You are on page 1of 16

Analysis of Simultaneously Measured

Pressure and Sandface Flow Rate in


Transient Well Testing @
F. Ku?uk, SPE, Schl.mbc%er Well Services

L. Ayestaran, SPE, Schl.mberger Well Services 5R!5 lal’71

SUrmnary

New well test interpretation methods are presented that tkd. penetration, perforation, aciduing, fractures, nOn-
eliinate wellbore storage (aftertlow) effects. These new Darcy flow, and permeability reduction caused by gas
methods use simultaneously measured sandfuce flow r“te saturation around the wellbore) dominate pressure
and weffbore pr~ure Ma. It is shown that formation behavior. The stratification and dud porosity also may
behavior without storage effects (unit response or in- affect wellbore pressure during this period.
fluence function) can be obtained from deconvolution of
Jkmdface flow rate and wellbore pressure data. The Middfe-Time Period. During thii period, radial flow is
storage-free formation behavior canbc analyzed to iden- es@blished. Conventionally, semilog techniques are used
ti& the system (reservoir flow pattern) that is under testing to determine formation kh and initial pressure and skin.
and to 6stimate its parameters. Convolution (radial multi-
rite) methods for reservoir parameter estimation and a Late-Time Period. During tbk period, outer boundary
few synthetic examples for deconvolution and convolu- effects start to distort the semilog straight line. For ex-
tion also are presented. ampIe, the gas cap shows a curve-flattening effect on log-
log and Homer plots.
Introduction Sometimes the separation of these periods from each

Welj testing with measured sandface flow rate can be otier is impossible; particularly, the effects of bottom-

traced to tbe beginning of reservoir engineering. The rate water influx andtor gas cap may start during the. middle-

must + measured over time to calcuk and{or approx- time period. Thus, the semilog approach sometimes can-
imate constant rate to obtain even a single reservoir not be app~,ed at all.

parameter from pressure measurements. Tfds approiimatc Furtbeirnore, the drawdown or buildup tests as con-

.comstant rate has been sufficient for estimating


ptmneabili- ducted today tend to homogenize tie reservoir behavior.

ty, skin, and initial formation pressure during the ra&d In other words, ” most of the reservoirs behave

infinite-acting period. During tbk period, the well should homogeneously during the storage-free radial intinite-

produce at? conkant rate at the sandface or at a zero rate acting period because most of the heterogeneous behavior

if a buildup test is condu~ed. Because of compressible takes place during the early-time period.

fluid in the production string (weflbore storage effects), The &pe-cume approaches have been introduced to

it (alces a long @e to reach the radial infinite-acting overcome some of these problems. The theories, applica-

period. The effect of outer boundaries also may stat tions, and elaborations of the type-curve methods, as welf

before the end of the wellbore storage effects. as WY” references, can be found in Ref. 1. In 1979,

In gener~, the storage ~pacity of the wellbore, Gringarten ez al. 2 introduced new type-curves that use
wellbore geometry, rim-wellbore complexities, and ex- different pammetetiation than the earlier ones, namely
ternal boundaries affeqt transient bebavior of a well. Dur-
Rhmey, 3 Agar.val et al., 4 McKinley, 5 and Eafkmgher
ing the. aualysis of pressure-time data, each of these and Kemch6 types. All the type curves presented by these

phenomena and its dtition must be recognized for the au~ors, and many others, were developed under tie
assumption that ~e fluid compressibility (density) in the
aPPEcatiOn Of SeMilog and typecurve techniques to deter-
mme formation flow capacity (k/z), @mage skin, and tubing and anmdus remains constant during @e test period.

average formation pressure. The influence of these During the early time, pamicularly for buildup tests, shut-

phenomena on transient behavior of a wg?fl progresses over in pressure incrkases ve~ rapidly; thus, the compresiibJl-

time. For the sake of convenience, the test time can be ty is usually higher than the compressibility of the fluid

divided into ~ periods according to which phenomenon in the reservoir for producing wells. Since the piessure

is affecting the pxessuze. These periods are defined as in the wellbore is a fgnction of the depti, the com-

follows , pr&ibfity of the fluid at the wellhead can be 10 or even


100 tink geater than the compressibility of the fluid at

EarIy-T@ Period. The combmed effects of wellbme the @ttOm. T@, the assumption that the ‘wellbore storage

storage, +nmge skin, and pseudoskin (which include par- coefficient is constant during the drawdown, and pal-
ticukrly during buildup, may not be correct. A variable
mwmt f9.33 sww .f P.t,.h+ EWW.IS weflbore storage coefficient alone makes the application

REBRU.4RY 1985 323


of type-curve methods afmost impossible. The combina- More recently, Metier et al. 17 have used sandface
tion of variable or even constant wellbore storage with flow measurements with pressure data fo? buildup test
wellbore “geometry timber complicates the type-curve amafysis. This has been the first successful attempt to use
~tching process. Moreover, wellbore pre.wure &ta afone direct measurements of sandface flow rate data in weU

tiY not indicate changing wellbore storage. testing. They showed that the Homer method can be
Some of the problems inherent with the use of type- modilied to ‘&ach a semifog stmight line earlier km type-
cume methods can be efiiated by the sbnuftaneous use curves or the 1 I%-cycle @e irdcates.
of measured sandface flow rate and pressure data.
‘fhe purpose of this work is to study the use of the Theoretical Developments
measured sandface flow rate in a broad sense with regard During the last 4 or 5 decades, many solutions have been
@ transient weU”testing. Furthermore, we explore the use developed for transierit fluid, flow through porous media.
of convolution and deconvolution $1 the interpretation of me superposition theorem (Dahamel’s tieorem) has been
pressure beliavior of a well with afterflow (buildup case) used @derive solutions for time-dependent boundary con-
or wellbore storage (drawdown case). ditions from time-independent boundary conditions. For
example, the multiple-rate testing is a special application
Background of the superposition theorem.
The use of tie sandface flow rate in transient testing is fn their classic paper on unsteady-state flow problems,
not riew. To oti knowledge, van Everdingens and van Everdingen ~d Hurst’s presented the dimensionless
Hurst7 were the first to estimate and use the sandface wellbore pressure for a continuously varying flow rate as
flow rate to calc~ate the wellbore pressure. To do this,
they approximate the stidface flow rate by the forr+nda
PwD(b)=mJ(f9Pso (b)

. . . . . . . . . . . . . . . . . . . . . ...(1)
qsf=(’-’-”o$
+ jtDqD‘(T)p@ (f~ - ~)dr. . . (2a)
where 13 is a positive constant. These authors stated that
0“
the constant, ~, cm be determined from well aad res?r-
voir parameters. Using tie above formula and the con- An alternative form to E.q. 2a can be obtained by an in-
volution integd, van Everdingen7 and Hursts presented tegration by parts as
an expression for the weUbore pressure with a variable
wellbore storage effect. Gladfefter et al. g presented’ a
method to determine the formation kh from pressure and
PwD(tD) ‘qD(tD)psD@)
aftertlow ikita. The afterflow &ta were obtained by
measuring the rise in the liquid level in the ivellbore.
Ramey 10 applied the Gladfelter approach to gas’ well +~fZD(7h’D&r~)dC . . .. . . . . . ..(zb)
buiklup tests.
0
A considerable amount of work afso has been done on
multirate (vimiible) rate tests dining the last 30 yeys. where
However, these are basically sequential constant-rate
dratidowns; only transient pressure is measured and rate
is assuined. constant during each drawdown test. The
pwD = *1%-PM$l!
techniques related to thk type of multirate tests afso can
be found in Ref. 1. AU the ivork mentioned so far de#s
with the direct problem. In other words, the constant-rote
0.0002637kt
solution (the influence or the unit response function) is ~D =
convolved (superimposed) with the the-dependent inner qipctr$
@~dary condition to obtain solutiom to the diffusivity
equation. This process is calfed “convolution.” psD = PD+L$,

