Você está na página 1de 8

Variable Flow Rate Reservoir Limit Testing

Robert C. Earlougher, Jr., SPE-AIME, Scientific Software Corp.

Introduction
The conventional reservoir limit testl- 4 can be used We shall present here a theoretically sound reser-
to estimate reservoir size if the test well operates at voir limit testing procedure for a cyclic two-flow-rate
a constant flow rate during the test. Unfortunately, situation. Simulated pressure-time data for more
it is not always feasible to perform a test at a con- complicated rate schedules experimentally extend the
stant rate, at Jeast for more than a few hours. This analysis technique to more realistic situations. The
is particularly true when flow rate varies cyclically analysis techniques and suggestions for its use are
as in oil wells that are pumped cyclically, water sup- presented, and an example, using data from an exist-
ply wells, and waste-disposal wells. ing industrial waste-disposal well, is included for
This paper shows how to estimate reservoir size illustration.
from pressure data in wells operating with a cyclic
rate schedule. The method applies to both injection Reservoir Limit Test Analysis
and production wells. The important requirement is The familiar reservoir limit test ' - 1 requires that a well
the cyclic rate: rate must oscillate about a mean be operated at constant rate until pseudosteady-state-
value within some relatively regular period. When flow conditions have existed for some time. Then,
this condition is satisfied, reservoir limit testing is reservoir size can be calculated from
fairly simple. This requirement can be relaxed, in
which case there will be a corresponding reduction - - O.23395qB f
A ~h - ~ ,cut (1)
in the accuracy of the calculated result. This is an moel
important consideration when the engineer has no where: q = flow rate for the well, STB/D
control over the quality of data - for instance, when
the engineer has several years of historical data from q > 0 for production, q < 0 for injection
a well. If he knows only rate and pressure at several m* = slope of the linear portion of the pres-
points in time, he cannot apply a very accurate, sure vs flowing-time plot, psi/hr
sophisticated analysis technique. The method of this
paper allows him to estimate reservoir size from such The slope, m*, is the constant rate of pressure change
data, although the accuracy may be less than de- during pseudosteady-state flow; 111* is negative for
sired. This method should not be used if a better production, positive for injection.
approach is available. Its value lies in providing rea-
sonable engineering estimates when conditions pre- Two-Rate Test
vent testing and analysis under more ideal conditions. Consider the cyclic rate history of Fig. 1; the rate is

With this technique, reservoir size can be estimated from pressure data taken from
either production or injection wells operating on a cyclic rate schedule. So long as the
requirement is satisfied that the flow rate vary cyclically rather than increasing or
decreasing monotonically, reservoir limit testing is quite simple and an engineer can
accomplish it without computer aid.

