tramien! Anafysi.sMethodsfor Bounded
39, Chen, C. sl al.: Ylessure
Nafmally Fractured Reservoirs,,’ SPEJ (J.m 19S5) 451-64. Here, P,Fm denotes the equivalent normalized drawdowo
40. Bmudet, D. and Gr@rfea, A,C. : , ‘Detaminadm of Fissured .
Volume and Blcck Siie in Fractured Rewwoirs by Typ#2.Ive pressure drop we wish to generate-i.e., P.Fa represents
Analysis,” paperSPE9293presentedat the 19S4SPEAm.af the normdiied pressufe dfop we would obtain for a con-
TechnicalConference and Exhibition, Dallas., Sept. 21-24. stant sandface rate of production. Notationally,
41. Reynolds, A.C. et al.: ‘<The Wellbore Pressure Response in
Naturally Fractured Reservoirs,,, JPT (May 1985)90S-20.
42. Tariq, S.M.: ‘<AStudy of fhe Behavior of Layered Reser@irs wifh
PFcN=Ap/qr, . . . . . . . . . . . . . . . . . . . . . . . . ..(A-3)
Wellbm Storage and Skin Effect,” PhD dissertation, Smnford U.
(1977). where Ap is the pressure drop in pounds per square inch
43. Larsen, L.: “Well$ Producing Commingled Zmm with Uneqml that would occur if the well were produced at a constant
Jnhial Freswlres and Rese.t-mir Pmperdcs,,, paper SPE 10325 sandface rate equal to q,. The methods mnaidered here
presented at the 19S1 SPE Ammal Technical Conference and
Exhibition San Antonio, Oct. 5-7. afe used to generate PPW and knowledge of Ap snd q,
44. Fefkovich, M, J,: ‘&DeclimCwve Analysis Using Type Curves,,> is not necessary or desired.
JPT (Jwre 19S0) 1065-77. The drawdown analysis scheme referred to as Method
45. Gringartm, A. C., Ramey, H.J. Jr., and Raghavan, R.: VJnsteady- 1 for n> 1 is given by
Stafe Presmre-DEmMicms Created h a WdI vfiti a Sinde -. Jnfinite-
Co”ducdvi& Verdcal Fracture,’, SPEJ (Aug. 1974)347+ Tram.,
AJME. 2S7. [Pi–P@”+l)l-sm, ,..,,,,,, (A4)
46, Chin-L., H,, %maniego, F., and Domiuguez, N.: ‘STransiem PFcN(in)=
Pr.s$sureBehavior for a Well wifh a Fhite ConduXivily VerdcaJ q(t”+ , –t”)
Fracture;x .7PEJ (Aug. 1978) 253-6-$.
47, Agarvd, R.G., Caner, R. D., and Pollock, C.B.: “Eva.lu&m and
Performance of l.ow-Pemmability Gas WelJsSdnudated by Massive where
Hydrmlic F..acmring,>, JPT (March 1979) 362-72. -. . .
48, Ci”co-L., H. md Sa.mmiego-V., F.: “Effect of Wellbore Storage “—l
and Damage in tie Transient Pressure Behavior of Vmdcafly
Fractured Wells, ” paper SPE 6752 presented at the 1977 SPE
s~= Z
j=O
pFC,V(~)[dr.+1 ‘fj)-dt”+l ‘fj+l)l.

Atm.al Tecb”ical Conference and Exhibition, Denver, Oct. 9–12.