Hutchison and .%kora, 11 Katz et af., 12 and Coates er


PD(tD) = the dimensiodws sandface pressure for
af. 13 presented methods for determ.inin g the influence
the constant-rate case without wellbore
function ~ectly from field data for aquifers. The proc-
storage and sf& effects,
ew of determining g fie influence function is cafled’ ‘decon-
volition. ” S = steady-state skin factor,

Jargon and van Poolen 14 were perhaps the first to use


the deconvolution of variable rate and pressure data to
compute the constant-rate pressure behavior (the inffnence qD (tD) = !hf(tD)fq, ,

fiction) of the formation’in well testing. Bostic et al. 15


qi(b) = dqD(tD)/dtD ,
used a dwonvolution tecbni@e to obtain a constant-rate
q, = reference flow rate—if the stabilized
solution from a variable rate history with a known pressare
constant rate is available, then q,
history. They also extended ~e deconvolution technique
to combine production and buldup data as a single test. should be replaced by q~,

Pascal 1s also used deconvolution techniques to obtain a q~t) = variable sandface flow rate (flowmeter

constant-rote solution from variable rate (measured at the readings) , and


surface) and pressure measurements of a drawdown test. qD(2Li) = dheUSiOdeSS sandface rate.

324 JOURNAL OF PETROLEUM TECHNOLOGY


1.0.
TABLE l—HOMOGENEOUS RESERVOIR ROCK

AND FLUID DATA

8, bbl/STS 1.0
,..5
0,, psi-’
ccl
1,000
h. ft 100

40

2,666,94

3,000

0.35

10,000

0.76 x10-4

p, Cp 0.8

.6 0.2

shut-m Time, AI, hrs

Fig. I—Dimensionless sandface flow rates for constant The purpose of the test welf interpretation, as stated by
tiel Ibore storage and exponential decfi ne cases. z is to identify the system and deter-
Gringarten et al.,
mine its goverrdng psmrneters from measured dsts in the
wellbore and at the wellhead. Thk problem is known as
Although traditionslfy the skin effect is considered a tie inverse problem. Tbe solution of the inverse problem
dimensionless quantity different from the dimensionless usuafly is not unique. As Gringarten et al. 2 pointed out,
formation pressure, the skin effect will be treated here if the number snd tie range of measurements increase,
as part of the imer boundmy condition for the solution the nonuniqueness of the invers.k problem will be reduced.
of a unit rate production case. This boundsry condition Thus, combining sandface flow rate with pressure
is known as the homogeneous boundary condition of the measurement will enhance the conventional (including
thiid kind.. type-curve) weU test interpretation methods.
It shoufd be emphasized that Eqs. 2a and 2b csm be ap- AS m inverse problem, the ssndface pressure, psD(tD),

plied for many reservoir engineering problems. The hss to be determined by the deconvolution of the integral
linesrity of the diffusivity equation allows us to use Eqs. in Eqs. 3a, 3b, and 3c. As stated previously, p,D(tD) is
2a and 2b for fractored, layered, anisotropic, snd the solution for the constant-flow-rate (sandface) case.
heterogeneous systems as long as the fluid in the reser- Taking the Lsplace transform of Eq. 3a and solving for
voir is single phase. F@. 2a and 2b cm be applied to both ~@(s)* yields
dmwdown and buifdup tests if the initial conditions sre
~~D ($)= ~:D ‘s)
known. For a reservoir with m idkd constant snd uni- =, . . . . . ..8. . . . . . . . . . . . . . . . . . (4)
form pressure distribution [pD(0)=O], Eqs. 2a and 2b
can be expressed as
where s is the Laplace transform variable.

‘D
The Laplace Esnsform of P,D in Eqs. 3b and 3C will
PWD(fD)=f q~(7)p@(tD–T)dT . . . . . . . . . . . .(3a) be the same m Eq. 4, keeping in mind that p,D(o) =S.

0 Thu3, we hzve only one operational form of the convolu-


tion intograf given by Eqs. 3a, 3b and 3c. The superposi-
=SgD(tD)+ fDqD(T)p,’D(tD –~)dr. (3b) tion thmrem io this case is notliig more thsn the
0 convolution of P,D (tD) aod qD (tD). The Laplace tmns-
form of the convolution intepal allows us to express the
Furthermore, Eq. 3a sfso can be expressed as convolution integral in msny different forms. Further-
more, the kemef solution csn be a solution of constant-
rate or comtsnt-pre-wure case for the convolution integral.
PWD(tD)= ~’DqA(iD–T)P@(r)dT. . . .’ . . . . . . . . (3c)
If the wellbore storage is constant, the dmensiordes:
.0
sandface flow rste can be expressed as4,1s

fn Eqs: 3a tid 3c, it is assumed that q~(rD) exists. If


dp ~,D (tD)
qD(2D) is constant, then Eq. 3b must be used. ls@. 3a q~(tD)=l–c~—, . . . . . . . . . . .. . . . . . (5)
snd 3C we known as a Volterra integral equation of the dtD
first kind aod the convolution type.
Akhough it is assumed that PO is a constant-rate solu-
where

tion without stomge effect, mathematically and physical-


qD(tD) ‘q,f(t)fqr.
ly it can be a solution of a constant-storage case. In
pmctice, As a special application of Eq. 4, the dimensionless
p$D afways will be affecti by the welfbore fluid
wellbore pressure solution for the constant-storage case
that occupies the volume below the flowmeter u,!dess the
cao be written directly from Eqs. 4 and 5 as
ssndface rate is measured through perforations. However,
the volume below the flowmeter will lx sms.lf for most
~sD ($)
wells, since the flowmeter and the pressure gauge usual- ~wD (s)= (6)
ly csn be placed just above perforations. Therefore, 1+ CDS2F,D(S)’ ““””’”’””””””’”””””

throughout this paper, we will assume that p,D is not af-


Vhmughout lhls paper, the runclian ?(.9 w be Mod the Wlace tram form.! tha

fected by the fluid volume below the flowmeter. f.ncsion F(t).