DECEMBER, 1972 1423


q for time t:..t, then 2q for time t:..t, then q for time t:..t, TABLE 1-SYSTEM PROPERTIES FOR SIMULATED
etc. In the Appendix we show that the pressure data RESERVOIR LIMIT TESTS
taken during this rate history can be used to estimate Geometry One well in the center
reservoir size: of a closed square
Area, A 9 sq mi
2.509 x 109 sq ft
Acph = - 0.23395qB ,cu ft (2) Length of side, L 3 mi
m*ct 15,840 ft
Thickness, h 100 ft
where q is the arithmetic average flow rate over the Wellbore radius, r w 0.5 ft
test period. Eq. 2 is valid only when the well is op- Permeability, k 100 md
erating in what would be the pseudosteady-state re- Porosity, ¢ 15 percent
gion if there were only one rate. The time required Total pore volume, A¢h 3.76 x 10' cu ft
to reach such conditions can be calculated 3 ,5 if the Viscosity, f1. 2.0 cp
reservoir size is known. In practical situations this Compressibility, Ct 5 x 10-' psi"
time may vary from a few days to a few year;; the Initial pressure, Pi 2,000 psi
Formation volume factor, B 1.0 RB/STB
larger the reservoir, the longer the time until the
technique of this paper, or any other reservoir limit
testing technique, can be applied. STB/D, is used with both the 1,700 and the 3,400
Fig. 2 presents simulation results that illustrate the STB/D data points.) Both values agree adequately
validity of Eq. 2. (All simulated results in this paper with the value used to simulate the data, 3.76 X 109
are for production situations; results apply equally to cu ft. The slight discrepancy may be in plotting,
injection, in which case pressure would increase with drawing the straight line, measuring the slope, or
time. To help demonstrate this, we include a field considering points before the beginning of pseudo-
example for an injection well.) To construct Fig. 2, steady state in drawing the straight lines, and thus
we generated pressure data using a technique similar drawing a line with too steep a slope. We conclude
to that described earlier6 for a single well in the cen- that the simulated results agree adequately with
ter of a square. System properties are given in Table theoretical predictions.
1. For Fig. 2, flow rates are 1,700 and 3,400 STB/D
with a period of 10 hours. The points shown are at
More Complex Situations
the end of the rate periods, but there is not a point Although the preceding material is too simplified to
for each rate. All calculated points fall on one of the be applicable in actual situations, intuition says it
two lines shown. To avoid clutter on the figure, some might help. To learn if it would help, we tested Eq_ 2
points are not plotted. Note that there is one pressure- with eight simulated reservoir limit tests. Eq. 2 pro-
response curve for each flow rate; each response is vided successful results in all cases. We describe two
linear during pseudosteady state. For the system of of these tests here.
Table 1, pseudosteady state starts at approximately First, we consider a production situation in which
1,400 hours. Fig. 2 indicates that pseudosteady state the rate changes every 3 hours over a 24-hour in-
may start earlier for practical purposes. terval. The rate sequence is 800, 1,200, 1,600,2,000,
The slore of the straight line for the 1, 700-STB ID 2,400, 2,000, 1,600, 1,200 STB/D. Otherwise, con-
pressure points is - 0.0348 pSi/hr. This corresponds ditions of Table 1 apply. Fig. 3 shows the results of
to a reservoir pore volume of 3.43 X 109 cu ft. The this simulation after the beginning of pseudosteady
slope of the straight line for the 3,400-STB/D data state. There is a separate straight line in Fig. 3 for
is - 0.0364 psi/hr, corresponding to 3.28 X 10 9 cu each rate. Each line has the slope m* = - 0.020 psi/
ft of reservoir pore volume. (Note, in using Eq. 2 hr. Using Eq. 2 with the average rate, q = 1,600
with these slopes, that the average flow rate, 2,550 STB/D, we calculate a reservoir pore volume of

:c ~1600
L PSEUDO-STEADY
I PERIOD -

2q ... ~ o
o 0 0
*
/ m '-0.0348 PSI/HR

cr
.
:f
:::>

:1400
1500 /Q'1700 STBIB

...- .
~
-= ~ 1300
i *
21200~~:;~~:::m::'-:~0:3~64;P~S;I/:HR~~~~;';~~J.
o
.,
/Q'3400 STIlID

11000 200 400 600 800 1000 1200 1400 1600 1800 2000 2200
; .. 2 ;=3 FLOW TIME, t, HR
i=n
Fig. 2-Simulated pressure drawdown data for a two-rate
cyclic history, Rates are 1,700 and 3,400 STB/D,
TIME, ,
changing every 10 hours. = 2,550 STB/D. q
Fig. 1-Two-rate cyclic flow history_ Conditions of Table 1 apply.