49. L@&S. and Brockmbrough, J.: ‘<ANew A@tical .%lutirm for
Finite Conductiviw Fractures with Real Tine and ldwe Smace . . . . . . . . . . . ..’ . . . . . . . . . . . . ..(A-5)
Parameter Estimation,” paper SPE 12013 presented it the i9S3
SPE Ammal Teh”icd Co”fereme and Exhibition, San Fmncixo, men ~qo, ~. A-4 is replaced by
Oct. 5-8.
50.JWwm, R.: ‘“TheEffecfofF’rdcingTim m ‘ryp cuw AMI.
ysis,39JPT (June 19s0)
1 1053-64. pF&(;o)= [Pi ‘Pwf(tl )1
51.Bmwn%mmk, E.FR.: , CAfteflows and Buildup Inta’pretatiom on . . . . . . . . . . . . . . . ,. (A-6)
Pumping Wells,,, .JPT
. . (Feb. 1982) 397-405. q(t ~)
52.Pemine, R.L.: “Analysis; of pressure Buildup Curves,,> DrilL ad
Prod. Pmc, , API (1956), 4S2-509.
53.Martin, J. C.: “Sirnpli!ied &@ions of FJow in Gas Drive Here ~ denotes a point in the submterwd [tj, tj+t] Ofthe
Hesermirs md the ThmretimJ Foundado” of Multiohase Flow psmition of 3?q. A-1 and is given by
Buildu~ Analysis., s JPT(Oct. 1959) 321-23: Trans.. .iJ2.fE, 216.
34. Chn, W. C.,- Reynolds,’ A. C., &d R@&.n, R“.: T&sure
Tramient AmJvsis of Two-Phase F1OWProblems.,, SPEFE (June
;=ejtj+,+(1–e.)t., ,, . . . . . . . . . . . . . . . . . . . . (A-7)
1986) 151-64.”
55. Awad, M.A. and Tergim, J.: ‘%npx’tance of Bouom-Hole Flow where
Monitoring During Transient Pressure Testing,r> Proc., SPWL.%
New Orleans (June 1934). Osejsl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..(A-8)
‘Appendix A-Dimensional Vereion of
Methods 1 and 2 Selection of the tJj values wifl be discussed later. (lf we
IIJRef. 10, using dimensionless variobles, we derived choose Ej= % for all j, then the method of Eq. A-1 rep-
methods for fhe generation of constant-sandface-pressure resents the dinzensionsl version of a method presented in
data given pressure and vsriable-sandface data mesaured Ref. 8.) For buildup &ta, Method I is given by
as a function of time during drawdown or buildup. Here,
the difnensionaf versions of the methods of Ref. 10 are pFm(;n)= [Pws(fn+l)–Pwd +sum
recorded, and guidelines for the practical application of , . . . . . ..(A-9)
these methods are presented. fmp)-lxzp +1.+1 –k)

SPE Formation Evaluation, OcfoteI 1986 467


where to generate PFm aa a function of time. Oiven buildup Y.
pressure and sarrdface rate data, the schemes of Eqs. A-9
and A-15 can be used recursively to generate PFN as
a furrction of shut-in time. If both drawdown and buildup
pressure and rate data are available, F.qa. A-4 and A-12
can te used to obtain PFa aa a function of total test time.
–q(tp+tn+l–tj+l)]. . . . . . . . . . . . . . . ..(A-IO) In SUcases, PFCN represents the normalized drawdown
pressure respnnae that would be obtained if production
when n=O, Eq. A-9 is re~laced by were at a constant aarrdface rate.
when necessary, linear interpolation is used to obtain
PFCN and q vahres appearing on the right sides of Eqs.
~Fa(;o)= [Pw$(fn+l)–PWl
. ...... (A-11) A-5, A-10, A-13, and A-16. If we choose dj=l, for dl
q(~p)–q(tp+tl) j, then except for this linear interpolation procedure, Eq.
A-12 is equivaferrt to Bostic et al.’s method.
Here, the partition of F.q. A-2 is US@ pm representa the As discussed in Refs. 10 and 11, it is necessary to usc
measured shut-in pressure, p ~ represents the flowbrg an iterative procedure when applying Metbcd 2. For build- . .
wellbore pressure at the instant of shut-in, and fp repre- up data, if the change in sandface rate, @t=O) –q(At),
sents the producing time. The q’s represent the sarrdface is increasing as in wellbore-storage-domimtcd data, then
rates r~orded during shut-in. Bccawse tP cnr’responds to Method 2 @q. A-15) will be unstable udess the partition.
At=O and At=tn+, @q. A-2), we can use the notation of Eq. A-2 ia selected so that
q(At=O) for q(tp) and q(At–tj) for ,q(tP +?”+ I ‘Ij),
which indicates that knowledge of tp K unnecessW.
tj+, –tjstj–tj-, . . . . . . . . . . . . . . . . . . . . . . (A-is)
For drawdown analysis, Method 2 for n= 1 is given by

for SUj. Because we normally wish to u= partitions such


~Fw(tn+,)= [n-p~~.+])1-sum
~(;o) , . . . . . . (A-12) that tj+, –tj > tj ‘tj- I, this restriction is incOnve~ent.
For drawdown cases where q(t) is increasing, it is also
necessary to require that the inequahty of Eq. A-18 holds
where when Method 2 ia applied (Eq. A-12). On the other harrd,
if q(t) [q(At=O) —q(w) for buildup] is arr incrcaaing func-
n-l tion of time, it is necessmy to choose the partitiom of
sum= ~ PFm(~.+l-tj+l) the time intervals such that