FEBRUARY 1985 ““ 325


As Ramey 10 noted, i3 cannot be estimated as re@ily
as the weflbore storage constant, CD. However, in prin-
ciple, Eq. 9 has a much more ~ediate cO~~tiOn wi~
the real systems. In fact, the pressure increwes more
-...O.W9 rapidly during the early-time buildup tests; then it slows

,/’,
down during the transition period and boilds up vay sfow-
,,,.”.., !.!

k&
ly during the semilog period. Because of the rapid change
of pressure, the wellbore storage wilf decrease condnuoLls-
Iy except in the case of phase redistribution.
In many cases, it is difficult to recognize changing
wellbore storage effdcts because it is a gradual and con-
tinuous change. Furthermore, tie fifi~ clOsing time Of
~.. ~.z , ~.,
10“ the welffread valve also will affect pressure at tie same
shut-to n.., M, hr.
time.
Most of the work thus Em in early-time analysis has been
Fig. 2—Shut-in pressures for constant wellbore storage and

exponential decline cases.


directed toward the construction of type curves from the
solutions of Eq. 7 for a constant wellbore storage for dif-
ferent wellbore geometries, such as fractured wells, par-
tial penetration, etc. Type curves for Eq. 8 for different
where ~$D (s) is the dimemionIess sandface pressure for vafues of O and skin also can be developed and used for
the constant-rate case witbout storage effect but including graduafly decreasing wellbore storage cases to determine
skin. skin snd W?.
Van Everdingen and Hurst 18 presented an equatiOn Fig. 2 presents a semilog plot of p ~, (At) vs. At,

simifa to Eq. 6, and Aganval et al. 4 presented the same calctiated from Eqs. 7 and 8 by using reservoir and fluid
equation as an integrodifferentid form for radial systems. parameters given in Table 1. A8 shown by thk figure,
Cmco-Ley and Mmaniego 19 {for fractured reservoirs), tle exponential decline case approaches a semilog straight
and Kufmk and Kirwan20 (for partially penetrated wells) line earlier than the constant-wellbore-storage case.
presented the same expression for the dimensionless The most important point to be made frO.m tbe abO~e
weflbore pressure as in Eq. 6. discussion and buildup data prc%ented in Fig. 2 is that the
on the other hand, the dimensionless sandface flOW raE principle limitation of the type-curve analysis stems from
can be obtained directly from Eqs. 5 and 6 in terms of the lack of information about the sandface flow rate
the dimensionless formation pressure as behavior. Thus, the type-mu-w analysis usuafly is used
for quabtive answers and supported by the semilog
1 analysis. For quantitative analyses of the early-time data,
~~(s)= . . . . . . . . . . . . . (7)
St 1 + CDS2ESD (s)1 it is necessary to measure the sandface flow rate. Fur-
thermore, the use of measured sandface flow rate also can

Fig. 1 presents vafues of qD wdculated from Eqs. 1 and improve the semilog anzfysis. In the following s&tiOns,

7 as a function of real time for a buildup test using reserv- the convolution and deconvolution of simultaneously

oir and fluid properties given in Table 1. As can be seen measured sandface flow rate and pressure data will be
tlom this figure, for the exponential&clime case, the samd- shown to obtain the formation pressure and parameters.
face tkw rate declines faster than the constant-wellbore-
storage cwe. IZamey and Agarv.m121 also pcesentd values Convolution (Superposition)

of qD(tD) as a function of tD for various skin and storage Continuous Multirate Method. The simplest approach
constants: to solving the convolution integral is to assume a psD

Another important application of the convolution in- function in Eq. 3a. This psD tb~On cO~d ~ a line-

tegral given inF.qs.’3a and 3C was resented by van Ever- source solution, infinite conductivity vertical fractured
*J
dingen, 7 Hurst, g and Ramey for calculating the solution, etc., for the constant-flow-rate case or the cOn-
wellbore pressure by using Ea.. 1 and the line source stant-pressure case. The chosen pa function can be con-
solution. volved with qD (sandface flow rate) by using the
For a finite wellbore radius, tie dimensionless wellbore convolution integril (Eqs. 3a tkougb 3c) to modify the
pressure solution afso can be written directly from Eqs. time function or pwD. For example, the cOnvOlutiOn
1 and 4 for the exponential sandface rate decline case as (superposition) of variable rate with the log approxima-
tion for tbep,D function and wellbore press~e CCJMMOR-

~wD(.,=y —. ,., ,.,. .,, ,,, ,. .,, . . . . . . . . (8)


ly is used for the anafysis of muldrate tests. The same
technique also can be used for the analysis of buildup or
drawdown tests with measured sandface flow rate data.
‘fhe exponential constant, i3, is giVeII M V~ Eve@gen7 USing the 10g approtition for P.D in f%. 3a! fie

as change in measured pressure in oilfield units can be ex-


pressed as
(3=m#yic,rw210.000264 k,

where u can be determined from welfbore pressure data,


Af)&t) =m@(,)@og(t-r) +~]d~, . . . . . . . . (9)”
such as the wellbore storage constant. Note that (3 is a
dimensionless constant like CD. 0

326 JOURNAL OF PETROLEUM TECHNOLOGY


where As noted earlier, the continuous variable-rate test sug-
gested here will not eliminate completely the effect of the
Ap ~f) ‘pi ‘p tit), wellbore storage, since there is a finite volume between
the bottom of the well and the flow meter, but it will
minimize the wellbore storage effect.

‘=”’’’+’”g(*)-’’’” The
simplifies
method
Eq.
suggested
12. However,
by Gladfelter
it does not
et al. 9 further
improve the
muhirate method. The disadvantage of the Gladfelter et
al. 9 method over the mukirate method is that here is a
pxsibfity of the existence of one to three different straight
lines.
and
Motiled Homer Method. Using the measured sandface
rate &ta, Meunier et al. 17 presented a modification of
162.6 @q
~.— the Homer time ratio, which they named “the rate-
kh
convolved buildup time function. ” They showed that the
time required for the start of the semilog straight line can
Eq. 9 can be rewritten as be reduced consi&rably by using the rate-convolved
buildup time function instead of of the conventional
Homer time ratio. Metier et al. 17 gave detailed ex-
planations on how to modify the Homer dme ratio if the
sandface rate measurements are available. In this section,
the fundamental mture of tie Homer and modified Horner
methods will be examined.
The buildup test is perhaps the most popular transient
where b= ,~m. testing practiced by the oil industry over tie last 3 decades.
Letus approximate the integral in Eq. 9 by the Riemamr The Horner method used for the analysis of the buildup
sum, which yields test is appealiig for its simplicity, generality, and ease
of application. The reliability of kh, skin, and the ex-
trapolated pressure estimated from the Homer method
AP.jc(/n) =
depends on the slope of the Homer semilog straight line.
qD (tn)
The assumption required for the Homer method is that
the sandface flow rate becomes zero during the semilog
period. Flom a theoretical point of view, as long as the
measured pressure increases at the wellbore, the sand-
face rate never will be zero unless the fluid in the wellbore
is incompressible. Thus, rhe effect of the decaying sand-
where t. is the measured (discrete) time point. This equa- face flow rate on the Horner semilog straight line wifI
tion has been presented elsewhere for multiple-rate be investigated in this section.
analysis. Eq. 10 gives the condnuous form of the muhirate The convolution integral given by Eq. 3b can be writ-
(variable-rate) equation. Any integration techniques, as ten for buildup tests as
well as the Riemarm sum used in Eq. 11, can be used to
evaluate the integral given in Eq. 10. A plot of the left PDS(~pD +AtD)=%D(ArD) +PD(t@ +AtD)

side vs. the first term of the right side of Eqs. 10 or 11


will yield a straight line with a slope m and an intercept
-~[1-qD(T)@;(MD-~)d~, . . . . . . . ..(12)
b. For buildup tests, qD(f) should be replaced by
o
1 –qD (At) and AP ~f(At) by Ap ~, (At) =p ~, (At)
–p~At=O) in Eqs. 9 through 11. where
The major advantage of continuous variable-rate test
(rate measure just above the perforation) over the con-
vention multirate testis tit the wellbore storage effects
are minimized. The wellbore storage effect has not been
discussed in the literature for the conventional mtddrate
tests. The wellbore measured pressure used in the con- and q is the constant rate. before shut-in.
ventional malysis should be taken from the storage-free For tie sake of simplicity, let us assume that after shut-
in ftite-acting period. Thus, the end of the storage effect in, the afterflow rate decliies exponentially, as suggested
should be determined. In other words, the sandface rate by van Everdingen6 and Hurst7 so that Eq. 1 becomes
should be equal to the constant-surface rate for each
drawdown or buildup period. qD(At) =e ‘-&, (13)
The second problem with the conven!iond multirate test
is the surface step rate change cannot be taken as a step where a is determined from measured sandface flow rate
rate change for the sandface. Neglecting the continuous data. For e~ple, if sandface rate measurements are
rate change from one rate to another will affect the result available, a in Eq. 13 can be determined by the least-
of the conventional multirate test analysis. square.s curve fitting of Eq. 13 to the sandface rate data