1424 JOURNAL OF PETROLEUM TECHNOLOGY


3.74 X 10" cu ft. This agrees well with the actual these data sets.) If no alternative for estimating reser-
volume of 3.76 X 109 cu ft. For the 1,200,1,600 and voir size were available, these numbers might be con-
2,000 STB/D rates, there are two lines in Fig. 3. sidered as passable estimates of the correct pore
One applies when the rate increases to that particular volume, 3.76 X 10 9 cu ft, even though they are 50
rate and one applies when the rate decreases to that percent in error. Nevertheless, it would be difficult to
particular rate. A similar simulation for this system, have much confidence in results calculated from the
but with a rate variation of ± 15 percent, produces data of Fig. 4, since the upper line has a wide data-
essentially the same results. Even though there is point scatter and the lower line has only five data
some random "noise" in the pressure time plot, the points. We do not, therefore, recommend intentional
analysis still works. The same is true when, in addi- data collection of the type reflected by Fig. 4. The
tion, the time interval (period) varies randomly by example presented later in this paper indicates that
-f- 15 percent. In these two cases, best results are this type of data collection may be adequate when
obtained by fitting a least-squares straight line to the data are taken over a period of years. In that situa-
appropriate points. tion, we at least have a large number of data points
For best results pressure data should be (1) col- in a given rate range.
lected continuously rather than only at specified times Fig. 5 shows simulated pressure data for a more
and (2) plotted at a fixed time interval after each general case. Rate varies randomly from 800 to 2,400
rate change. Rates must be known when pressure STB/D and time interval varies randomly from 1 to
data are taken. Fig. 4 illustrates the importance of 5 hr. The average rate is 1,610 STB/D. There are
handling data carefully. This figure shows calculated three sets of data points in Fig. 5; for the upper set
data using the basic rate schedule of Fig. 3 with a (circles), 1,000 S" q S" 1,100 STB/D when the pres-
-f- 15-percent random variation in the rate (resulting sure was determined; for the center set (squares),
in q= 1,599 STB/D), and in the time interval. How- 1,600 S" qS" 1,700 STB/D; and for the lower set (dia-
ever, pressures are plotted every 12 hours (without monds), 2,200 S" q ~ 2,300 STB/D. The lines are the
regard for rate) in Fig. 4 to illustrate data analogous
to recording pressures only two times a day. Clearly,
the pressure data alone are useless. We can, however, .1490-1710 BID
0 • 2100-2200 BID
estimate reservoir pore volume from these data if we 1700 0
0

know the rate associated with each pressure point. 0 0


0
0
To make this estimation, we consider only those 0 0
0 0
pressure points falling within a specific rate interval. 0 0
The solid diamonds in Fig. 4 are pressure points 0
0
0
0
0 0 00
taken when the rate was in the interval 1,490 to 0 /m*".0.0405
• • ."
1,710 STB/D. The solid squares correspond to the 0 0 0 •
rate interval 2,100 to 2,200 STB/D. The line shown
for each set of points is the least-mean-square straight
g 0
0
0
• • 0
:I: 1500 0 0 ."
line for those points. For the 1,490 to 1,710 STB/D g 0 0 0
,.- m*:-0.0452
points, the regression line has the slope - 0.0405 § 0
0
o 0
psi/hr, indlcating a pore volume of 1.85 X 109 cu ft. 0
~

For the points in the interval 2,100 to 2,200 STB/D, 1400


1500 1600 1700 1800 1900
n
2000 2100 2200
the slope is - 0.0452, so pore volume is 1.66 X 109 FLOW TI ME, t, HI

cu ft. (Note: q = 1,599 is used in Eq. 2 for both Fig. 4-Pressure data taken every 12 hours for random
rate and time variations. Rate varies as in Fig. 3, ± 15
percent. Time period is 3 hours ± 15 percent. = 1,599 q
STB/D. Conditions of Table 1 apply.

q' 800 BID,1 IN SEQUENCE


m'\:-0.020 PSI/HI
1700 1700

0 ~ m*:~0.0155
1200 2
..... nO
0 0
"0
0 0
'i 1200 8 J
.
'"
w" 1600
1600 J
w.. l600
! *

...
~
/ " " m '-0.0169
5 5 u
0 0 nn ,.,0
n 0
f 1600 7 OU 0 0'0
... 0 r!"

o=f 1500 ~ooo 4 ~ 1500 o 1000 -1100 BID


o 1600-1700 BID

~ o 2200-2300 BID ...