rj+, —tj>tj —tj-~ . . . . . . . . . . . . . . . . . . . . . . (A-19)


x[d++l)-q(i)l. . . . . . . . . . . . . .. . . . .. (A-13) \
for all j to achieve stabti~ if Method 1 is used. When
When n=O, Eq. A-12 is replaced by q(t) [q(At=O) –q(At) for buildup] ia a decreasing fmrc-
tion of time, no restrictions cmthe partitions of Eqs. A-1
~Fm(ti)=p-%@)
—. .. . . . . . . . . . .... . ..(A-14)
and A-2 nre rreceasary to achieve stabtity. A detailed sta-
q(;o)
btity analysia is given in Refs. 10 and 11.
It is convenient to chonse p-aintaof the pmthions of Eqs.
A-1 and A-2 to corm.$pnndto times at which pressure and
The corresponding scheme to be used in the analysis of rate &ta arc measured. However, it may bc rrccessary
buildup data is to add or to delete points from partitions selected in MIS
way to achieve srabtity andlor accuracy. At times where
~FcN(tn+, )= IPws(tn+i)–hf.l+sum they have not been measured, pressure and rate data are
.(A-IS) generated from the measured data by linear interpolation.
q(fp) –q(tp +io)
We recommend the foUowing guidelines for use of
Methods 1 and 2 in the analysis of field weU teat data.
for n z 1, where Plot, on log-log coordinates, q(t) va. t for drawdown data,
.—1 or q(Ar=O) –q(At) vs. At for buildup data.
NllIl= ~ P~~(tn+l ‘tj+l)[q(tp ‘~+1)
Csae 1. If the slope of this log-log plot is always zero
j=O
or negative, then both Methods 1 and 2 are stable regard-
less of how the partition of Eq. A-1 @q. A-2 for bdiid-
–q(tp+;j)]. . . . . . . . . . . . . . . . . . . . . . ..(A-bS)
UP)i$selected. In thk case, either Method 1 or 2 can be
used successfidly. Method 2 generally will give slightly
when rr=O, Eq. A-15 must be replaced by more accurate results but Method 1 always is more com-
putationslly convenient. In uairrg either method, one,
~Fm(t,)= rPws(tl)-Pyl . . . . . . . . .. . . ,(A-lV should use 9j = 1 for SUj. For either method, sufficient-
q(tp)–q(~p +fo) ly accurate results wiU generrdly be obtained with 10
pointa per log cycle irr the partition of Eq. A-1 @q. A-2
Given drawdown pressure and sandface rate data, the for buUdup) but adding more poirrta wiU increaae the ac-
schemes of MS. A-4 and A-12 can be used recursively curacy obrained.