PEBRUARY 1985 327


Fig. 5—OimenslonIess times for the start of the Horner and

modified Horner semilog straight lines as a function

of 0.

tremely long time period. It is obvious from Eq. 14 that


as At increases, 1/2. 303cYAi becomes smaller, but log
[[tP +Ar)/At] decreases as well. Thus, the Homer plot,
as it is seen in Fig. 3, asymptotically approaches the cor-
Horner & Modified Horner Time Functions
rect semilog straight line.

Fig. 3-Modified Horner and Horner plots for exponential


If the term (ZP +Ar)/At is very large compaed to the
dec~ne sandface rate, term l/2.303aAt, then the Horner and modified Homer
straight lines are almost identical at large At. In other
words, if tp is veiy large compaed to the maximum shut-
in time, then the correction caused by the aftertlow
becomes almost negligible as At becomes large.
Before determining g an approximate formula for the stint
of the modified Homer sefiog straight line, it will be
interesting to determine an approximate formula for the
start of the Horner sem-ilog straight line. Theoretically
speaking, the Homer semi30g straight line never yields
the correct straight hne. However, we can define an er-
ror criterion between the correct and computed Homer
slopes. Then, we. take. tie time at which the error criterion
is satisfied as the start of a Homer semfiog straight line.
As developed in Appendix B, m approximate formula
for the start of the Homer semilog straight line is given by

Fig. 4—Functions for the start of the Horner and modified


2ctPD –AtD[(rF +At)D]
1 @ ‘btiD [–in p –27

Horner semilog straight tines.


+-In 4+EXJ3AtD)+2S]- 1/AtD
1 =0, . . . . . .. .(15)

where

Substitution of the exponential integral solution forp~


and Eq. 13 for qD in Eq. 12 (the details of the deriva-
tion are given in Appendix A) yields

an error criterion,

pi-pw’’’)=~[’”g(%)+al (14)
The constant a in Eq. 14 can be considered an afterflow
%., = slOpe of correct Horner semi-log straight
parameter. The term l/2.303cYAt will modify the Horner
lie,
&me ratio, (tP +At)/At. A semilog plot of pw~ (At) vs. An-
tilog {log[(tP +At)/&] + l/2.303aAt} wilJ yield a ~ght m~~~ = slope of computed Homer semilog stmaight

liie with a correct slope. This straight line starts much line, and

earlier than the Homer “setnilog straight line, as shown tpD = dimensiOIdess producing time.
in Fig. 3 (approximately one-half cycle earlier). In fact, For a given e (relative error), 6, skin, and producing time,
Chen and Brighamz found that the correct semilog zeros of Eq. 15 with respect to AtD will give the sw
straight liie on the Homer plot is obtained after an ex- of the Homer semilog stmight line. It is interesting to

328 JoURNAL OF PETROLEUM TECHNOLOGY

—-—
observe that the start of the semilog Homer straight line other words, the time required for the stat of the modified
is a function of the”producing time, as expected. Homer stmight line is half of the time required for the
A simple formula cannot be derived for the start of the Homer straight line, as can be seen from Eqs. 16 and 19.
Homer semifog straight tie from Eq. 15 because it is Depending on the formation and fluid parameters, hours
a transcendental function. Fig.. 4 presents values of could be saved on the testing time.
9
15 as a” function” of dimensionless time for tPD = 10 , Very simple qD and pD functions me used to explore

(3=10-4, S=0.0, and E= O.01. As can be seen from Fig. the effect of afterflow on the Horner analysis. It must be
4, E41. 15 has two roots (zeros) for the values of tPD, 6, recognized that the sandface flow rate (afterflow) has to
S, &d e given previously. The first zero resufts from the be measured to obtain accurate and reliable restdts from
early-time period, which can be seen easily from Fig. 2. the modified Homer analysis.
The second zero results from the radial inkinite-acting Even though the modified Horner anafysis improves the
period..l%e upper curve in Fig. 5 presents dimensionless semilog anafysis, it cannot be applied to very early-time
time for me start of the Homer semilog straight line as pressure data. The otier methods, such as continuous
a function of P for tPD = 10 6, S=0.0, and 6=0.1. As Cm mukirate, require prerequisitepD functions. Thus, decon-
be seen from Eq. 15, the dimensiordess time for the stwt volution methls will be used to obtain p,D functions (in-

of the Homer semilog S-might line is a very wesk func- fluence functions) and formation parameters in the
tion of skin. It is basically a function of the production following section..
time and IS. This had been observed by Chen and
Binghamzz for tie constaht-wellbore-storage case. Eq. Deconvolution
15 slso can be used for the constant-we~bore-storage case Deteminin g wellbore geometries and reservok types
by substituting lICD for 13. However, UCD is a ve~ (fractured, layered, composite, etc.) is an important part
crpde approximation for “B. of well testing. The reservok engineer must have stiffi-
Eq. 15 cm be quite useful for the design of buildup tests cient information about the system being analyzed. For
for an optin@ value of a producingdme to achieve a cer- example, if w curves for fully penetrated wells ae used
tain accuracy for the Homer semilog straight line. Par- for partially penetrated wells, both kh and damage skin
ticularly, Eq. 15 will be usefuf for driflstem tests, since will be underestimated. Thus, the system identilcation
production time is limited by production facilities. becomes a very important part of well testing. For in-
For large producing tie, tPD, Eq. 15 can be simPlifi~ stance, identifying a one-half slope on a log-log plot of
to tie pressure data will indicate. a vefically fractured well,
as two paral!el straight limes on a Horner graph wi13 in-
AfD=~. . . . . . . . . . . . . . . .. . . . . . . . . . . . .. (16)
dicate a fractured resemoir. However, either wellbore
storage or’afterflow” usually dominates these chamcteristic
l+q. 16 also can be derived from drawdown solutions
bebaviors of we.ffs and reservoirs durtig the early-time
by using the same principle given in Appendix B.
period. Thus, the pressure behavior of formation without
If p i.iapproximated by I/CD, Eq. 16 then can be writ-
the weflbore storage effect must be cahdated or the
ten as
wellbore storage effect on the formation pressure should
AfD=:CD. . . . . . . . . . . . . . . . . . . . . . . . . . . ..(17) be minimized for the conventional identification of the
system. Tbis is merely the deconvolution of F@ 3a
As noted previously, Eq. 17 will yield very optimistic tlqough 3C to calculate p,~(r~) from P WD (tD) and

values for tie stat of Homer semilog straight line if in- qD (rD). seve~ graphs Of PsD(t~) VS. tD , such 2$ liim,
deed the weflbore storage remains constant during the test. sphericsl, etc., will provide information about a given
The dimensionless time for the start of the motied wellbore geomet~ and reservoir.
Homer semilog stmight line (details of derivations are This approach to system idendi5cation is not genersl,
given in Appendm B) also can be given by but it uses our conventional knowledge about well and
resemoir behaviors. In general, the problem of the system
e[2@(tPAt)D +(tP+At)D] -b’2AtD2[(tP+AZ)Dl
identification is much more complex because often we do