~ 2000 6 ~r- m :-0.00435
g §
vo
0 00> 0
'v v
~
2400 5 <f 0
1400
14~00 1600 1700 1800 1900 2000 2100 2200 1500 1600 1700 1800 1900 2000 2100 2200
FLOW TIME. t, HI FLOW TIME, t, HI

Fig. 3-Simulated pressure drawdown for an eight-rate Fig. 5-Simulated pressure data for completely random·
cyclic history. Rates change every 3 hours in num- rate, random-time drawdown. 800 ~ q ~ 2,400
bered sequence shown. = 1,600 STB/D.q with q = 1,610 STB/D; 1 D.t 5 hours. < <
Conditions of Table 1 apply. Conditions of Table1 apply.

DECEMBER, 1972 1425


Ieast-mpan-square straight line for the points in each in Fig. 6. Formation volume factor, B, is 1.0; system
group. Calculated pore volume is 4.9 X 10 9 ,4.5 X 10 9 , total compressibility, Ct, is unknown, but is estimated
and 17 X 10" for the circles, squares, and diamonds, to be about 5 X 10- 0 psi-".
respectively. The first two values are within one-third The lines in Fig. 6 are least-squares fits to the
of the correct value (3.76 X 10" cu ft), whereas the pressure data for the three rates. The slopes are
calculated result for the diamonds is almost four m* = + 2.62 X 10' psi/hr, + 2.15 X 10- 3 psi/hr,
times too high. This might be expected from the and +3.16 X 10- 3 psi/hr for the -5,140, -10,280,
data. Note that in Fig. 5 the diamond data points and -15,420 STB/D rates, respectively. Using Eq. 2
show considerably more scatter than the circular or we calculate the pore volume, being certain to use
square data points. They also show two long periods the average injection rate, q = 9,660 STB/D.
during which the pressure appears to be increasing
with time rather than decreasing. These two features - 0.23395q-B
Ar/>h=--~----"--­ (2)
provide a warning to the engineer that these data m*Ct
should not be considered seriously. Normally, one
- 0.23395 (- 9,660) (l)
should choose another rate interval, plot more points,
and calculate volume again under those circum- (2.62 X 10- 3 ) (5 X 10- 6 )
stances. Figs. 4 and 5 illustrate the general accuracy 1.73 X 1011 cu ft
of this data analysis technique. By using sound engi-
= 30.7 X 10" STB
neering judgment, using as much data as possible,
and eliminating obviously questionable data, it should for the circular data points. Similarly, the calculated
be possible to estimate reservoir size within a factor pore volume is 37.4 X 10" STB for the square data
of two, if the pseudosteady-state condition is satisfied. points and 25.6 X 10" STB for the diamond-shaped
One of the eight simulation tests was for a 4-to-l data points. These volumes do not contradict the
rectangle with the well in the center. The analysis known geology of the formation.
worked just as well for this system as for that in The relatively good agreement of the three calcu-
which the well was in the center of the square; it lated pore volumes indicates we can have some faith
merely takes longer for the system to reach pseudo- in the results. The results for the diamond-shaped
steady-state conditions. We see no reason why reser- points (q = - 15,420 STB/D) should be considered
voir geometry should create any difficulties in reser- the least reliable since there are only five data points,
voir limit testing when flow rate varies cyclically. and they are scattered. The value of 37.4 X 10" STB
may be the most reliable, as most data points are for
Example Calculation q = - 10,280 STB/D.
Fig. 6 shows pressure data for the last 5 years of These calculations require that flow be pseudo-
the II-year life of an industrial waste-disposal well. steady state for the validity of Eq. 2. The long period
The well, with casing set at 1,800 ft, is completed of apparently constant pressure increase with time is
open hole to about 3,000 ft, just into basement. Only a good indicator of pseudosteady-state conditions. If
a part of the 1,200-ft section is porous and permea- this assumption is incorrect, calculated pore volume
ble, but porosity and net interval are not known. The is too small.
injection horizon is known to be very large.
Injection is with one, two, or three pumps, so the
Recommended Testing and
rate is -5,140, -10,280, or -15,420STB/D. Pres- Analysis Procedure
sure and rate data are reported by month only, but For best results in applying the preceding reservoir
no indication is given of how long the rate had ap- limit testing technique, we suggest the following:
plied when the pressure was measured. Cumulative 1. Measure both pressure and rate as a function
fluid injected is known accurately. The average in- of time. Recording pressure and rate instruments are
jection rate is -- 9,660 STB/D for the period shown preferred.