46a SPE Formation Evaluation, October 19S6


Case 2. If the S1OPC of the log-log plot mentioned above and
is pnsitive, Method 2 will be unstable urdess the inequal- n-1
ity of Eq. A-1 8 is satistied. In this case we recommend sum= ~ PFcN(tj+l)(cj –cj+, ). . .(B-3)
using Method 1 with !j = % for all j and”with a partition j=O
selected such that the mequali~ of Eq. A-19 is satisfied.
More specifically, aa shown irrRefa. 10 and 11, we should III f+q. B-2, the tam q’ (tn+, –tj) denotes the derivative
choose the partitions such that of the raie function evaluated at time f.+ I ‘tj.
The foflowing pmccdure ia recommended for the prac-
drn+l “j)-dtn+l ‘tj+l) tical application of Method 3.
1<1, . . . . . . ..(A-20)
q(tn+, –t”) 1. Let {tm}~~~ denote the aet of points at which pres-
sure and rate data are rneaaured. On each aaanciatd subii-
when analyzing drawdown data, and such that terval, t., which is strictly <t arrd s tm+,, compute the
derivative of the rate function with
, @A~-~)-XA1-rj+,)
Is,, . . (A-21) q(tm+] )–q(tm)
q(Ar–O)–q(Af–tn) q’(rm+~)= , . . . . . . . . . . . . ..(B4) -.
tm+l –tm
when analyzing buildup data. Sufficient accuracy can be
obtained by chonsing partitions that contain IO or more for rrr=O, 1.. .s. Here the “auperacript” - is used to
poirr~ per log cycle, but it may be necessary to uae fewer denote that Eq. B4 gives the value assigned to q’(t) at
than 10 pnirrta for some log cyclca to maintain stability. all times that lie irr the subintcrvaf t., which ia strictly
For either Case 1 or 2, accuracy wiff be enhanced by <t arrd St.+l. At this point, we have a table compnacd
using more tharr 10 poirrta per log cycle in regiona where Of v~ues t~, p~r~), q(t~), and q’(F~+I) for
the absolute values of the slope on the log-log plot of rate rrr=o,l. :.$.
q(t) [q(At=O) –q(At) for buildup] vs. time is large. 11 2. Define the sequence of times {tj} j~~’ at which
The dimensiordess version of Eq. AA is given in Refs. PFW data are deaircd.
10 arrd 11 and can be obtained by multiplying 13q. A4 3. Compute PFcN(tI ) from the following equation:
by kM141 .2P and letting
Pi–P&h)
P~c,./(t~)= . . . . . . . . . . . . (B-5)
khp~~ !&Lip q(t~)–o.5t~q’(f-~)
p IWD = —= — ,,. ,,, ,, . . . . . .. (A-22)
141.2P 141 .2qrft where p ~ is the initial pressure at the beginning of the
drawdown teat.
The dnerraiordess versions of Eqs. A-9, A-12, and A-15 4. Compute subscqucrrt values of PFm by applying Eq.
carr be obtainqd by the sarrre procedure and are also given B-1 recursively. At each arep of the prncedure, borb q
irr Refs. 10 arrd 11. -. . ..
arrd q‘ must be evaluated to compute Cj; see Eq. B-2.
For the analysis of trarraient data, either Method 1 or Terms of the form q(rn+l –tj) are evaluated by linear in-
2 shonfd be applied following the guidelinca defirreated terpcdation between meaaured rate vs. time &ta. Terrrra
abnve. When pressure and rate &ta are controlled by out- of the form q‘ (t.+, -rj) are evaluated by searching the
er boundary effects, Method 3 of Appendm B will yield table cmratructed in Step 1 tn tind the subinterval such
the most accurate pseudostcady-state P,ccN data. that tm <t.+, –tj =tm+l, arrd then set

Appendix B-Dimensional Version of q’(tn+, –tj)=qz(rm+, ). . . . . . . . . . . . . . . . .. (s-6)


Method 3
The derivation of Methods 1 arrd 2 of Appendix A are As shown in Ref. 11, Method 3 is stable if q(t) is a
given irr Refs. 10 and 11. The derivation of Method 3 k smrdrly decreasing furrcdon of tie. If q(r) is arrirrcrcaa-
similar and is given in Ref. 11. Here, we present ordy ing timction of time, then the pnirrta {tj }j~~ 1 at which
the dimensional version of Method 3. The analysis of PFCN is computed must satis@ the inequality
reservoir limit tcsta cmrducted under variable-rate condl-
tiorra appears to be the most important practical applica-
tj+L-tj>tj–tj_, . . . . . . . . . . . . . . . . . . . . . . . (B-7)
tion of Methnrf 3. Thus, for brevity, we include ordy the for all j to guarantee stabfity. Generally, if 10 pninta pcr
drawdown version of Method 3. The buildup analog is log cycle arc used irr the partition {tj} jl~’ , tkll Suffi-
given in Ref. 11.
cient accuracy will bc obtained when Method 3 ia applied.
By using the notation of Appendix A, Method 3 for the Further derails are given in Ref. 11.
analysis of drawdown data is given by ..
S1 Metric Conversion Faotore
~FcN(tn+l)= [Pi–P@(t.+l)l-sm, ....,.. ,@-l) bbl X 1.589873 E–01 = m3
c?! tl X 3.048* E-01 = m
psi x 6.894757 E–00 = kpa
where
w.m$nmlon ratio, is exact SPEFE
Cj=q(tn+l ‘tj)–0.5[(tn+l ‘tj)9’(tn+l ‘fj)l Orlghd mm.suipt mwN6d in the *MY ofPeomleumEngi.earsollbeSnp. 16,
l?W.FaPW-W forFubrkationP.&M
19,1%5 ,%v+sed
rmnusz6PIracakWFeb.
Z6,198$Psw(SPE 13WiJ)IimtPmn!ed e,, lb 1w4 SPE Annual Technics Cm.
. . . . . . . . . . . . . . . . . . . . . . . . . . . ..(B-2) fefnma and Extibi!lcm held i“ Ho”,!m SePt W-19.

SPE Formation Evahmion, October 1986 469

You might also like