.e-8~D(–1” &7-27+fII 4+2S)=0. .(18) not fmow the governing differhisl equation.
It is worth repeating that the fluid flow in the forma-
As in the Homer case, tie second root of Eq. 16 will give tion is described by the linear dh%sivity equation for the
tie dimensionless time for the sw of the modified Homer deconvolution methods given next.
semilog straight lime as a function of 8. Fig. 4 presents
vslues of Eq. 18 as a function of dimensionless time for There are several methods for tie deconvolution of Eqs.
tPD=’106, “13=10 -4> S=0.0, wid 6=0.01. 3a through 3c. These metiods will be discussed in the
The lower curve in Fig. 5 presents dimensionless time following section.
for the start of the modified Horner semilog straight line
w a fUIItiWI of j3 for rPD=106, S=0.0, and c= O.01. The
Lineariza tion of the Convolution Integral. In this sec-
sw of modified Homer semilog stmigbt we is also a tion, the “convolution equation,” Eq. 3c, will be solved
weak function of skin. directly by using the limaiization niethod. Eq. 3C can be
For large producing @ne, tPD, @. 18 cm 5iMP~i~ ~ discretized as
n
AtD=:, . . . . .... . . . . . . . . . . . . . . . . . . . .. (19)
PWZND”+l)= z ~“mlti%+l -~)
4@3
i-o ~Dt
The modified Horner semilog straight line star@ at least
one-half cycle earfier than the Homer straight fine. fn “ps~(~)d~. . . . . . . . . . . . . . . . . . . . ..(20)

FEBRUARY 1985 329


,6,0-

=
:

‘“:”’’’””:;:L:L---- “-”-” :,200

il’’’’”F”’-’F
i . . .. . .“..,. !. ah”..!.
‘.”., !,. s,..,.!.

, m, -
+ ./”/
., ./
. .0,
/“
:

i
-”

~’”:~,~l
--.
.-.
,
0 ..
10 ,o- ,0- 1,.’ d
,,!! ~~~~~~~~~
SW-i- ‘m., .!, b

,o- ,0-’ , 0“
Fig. 7—Calculated formation pressure drop using Hamming
slut-l. n.,, At, hrs
method and wellbore Pre?sure drop.

Fig. 6—Calculated formation pressure drop using linearization

method and wellbore” pressure drop.

where “sum” is equal to


n—1
By using the trapezoid rule for integration in Eq. 20,
,2 psD(bi+fi)[qD(tDrI+l ‘b)
n

PwD(tDfI+l)= X bSD(tDt+I)Qi(tDn+I ‘tDi+l)


‘qD(tDn+I ‘tDi+I)l,
i=!l

and the fist value ofp ,D is given by


+P.D(tDi)q~(2Dm+ I –2Di)l(tDi+I ‘2Di)/2. . . (21)

Eq. 21 givea a system of linear algebmic equations. The PsD(tD%)=:;;;;


— . . . . . . . . . . .. . . .. . . . . . ...(24)

coefficient inairiix of this system of equations is a lower


triangular niatri~ that is, all its nonzero elements aze in As can be seen from Fig. 7, Eq. 23 gives a stable and
tie lower right of the mat@ The system of equations nonosdlatory integration scheme for the convolution in-
can be solved easily ‘by forward substitution. tegral given in Eq. 20. Furthermore, the derivative of the
For field data, p ~D sho~d be replaced by fp i –p wj) sandface flow rate data, q~(tD), is not needed in Eq. 23.
and (p w, -PJ for drawduwn and b~dup tests, respec- However, for Eq. 21, q;(t) must be known. A iinite dif-
tively; and p,D should be replaced hy (pi –pw)f and ference appromixation also can be used for q&D), but
Q% –P&, ~bich we s@dface press~e +fferenc= some accufacy would be lost.
without storage effecm (formation pressure drop). qD fur The semilog plot of @ ~, –p ,,.,+)f vs. At, given in Fig.
drawdown and (1 –qD) for buildup must be used to 7 for radi+ synthetic buildup, test data, yields a straight
replace qD in Eq. 21 for ~eld case ch. If tie flOW ~S line starting kom a veiy early time (At=O.05 hours) @tb
radial, formation kh can be determined from the slupe of a correct slope. The lower curve in the same figure is tbe
the straight line if the (p ~, –pM)f vs. log’Af plot. Skin plot of (pW, –p;), which includes the we~bOre it~rage
factor also can be “determined from me conventiowf skin effect.
formula. However, the trapezoidal method used in Eq.
21 gives osciffatory result.% As can be seen in Fig. 6, at Laplace Transform Decunvulntiun. The convolution of
very early times, the cal~ulated values of @w. –P ~)f RI. 4 yields
oscillates. Furthermore, iugher-urder mefiods, inclu~g
‘D
the SiniPson rule, yield divergent results for the integral
pSD(tD)= \ K(T)PWD(tD –7)d~, . . . . . . . . . . . . (25)
in Eq. 20. ..” .’ .. . ~
Hamminz23 sumzested a stible integration scheme for
evaluation ~f the ‘%wolntion integrul~. He also showed where
that direct integration (discussed as the linearization
method)
owiflation.
of convolution integmfs usually will r@t in w
lr(z~)=c-1
— . .................
M
“1
(26)
s?D ($)
The intigral in Eq. 20 can be approximated as
K(rD) can be cumputed eifher from the Laplace
transforms of qD(tD) data or a curve-fitted equation uf
qD(tD) data. However, it would be @e consuming to
invert all the gD(tD) data in Laplace space and transform
it back to real space in accordance with Eq. 26. Thus,
‘PSD(~Di+%)~1q~(2Dn+l-7)dT. . . . . . . ..(22)
aPProhtion functions must be used for qD (tD) data.
r~i
Once an approximation is obtained, it will be easy m com-
The right side of Eq. 22 can he integrated directly.
pute K(tD) and integrate Eq. 25 to determine p,D(tD).
Substitution of the integration resuhs in Eq. 20, and solv-
A few types of approximation functiuns can be used to
ing for p,D yields
approximate qD (tD). We have tried rational functions,
p,D(tDn+,A)=~”~(f~+l)–s~ power series, and ex~nentid functions. Expmentid func:
,. .,.... . . . (23)
fZD(tDn+ I ‘tD.)
tions give a good representafiori of qD (tD) data. Ex-