°
2. Pick pressure-time points in a specific rate in-
terval. A rate interval of 5 to 1 percent of the total
rate variation should provide good results. Plot these
o - - - - - - - 5,140 BID data as in Figs. 5 and 6 for several rate ranges.
...'" o -10,280B/D (Ideally, the plotted pressure data should all be at
15,420 BID
<) - - - - - -
<)
the same time interval after a rate change.)
~ 400 3. Use the least-mean-square method to calculate
w
OK
~~~-->~~
o the best straight-line slope for each set of pressure
~
<)
o
oo 0 __0-
data. Results will be better than if "eyeball" straight
...
OK

~300
.
~
-- -- -- -- -- -- _~_--o-: lines are drawn.
4. Use Eq. 2 to calculate reservoir pore volume
~
.5l-0'- ---- o for each set of data points. The over-all average rate
must be used for each set of data; this is not the rate
-0
for that data set.
INJECTION TIME
5. Apply engineering judgment to decide what
Fig. 6--Data for example calculation; 5 years of injection
reservoir pore volume to use. Be wary of pressure
history for a disposal well with three injection rates. data that show many slope reversals, such as the
q = - 9,660 STB/D during period shown. diamond-shaped points in Fig. 5. These data should
1426 JOURNAL OF PETROLEUM TECHNOLOGY
probably be ignored. Once questionable data are (tDA)PSS = dimensionless time based on area, at the
omitted, choose the most frequently occurring result beginning of pseudosteady - state flow
as the best estimate of reservoir size. There appears t::..t = time period, hours
to be little value in averaging the results from several fJ. = viscosity, cp
rate-interval data sets. cp = porosity, fraction
6. If the system has not been operated long enough
for pseudosteady-state conditions to apply (Eq. A-5), Subscripts
calculated results will be too low. In this case, one D = dimensionless
has a conservative estimate. For large systems it may DA = dimensionless based on area
be months or years before pseudosteady state is i = index on time
reached and the technique can be applied. j = index on time
There is a possibility that a complete regression pss = beginning of pseudosteady state
analysis could be used to estimate shape, well loca-
tion, and size. 7 Acknowledgment
R. C. Earlougher, Sr., posed the problem that pro-
Conclusions vided the idea for the analysis technique given here.
Reservoir limit testing is feasible when flow rates vary
cyclically. Reservoir pore volume is calculated just References
as in the better known techniques, except that aver- 1. Jones, Park: "Reservoir Limit Tests," Oil and Gas I.
age flow rate is used. This approach has a sound (June 18, 1956) 184.
2. Jones, L. G.: "Reservoir Reserve Tests," I. Pet. Tech.
theoretical basis when only two rates occur in the (March, 1963) 333-337.
cycle. Analysis of simulated pressure behavior for 3. Matthews, C. S. and Russell, D. G.: Pressure Buildup
varying rate drawdown situations indicates that the and Flow Tests in Wells, Monograph Series, Society of
technique is applicable to much more complex situa- Petroleum Engineers, Dallas (1967) 1.
4. Earlougher, Robert C., Jr.: "Estimating Drainage Shapes
tions. If the system is large enough to be a good pro- from Reservoir Limit Tests," I. Pet. Tech. (Oct., 1971)
ducing or disposal reservoir, it may take several 1266-1268.
months, or years, to reach conditions for which this 5. Dietz, D. N.: "Determination of Average Reservoir