33U JOURNAL OF PETROLEUM TECHNOLOGY

.—

TABLE 2—FRACTURED RESERVOIR ROCK TABLE 3—COMPARISON OF

ANO FLUID DATA COMPUTED AND ANALYTICAL

PRE?SURE FOR A FRACTURED

B, bbl/STB 1.0 RESERVOIR


,..5
~, psi-l

1,000 At
~ .- *
h,Dft 100 131.96 65.33
0.02 131.93
k,, md 400 133.78 98.32
0.04 133.57
pti, psia 4,216.94 134.86 115.30
0.06 134.66
q, STB/0 3>000 135.64 124.37
0.08 136.47
0.35 136.25 129.44
rw 0.10 136.12
R 10
0.20 138.27 138.27 137.03
tp, hours 10,000
0.30 139.62 139.55 138.87
,.-5
k (spherical matrix used) 140.53 139.83
0,40 140.62
~, Cp 0.8
0.50 141.42 141.33 140.85
+, 0.2
0.60 142.10 142.02 141.59
. 0.06
0.70 142,68 142.62 142.24

0.80 143.20 143.15 142.81

0.90 143.66 143.63 142.32

1.00 144.08 144;07 143.78

ponentiel functions approximation for qD can be 2.00 146.94 146.98 146.84

3,00 148.!36 148,70 148.60


expressed as
4.00 149.92 149.92 149.85

5.00 150.88 160.86 150.80

6.00 151.66 151.63 151.59

qD(tD)=cleB1tD +c2e82tD +...cne8~tD, .:. (27) 7.00 152.% 152.29 152.24

8.00 152.85 152.85 152,81

9.00 153.33 153.35 153.32

where i= 1,2. .,.n.


After determining Ci xnd L3i from qD(tD) data, apply-
ing the Lapktce transformation to Eq. 27 and substituting represents the behavior of p,D. As noted earlier, the 10g

the resulting expression in Eq. 26 yields “.. apPrOfimatiOn was used to approximate p,D (tD). In tis
case, the natural choice will be the power series of in tD
such tbatzz

1.” m
i?(y)= ~ , “ ..!:.’: ””””.”.:.:”’: ’(2. !)4
“’””” >sD(tD)= x C@ I tD)i-l. . . . . . . . . . . . . . . ..(31)
. .. . . .—, {=1.
,.,

Substitution of Eq. 31 into Eq. 20 yields n number of


equations and “m number of unknown pxmneters,
AS SugEeSM bv Sneddo”. 24 if the inverse tramfOrma- .=(cI ,c2 .:.c~)T. In fact, obtxining these paramters, .;,
becomes an unconstrained opdmiiation problem. We have
tion d~~s not &fist, Eq. “28 should be divided by”p until
its inverse’ trxnsform is possible. For each ~vision ofp, appEed a fotirtb-degree polynomial, &f. 31, to weIfbore
pressure and rxte data (synthetic) that were obtained from
P~D(tD) fi Eq. 25 has to be differentiated.
a fmctured reservoir for the resewoir and fluid pam.meteis
We have seen from vw Everdngen’s7 and Hurst’sg
gjven in Table 2. The second column of Table 3 presents
works thxt qD cti be approximated by m exponential
fonction as the cxkdated values of Apti from curve-fit approxima-
tion for the fluid and reservok date given in Table 2, while
qD=l_e-@D the third column of Table 3 presents the vxlues of Ap,f
,. ..,. . . . . . . . . . . . . .. . . . . . . (29)
calculated directly from q amlyticel solution without

which is the simplest form of Eq. 27. For this case, P$D storzge effect. The fourth column in Table 3 presents

cm be wriiten directly from Eq. 3 as &~, with the wellbore storage effect.
As seen from Table 3, dMerences in Ap,f from the
curve-fit approximation and the anxlydcel solutions are

mm=?’”’”;’ ‘+PwD(tD). . . . . . . . . . . (30) xlmost identicel. The relative error decreases for lerge
times.

From Eq. 30 it is very simple to calculate p,D if qD dat6 C0nc1u8i0nx

cm be approximated suc.csss~y ,by .,,using ,:.Eq. 29.. 1. The convolution integml (superposition theorem) is
qD (tD ) data ds.o can be approximated by piecewise ex- used for the analysis of continuously varying wellbore
ponential fun~om for a selective interval for Eq. 26. flow rate and pressure. This analysis is ve~ similxr to
the conventional mukirxte meth0d3.
Curve-fit App:oximmtioiis’o fp,D”. p~~(tD) in Eqs. 3a 2. Wellbore storage (afterflow) effects can be present
and 3:..c~wb&approximated by choosing suitable func- to a significant degree in the Homer semi-log strxight Iiue.
tion~ approximations. These ~ctionel approximations The Horner an~ysis is modified by using measured sxnd-
copld be power series, continued functions, rational func- face flow rate data to obtain a correct semilog straight
tions, or exponential functions. The success of this method liie. The modified Horner semilog straight line steits”xt
depends on how well the approximation function last one-half cycle earfier tlmn the conventional Horner

FEBRUARY 1985 331


straight line. Approximate formulas afe presented for the tPD = dimensionless production t@e
stat of the modified Homer and Homer semilog straight T = transpose
limes as a function of sandface rate decline, production At = running testing time, hours
time, and the relative error between the correct and com- ArD = shut-in time, dimensionless
puted s30pes.
a = O .000264kLWpc,r;
3. The formation pressure (influence function) can be
# = a positive constant
cakulated ffom the deconvolution of measured wellbore
.y = 0.57722 ... = Euler’s constant
pressure and sandface ffow rate data. Some new decon-
p = viscosity, cp pa-s]
volution techniques are introduced to compute the format-
f = rate-preasufe convolved time function
ion pressure (iiuence function) without wellbore storage
~ = dwmny integration variable
(atlertlow) effects. W&bore and reserv@r geometries ca
be identified from this computed fofmation pressure. Fur- + = porosi~, fraction

thei’more,, the conventional methods can be used to analyze - = LaPlace transform of


this computed formation pressure to detefmine formation
parameters. References
4. Deconvolution of synthetic data from a homogeneous
1. Earlo.gher, R.C. JI.: Advances in well Test Analysis, Monograph
and ffactured rekervoir shows tiiat it is possible to com- Serk, SPE, Richardson, TX (1977), 5.
pute the formation pressure from tbe”beghing of the test. 2. Gri”gamen, A.C. er .1.: ; ‘A Cmnpmism Bctwee. Different Skin
and Wellbore Storage Type Cm’ves for Early-Time Transient
‘IW computed pressure reveak the characteristic behavior
~Y@,” pawr SpE 8205 p=nted at tie 1979 SpE AIInuaJ
of homogeneous and fractured reservoirs.
Tedmicaf Confer . . . . and Exhibition, Las Vegak, Sept. 23-26.
3. Ranmy, H.J. Jr.: “ShorGTime Well Test Dm Inteqxctation in e.
Acknowledgment
time of S!dn Effyt and WelJbore Storage,,, J. Pet. Tech. (Jan.

We thank Scblumberger We! Services for its peffnission 1970) 97-10% Tram, AJME., 249.
4. A&u-waJ, R. G., AJ-Hussainy, R., and Rarmy, H.J. Jr.: “AD 3n-
o pub3ish OdS paper aid acknowledge the en~u~gement
vesdgatim of Wellbore Storage and Skin Effect in Unsteady Liq-
and help of Gerard Catala of Servic& Techniques
uid Ffmw 1. AmJMcaJ Treatment,!, L%., A% EW J. (Sept. 1970)
Scblumberger during the initial stage of this work. We 279-90; Tram, , AJME, 249.
alio t@& H.J. Ramey Jr. for providing a solution for 5. McKinley, R. M.: ‘YWeUbore TransmissibiShy from Afterflow-

the integral in Appendm A. Dominat&J Pressure Buifdup Data,” J. Pa Tech. (July 1971)
S63-7% Twns. , AIME, 251.
6. Ead.mgk, R.C. Jr. and Kersch, K. M.: “AnaJysis of Shofl-Time
Nomencfatnre
Transient Test Data bv TVWCIUW Matching,,> J. Pet. Tech (JuJY

B = oil fofmation volume factor, RBISTB 1974) 793-SC@ Tr&. ,-&JME, 257. -“
7. van EverdimgeII, A. F.: ‘“i% S!dn Effect and Its Infken.x on the
[res m3/stock tank m3]
Prcducdve CPpacity of a Well, x,J, Pet. Tech. (June 1953) 171-76;
ct = system total compressibility, psi-1 Tram. , Am.-.. -, 1. QR
..
ma-l] 8. Hurst, MIV.: ‘ ‘Establishmerd of the Skin Effect a“d Its kopediment

C = wellbore ,storage coefficient, bbllpsi t. Fluid Flow into a WdJ Bore, ,7 Pet. E.g. (Oct. 1953) B6-B16.
,9. Gladfelwr, R. E., Tracy, G. W. , and Wikq, L.E, : ‘ ‘Selecting Well%
[m3/kPa] Which V?ii Res.mmd to Fmducticm3dmulation Tresxnent ,’, Drill.
CD = wellbore storage constant, dimensionless and Prod. fi!lC. , API, DaJlas (1955) 117-29.
10. Ranw, H.J. Jr.: “NomDaIcy Flow and Wdl&me Storage Effects
h = formation tilckness, ft [m]
in Pressure Build-Up and Drawdown of Gas Wells,,, J. Pet, Tech.
k = fofmation permeability, md
(Feb. 1965) 223-33; Trcms. , AIME, 234.
K = kernel of the convolution integraf 11, H.tchimm. T.S. and Sikora. V.J,: . ‘A Generalized Water-Drive
Am fvsii., 9 J. Pet. Tech. (Jui 1959) 169-77: Tram.: AfME. 216.
p: = deferential of pD
12. X& D.i., Tek, M.R., kd.Joms,’ S. C,: “i Gene&ized Mcdel
PDS = shut-in pressure drop, dimensionless for Predicting the Pa’fomce of Gas Reservoirs S.bjec4 ta Water

PD(tD) = fOfnJatiOn pressure, dimensionless Drive,>> papr SPE 42S, presented at 1962 SPE Anmd Meeting,
Los .@eles, Oct. 7-10.
pi = initial pressure, psi [kPa]
13. Coats, K,H. et al.: “Determimtion of Aquifer fnfluence Functions
PSD = PD +S = fofmation pressure includlng From FieJd Data,’- J. Pet. Tech. (Dec. 1964) 1417-24 Tram.,
skin, dmensi0n3ess AJME, 231.
14. Jargon, 1.R. and van PcalJen, H.K.:. “Unit ResFon% Funcdon From
P WD = pressure drawdown, dmensicmless
V&nz.R?@ Dam. ‘y J. Pet. Tech. (Aug. 1965) 965-69; “Trans.,
PW = bOttotiole flowing pressure, psi [!-@a] tiE,-234.
P~. = bottomhole shut-in pressure, psi ~] 15. Bred., J.N. et af.; ‘, Combined Analysis of Pmtfracmrin’g Perfor-
mance and Pressure BtdJdup Dam for Evaluating an MHF Gas
q = stabflkid constant ate, STB/D
Well,,, J. Pet. Tech (Oct. 1980) 1711-19.
[stock-tank m3/d] 16, PascaJ, H.: ‘ ‘Advanm in EvaJuatinE Gas Well Deliverability Us-

qD = s@face flow rate, dimensimdezs ing Vtible Rate Tests under No”-Darcy Flow,’ 3 paper SPE 9841
presented at tie 1981 SPEIDOE LOW Perrneabili@ Symposium,
qr = reference flow rate, B/D [m3/d]
Denver, May 27-29.
qR = resewoti flow rate, BID [m3 /d] 17., Meunier, D., Withna@ M.J., ed Steward, G.: ‘Tnterpretatio.

q~ = smdface flow raw, B/IJ [~3/d] of pressure Buildup Test Using IreSim Measnrmnert of Afterglow,”
J. Pet. “TecJa, (Jan. 19S51 143-52.
– wellbore radius, ft [m]
~w —
s = Laplme m~~fO~ vfiable
Trans format& to FlrJw Pmble& in R&voirs,x, Tram., .&E
S = skin factor (1949) 1S6, 305-24.
19, CkO.bY, H. ~d sanxmieso, F.: ‘aPressure Transient AmlY$is
t = time, hops
for Naturally Fractured Reservoirs,, s paper SPE 11026 presented
t~ = time, dimensionless at the 1982 SPE AnnuaJ Te&dcaI Co” femnca and Exhibition, New
tp = production time, hours Orleans, Sept. 26-29.

332 JOURNAL OF PETROLEUM TECHNOLOGY


.


20. K@uk, F. and Kirwan, P. A.: “New skin and Wellbore StoraSe where
TYCX CUweS for PardallY Penetrated Wells, ” paper SPE 11676
pms.nmd at fie 1983 SpE c~.~ Region~ Meeting. venti~,
y= O.5772
hfarcb 23-25,
21, Kamey, H.J. Jr. and A&val, R,G. : ‘ ‘Annulus Unloading Rates
as Influenced by W.lhm WOW. and Skin Effect,’+ .%c. Pet. Ew. and
J. (Oct. 1972) 453-62.
22, Chin, H.K. and Brigham, W,E.:’ ‘Pressure Buildup for a Well with
Storase and Skin in a Closed Square,,, J, Pet. Tech. (Jan. 197S) Et(aAt)= \8M:du.
14146.
-m
23. Hammm g, R. W.: Numerical Mefiods for Sckvu&s anAEn@zeers,

McGwv-HiU, NW York City (1973) 375-77.

24. %eAdcm, I, M.: & Use of huq’ml Trmqfom, McGmw-3fiU, New Neglecting the imaginary term .i, the van Everdingen7
York City (1972) 207–14. and Hursts forms are obtained as
25. Stehf..st, H.: “Ntuncricaf fnversion of Lapk.ce Transforms,” Com-
numications of the ACM (Jan. 1970) 13, No, 1, &OriIhUl 368.