analysis can be applied. The larger the system, the Pressure from Buildup Surveys," I. Pet. Tech. (Aug.,
1965) 955-959.
longer the time until this technique can be used. 6. Earlougher, Robert C., Jr., Ramey, H. J., Jr., Miller,
F. G. and Mueller, T. D.: "Pressure Distributions in
Nomenclature Rectangular Reservoirs," J. Pet. Tech. (Feb., 1968)
199-208.
a = coefficient of time in generalized pseudo- 7. Earlougher, Robert c., Jr., and Kersch, Keith M.: "A
steady-state representation of dimen- Computer Method for Automatic Analysis of Transient
sionless pressure, Eq. A-7 Test Data," J. Pet. Tech. (Oct., 1972) 1271-1277.
A = area, sq ft 8. Ramey, H. J., Jr.: "Short-Time Well Test Data for
Oil Wells in Presence of Skin Effect and Wellbore
b = constant in pseudosteady-state represen- Storage," J. Pet. Tech. (Jan., 1970) 97-104.
tation of dimensionless pressure, Eq. 9. Agarwal, Ram G., Al-Hussainy, Rafi and Ramey, H.
A-8 J., Jr.: "An Investigation of Wellbore Storage and Skin
Effect in Unsteady Liquid Flow: I. Analytic Treat-
B = formation volume factor, RB/STB ment," Soc. Pet. Eng. I. (Sept., 1970) 279-290.
Ct = total compressibility, psi- 1 10. Ramey, H. J., Jr., and Cobb, William M.: "A General
C A = shape factor Pressure Buildup Theory for a Well in a Closed Drain-
age Area," I. Pet. Tech. (Dec., 1971) 1493-1505.
h = thickness, ft
J = index at which pseudosteady-state condi- APPENDIX
tions begin, Eq. A-9
k = permeability, md Reservoir Limit Testing Theory for
m* = slope of pseudosteady straight line used A Square-Wave Rate History
for calculating reservoir pore volume, A square-wave rate variation pattern is shown in Fig.
psi/hr 1. We can write the pressure behavior for that rate
n = upper limit on time index schedule using the principle of superposition3 :
P = pressure, psi n
PD = dimensionless pressure 141.2Bft ~ ( . _ . )
Pi = initial reservoir pressure, psi Pi - Pwi ()
t = kh i.J. q, q,.,
i=!
Pwf = flowing bottom-hole pressure, psi
q = flow rate at surface conditions, STB/D;
q > 0 for production, q < 0 for in-
• {PD ([t-U- 1)t::..t]D) +s}
jection (A-I)
qi = flow rate during time interval i
q = average flow rate, STB/D where
r w = wellbore radius, ft 0.0002637 kt
s = skin factor tD = (A-2)
f = time, hours
cpfJ. C t rw 2
tD = dimensionless time The dimensionless pressure, PD, may be calculated
fDA = dimensionless time based on area with well known approximations or determined from
DECEMBER, 1972 1427
tabulated or plotted data. 3, 6,~, 9 Note from Fig. 1 that
qi -qi-1 = q for i even (A-3a)
qi-qi-1 = -q for i odd (A-3b)
::::; Y2 at - 71'(tDA)P8,' (A-II)
Using these relations, Eq. A-I becomes
We use Eq. A-II and
_ 141.2 qB fL {
Pi - Pwr(t) - kh . PD(tD) +S -
q = Y2 (2q + q) = T3 q (A-12)
n/2

+1: [PD ( [t-(2j-l)Llt]D ) in Eq. A-lO to obtain


1=1
_ () _ 141.2qB fL{ 2 -I- 2 271'
Pi PwJt- kh at+Tb'TS-T(tDA)P,<S
-PD ([t-2 j LltJD )]} (A-4)
n/2
If n is odd, then the upper limit on the summation + ~ . 1: [PD ( [t-(2j-l)M]D )
is (n + 1) 12 and the last term in the summation F,J+ I
must be dropped when j = (n + 1) 12. For simplicity,
we assume n is even without affecting generality. -PD( [t-2 j Llt]D)]} . (A-13)
For any closed drainage system, a pseudosteady-
state Pn can be used when 4 ,5 The summation term in this equation requires spe-
cial attention. Note from Fig. 1 that
(A-5)
n ::::; tl Llt = tDI M D , . (A-I4)
For a well in the center of a square or circle, (t D '!)P8.,::::; where, of course, n must be an integer. Thus we see
0.1. Values for other shapes are available. a",6 Pseudo- from Eq. A-9 that for all practical purposes the
steady-state expressions for PD are of the form 4 ,10 difference
PD(tD) = at + b, (A-6)
n _ 2J = (tn)/l88
LltDr//)
A
2 • (A-I5)
where
(2,,) (0.0002637) k is constant since we have stipulated constant M. Keep-
a = ~--~----~~-- (A-7) ing this in mind, and using the approximation of Eq.
1>fL c t A A-14, we see that
h= Y2 In(2.2~58
r", C.!
A). (A-8) n/2

c,!is the shape factor. 3-:; From Eq. A-5 we see that
1:
Eq. A-6 can be used in Eq. A-4 if
(tn.1) p8s A
rw 2
This is equivalent to -PD ([(n-2j)M]D)] (A-16)

tn - (tn.J p 88 ( A2 ) Now we apply Eq. A-15 to Eq. A-16. For brevity,


.< rlO <J (A-9) we consider only the first term of the summation,
] - 2Lltn -
j = J + 1. The same reasoning applies to all other
where J is the smallest integer greater than or equal to terms. For j = J + I the first term is

tn -- (t IH )P88 (~)
rw
pn( [t-(2J+l)Llt]n ) -pn( [t-(2J+2)Llt]D)
2 LltD
= pn( [(n-2J-l)Llt]D) -Pn ([(n-2J-2)Llt]D)
When j ~ J we can substitute Eq. A-6 into Eq. A-4
and write
J
_ 141.2qB fL { ~
Pi - Pwr(t) - kh at + b + S + ~ aLlt
j=1

n/2

+ j-C~I [PD( [t-(2j-l)M]D )


Thus,

Original manuscript received in Society of Petroleum Engineers


(A-lO) office July 12, 1972. Revised manuscript received Sept. 5, 1972
Paper (SPE 3892) was presented at SPE 47th Annual Fall Meeting
held in San Antonia, Tex., Oct. 8·11, 1972. ® Copyright 1972
American Institute of Mining, Metallurgical, and Petroleum Engi·
Using Eqs. A-2, A-7, and A-9, we get neers. Inc.

1428 JOURNAL OF PETROLEUM' TECHNOLOGY


Using Eq. A-I8, we can arrange Eq. A-13 to read
PD ([t-(2J+1)tit]D ) -PD( [t-(2J+2)M]D )
2m* [ b+S-7T(t DA )pss+R ]
Pw((t) = m*t+Pi+ -3-
= PD ((tDA)pssA
2
-::..t[) ) _ PD ((tDA)pssA
., - 2 tit D)
~ ~-

(A-17)
= m* t + constant, (A-I9)
This term (for j = J + 1) depends only on the interval where
over which the cyclic rate acts, tit. It does not depend
on t, n, or J. Thus, the entire summation term of Eqs. - 0.23395qB
(A-20)
A-13 and A-16 must be a constant, R, as long as t is ¢ct hA
large enough to satisfy Eq. A-5: Eq. A-19 indicates that a plot of Pw( vs t should
nl2 be a straight line with slope m* given by Eq. A-20,
R = .1: {PD( [t-(2j-l)tit]D ) as long as t is large enough so Eqs. A-5, A-6, and
J='J,. 1 A-I5 are satisfied. Thus, the mechanics of reservoir
limit testing are the same for this situation as for the
- PD ( [t - 2j M]D)} = constant (A-I 8)
constant-rate case 3 ,4 with q in place of q.

Discussion (SPE 4262)


Hossein Kazemi, SPE-AIME, Marathon Oil Co.

The paper by R. C. Earlougher, Jr., has a sound to the boundary for the periodic flow-rate compo-
and practical approach. It must be emphasized that
the method of the paper applies to injection and pro- nent, A e iwt •
duction reservoir situations where: (1) the flow rate
oscillates periodically about the average flow nite,
When t > t[IRS' then PD(tD) = at + b,
where
(2) the period of oscillation is a small fraction of total
time t, and (3) t > tJiSS (tJiSS is the beginning of pseudo- (27T )(0.0002637) k
steady-state flow for constant flow rate). This can be a=
¢fJ.ct A
proved by superposing the pressure change due to