&(AtD)= –%e-@MD [–III j3-2-y+fn 4


APPENDIX A

Substitution of the exponential integral solution for pD +Ei(@tD)]. . . . . . . . . . . . . . . . . . . . . ..(A-7)


and Eq. 13 for qD in Eq. 12 yields

1 The values of integral :(&D) from f3kf. A-7 were com-


pSD[(tP+&)D]’ ‘~~i
[
‘—
4(tP +At)D 1 psred
munericaf
with the
Laplace
vafues
trsnsform
computed
inversion
from the
technique
Stehfest
using
z

Eq. A-3. When AtDs 30, the difference between two


vslues of @tD) becomes less than 1%. The difference.
.%E,(.&)-j””e-Bexp[ -&] becomes smaller as AtD increases. Substitution of the log
0 approximations for the exponential integrafs given in
Eqs. A-I and A-7 for the integral yields (in pmcticd units)
dr
.—+SqD(AtD), . . . . . . . . . . . .. . .. (A-l)
2(AtD -,)
~-pw(~)=~[l”g(%)+~(A’)l,
-(A-8)
where

where
~D=e —d==-LmrD. . . . .. . . . . . . . . . . . . . . ..(A-2)

Thus, tie two forms of gD wifl be used interchangeably.


i(At)=*e ‘aAr[–fn f3-27+fn 4+ Ei(aAf)
The integral in Eq. A-1 cannot be integrated readily and,
unfortunately, van Everdingen7 and Ramey 10 dld not
present the &tds of integration. Ramey* provided me
+~. . . . . . . . ... . . . . . . . . . . . . . . . . . . . ..(A-9)
a heurirtic derivation that wss edited out of his paper. 10
Rafney’s integration method is given below. The LapIace
The long-time approximation for g(At) maybe obtained
transform of the integraf given in Eq. A-1 is
by substituting limiting fbims of ci+tain terms in Eq.’ A-7
for long times. For lsrger values of At, the term
e ‘@(-bI j3-27+ln 4+2$ in Eq. A-9 approaches
Z(S)= *KO(J). . . . . . . .. (A-3)
zero. The term e ‘aNEi(aAt) can be approxinsted as

The long-time approximation for KO(~ is e-a&Ei(aAt)=~, . . . . . . . . . . . . . . . . . ..(A-1O)

Koc~)=–lny+y.
(v . . . . . . . . . . . . . . . ..(A4) when &+l.
Substituting Eq. A-10 in Eq. A-9 yields

Substitution of Eq. A4 in Eq. A-3 yields ~(At) =~ (A-11)


2.3026dr” ““”””’”””””’’”””””””

1
j(s) = –— (h s–in 4+2T). . . . . . . . . ..(A-5) Further substitution of Eq. A-11 in Eq. A-8 gives Eq. 14
2(s+.0)
in the main text.

The inverse Laplace transform of Eq. A-5 yields APPENDIX B

The dimensionless form of Eqs. A-8 and A-9 cm be writ-


@ro)= –fie-@tiD [–in &~i-2y+fn 4 ten as

+Ei(6AtD)], . . . . . . . . . . . . . . . . . . . . ..(A-6)

‘Ranw H.J, Jr.: P+rsmal eommunkallon S$anfwd U., Stanford, CA (Mm+l 1984). PD(A’)=05[”[-1+’(A’D)]
w
FEBRUARY 19S5 333
where from Eqs. A-1 and A-2 as follows. Let us rewrite Eq.
A-1 in terms of the modified Homer time
~(AtD)= %. ‘~ND [–In (3-2y+ln 4+ Ei@AtD)

+2,g. . . . . . . .. . . . . . . . . . . . . . . . . . ...03-2) f’D@@=”4w%31”


1
The

In{[(tp
differentiation

+At)D]/AfD}
of
yields
Eq. B-1 with respect to +—
26AtD I +%$(AtD), . . ,. (B-7)

where

‘com=0.5++[g(Ad],
..................@-3) $(AtD)=!4e-5ti.c (–fn fl-2y+ln 4+2S). .: .(B-S)

lA$(AtD) in Eq. A-7 is the ordy remaining term that is


where. mcom is the computed dimensionless slope not included in the modified Horner time. Thus, the
of the Homer semifog straight line and x equals relative error of slope of the computed modified Homer
ln[(f,n +At)D/AtD] . semilog straight line (as in the Homer case) can be ex-
Let us &fine a relative error for tie computed slope as pressed as

df(AtD)
~. —...- . . . . . . .. . . . .. . . . .. . . . . . . . (E-9)
~= I
=Ij ...... .... ......... . (B-I) Ax

where

where

mcom> 0.5.
‘=1”[(%%]++ ‘@-’o)
Differentiation of Eq. B-8 with respect to x and substhw
Substitution of Eq. BzI in Eq. B-3 yields
tion of the result in Eq. B-9 yields

df(AtD)
~= ~~
—.. . .. . . . . ... .. . . . . . . . . . . . . . . . . . (B-5)
Ax

.(–in 6–2y+ln 4+2S). . . . . . . . . . . . .(B-11)


Taking the &rivative of $(AtD) with respect to x and
substituting in Eq. B-5 yields Eq. B-11 gives the dimensionless time for the start of the
modified Homer semiIog stiaigbt line as a fiction of e,
o, tPD, and skin.
AtD[(tP +M)DI
~= A3e-~&D [–1. 13-2-y+ln 4
2tpD [
S1 Metric Conversion Factors

bbl X 1.589873 E–01 = m3


Cp x 1 .“* E–03 = Pas
ft X 3.048* E–01 = m
+.Ei((3AtD)+2S]-~
.1 . . . . . . . . . ...@-@
psi x 6.894757 E+OO = kPa

Eq. B-5 gives “tie dimensionless dme for the start of the
.Gmversl.m Iaclor is exact JP’f
Homer semilog straight line as a function of e, 0, fPD,

and S. 0ri@n=4 rmnusuiDt rec.stved in the Satiety of Pe,,%um Engineers ollb Od, 5, ? 9S3.

Pew accaPted for P.MWO. M.,ch 11. I ss,. ~~vi,ed manuscript ,ece~ved ‘e*
A formula for the dimensionless time for the start of
14, 19%. PaPer (SPE 121771 first Prem.ted at the 19S3 SPE Annual T@ch”icnl Con.

the modified Homer semilog straight line can be found term. and mhlbitlo. held in S.. Franckm 0~. ~s.

334 JOURNAL OF PETROLEUM ~CHNOLOGY


4aoo.o

4a@o.o
LEQEND I
horner
_ mo ,dKde l!w!!,e~
4280.0

4970.0 —. —

4ZSOm0

\
4260.0 . — —

4240,0
Y ‘\
\\

4aao,o — .
\

\.
4220,0 —

\
‘\
4210s0
T

4aoo.o
lb’
HORNER & MODIFIED
Fig, 3-Modlfled Horner and Horner plots decllne sandfaco rate,

)2/77
1800.0 . .

1700,0

1600,0
II Ill II

1600,0
P

1400.0

Iaoo.o 1

TTll 111
1200s0

LEGEND
G wdlbOre DKM3SUWi
0
_m!M?lm.EE !E!M!!u
1100.0

1000.0 .

1’

Fig, 4-Calculated formation pressure


dropusingIkwkatkm
method and wellbore pressure drop,

1A/7,7
m
1600,0

?Emh n [ II’
1400,0


0

I
1s00.0

1200,0

1100,0

100 .0 1--1
10-8
TIME, HOUR8
10’ 10R

Fig, 5-Calculated formation pressure drop using Hamming method and wellbore pressure drop,
aooooo . .

2800.0 —— —

LEQEND
o formation pre8sun3
, ixessurp_
LY?s!!!l ore
2600.0 —

— —
@@”—
z S400.O i
a r
)

>
w“
e
g 2200,0 — — —
$

a
an —— — — —
2000.0
z
T
t=
s
x 1600.0 — — —
@
~
d
w — — —
1600.0
a

1400.0 — — —

1200.0 — — —- -

1000.Q — . .

I
Ii)-’

Fig, 6-Calculated formation pressure drop using polynomial approximation and welibore pressure for a
fractured reservoir,

You might also like