the average flow rate, q, on that due to periodic flow b= Y2 In (2.2':58A)
rw-C.!
rate variation about q. For instance, we apply such a
principle to a well in a rectangular reservoir: If the Thus, a plot of Pw( (t) vs t has a trend whose slope is
average flow rate is q and if the periodic part about m* = - 0.23395 qB (D-2)
_ ~ iwt _( A iwt ) ¢CthA
q is Ae , then the flow rate q is q = q 1 + e q •
From Eq. D-2 we obtain
Now, based on the work of Carslaw and Jaeger' (their
Eq. 13, Page 263) we write A h = - 0.23395 qB (D-3)
¢ m*Ct

+ A_ 't" e iwt If the pressure trend slope, m*, is known we can


t = 141.2qBfJ.[
Pi - p",r () kh PD (tD) i.J calculate the contributory pore volume A ¢h.
q
On the basis of the preceding analysis, we believe
J iWW/£t
• Ko ( ,,0.0002637 k rw
2) that if the flow rate is not oscillatory about the aver-
age flow rate with nearly constant periods, the pro-
cedure does not apply. In this case the results can be
only qualitative - as Earlougher has pointed out in
(D-I) ~~~~ ,

The illustrative problem that is presented here can


PD(tD) is the dimensionless pressure for constant flow be extended to practical pulsed (rectangular wave)
rate; s is the skin factor; and the terms involving e iwt flow-rate situations by applying the theory of images
constitute the periodic part of the solution. The sum- to the infinite case solution presented by Carslaw and
mation sign includes all the well images with respect Jaeger' (their Eq. 20, Page 402). Also, Culham's ap-
DECEMBER, 1972 1429
proach 2 may be used to obtain a pressure solution the average flow rate, STB/D
for a pulsed flow rate in a finite reservoir. However, i = V-I
the final analysis should yield the same conclusion Ko = zero-order modified Bessel function of
that we have arrived at in this discussion. the second kind
tpss = beginning of pseudosteady flow, hours
Nomenclature * w = frequency, radians/hour
a = coefficient of time in pseudosteady-state
representation of dimensionless pres- References
sure, hr~l 1. Cars law, H. S. and Jaeger, J. C.: Conduction of Heat in
Solids, Oxford at the Clarendon Press (1959).
A = amplitude of flow - rate variation about 2. Culham, W. E.: "Amplification of Pulse-Testing Theory,"
'All other symbols are as defined in the paper being discussed. J. Pet. Tech. (Oct., 1969) 1245-1247.

Author's Reply to Hossein Kazemi

Hossein Kazemi's contribution of a general solution


to the variable rate reservoir limit testing problem is t = St + 2n,,/w n = 0, 1,2, . . .. (R-l)
valuable. His application of integral transforms has where 8t is any arbitrary time increment chosen so
extended our results considerably and provided ad- that 8t < 2,,/ uJ.
ditional theoretical justifications for the analysis This indicates that a plot of pressures measured
method of our paper. He also correctly points out at times given by Eq. R-l vs time will be exactly
that different approaches can be used to treat the linear with slope m* when t > t pss • Thus, in Kazemi's
rectangular wave flow-rate case. general case, a plot of p vs t at a given rate will be
Note that the oscillating flow-rate, q, and the peri- exactly linear after t > t p ,,,' It is this fact that allows
odic part of Kazemi's Eq. 1 each assumes a fixed the reservoir limit testing technique to be applied.
value when JPT

1430 JOURNAL OF PETROLEUM TECHNOLOGY

Você também pode gostar