Вы находитесь на странице: 1из 7

Jg69

Use of Injection-Falloff Tests


To Evaluate Storage Reservoirs
G. A. Mistrot, SPE-AIME, Southern Natural Gas Co.
J. R. Dempsey, SPE-AIME, INTERCOMP Resource Development and Engineering Inc.
Richard W. Snyder, SPE-AIME, INTERCOMP Resource Development and Engineering Inc.

ktm!wtim
Probably the single most important piece of reservoir filtrate invasion, it is not uncommon for some solids
data required in the conversion of a depleted gas field to drop out, completely blocking the sand face and
to gas storage service is effective rock permeability requiring the well to be acidized before it will even
distribution, and this information is usually not avail- accept gas on injection. Acidizing even a small
able in sufficient detail during the primary depletion job is at best a necessary evil as it adds more
.. . . , .? 3-. llJVWeu
:- ....J A z~fie. n-on ~t+
of the field. Wells are drilled on relatively wide spac- Ilqulas to an aJ~fSdUy MWF -.. cevf.n=
w,-

ing; and core or buildup data, even where available, damage can affect pressure transients for a substantial
provide insufficient points of control for adequate period of time.
mapping. Matching history with a reservoir simulator A testing method must be designed that can yield
can sometimes fill in the gaps, but it is not uncommon usable information even at low rates and in severely
for areal history matches to be somewhat insensitive damaged environments. The analytical technique
to relatively wide variations in over-all reservoir per- must &e able to provide the follow~g:
meability and individual well descriptive parameters 1. Extent of damage;
(such as damage, turbulence, and crossflow) at the 2. Permeability of the undamaged zone;
producing rates at which the field was depleted. 3. Effect or significance of geological heteroge-
As the field is being converted to storage mode, neity;
infill drilling provides the required additional points 4. Predicted injectivity capability, including the
of control. Normally, good porosity limits can be ob- effects on injectivity resulting from the removal
tained through logging programs, but permeability is of damage or from the eventual perforation of
another matter. a previously urmerforated productive interval;
Drilling into pressure-depleted reservoirs with gas- 5. Degree of reliability of test analysis and in-
or nitrogen-lightened mud is often ruled out from a terpretation.
cost standpoint, and deep invasion is commonplace If the field data are taken carefully, their trend
when nonaerated muds are used. Additionally, costs should be valid, and the results of a test should be
in general and costs and problems resulting from lost accurate within plus or minus a specific and deter-
circulation Wually preclude extensive coring. minable percentage.
Completed wells more often than not have signifi- Injection-falloff tests provide a relatively inexpen-
cant damage and insufficient pressure to flow out in- sive means of obtaining information from which ade-
vaded fluids and produce at rates and pressures that quate analyses can be made. One or more of the
will permit an acceptable evaluation. In addition to original wells can usually be used as a gas source,

Injection- fallofl tests are useful during the conversion of a depleted field into a gas
storage system. When production testing is not possible, injection- fallofl testing can
provide the essential data, and an analysis O) them can yield acceptable values of
permeability and degree of wellbore damage.

JOURNAL OF PETROLEUM TECHNOLOGY


and the gas can be compressed with either an existing Porosities calculated from logs, and permeabilities
rn.a~hin~ or a portable single-stage field compressor. read from the above-mentioned correlations, were used
The test procedure described in this paper consists to provide the initial reservoir description for history-
of one or more injection periods, at diilerent rates, matching the tests in an unsteady-state gas model
each followed by a falloff. This approach ditlers from with r-z geometry.4 History match is effected by hold-
those previously described in that four or more sep- ing the porosity constant and changing the permeabil-
arate unsteady-state pressure-time sequences are ana- ity by multipliers, essentially keeping, where possible,
lyzed for near-wellbore effects and formation perme- the same slope of the porosity/permeability relation-
ability. This approach of alternating injection and ship. In any event, the permeability assigned to an
shut-in maximizes the transient effects and makes an interval was kept within the range or band of values
adequate history match more diiiicult, but once ac- that core data indicated could be expected for the log-
,-nmml; .haA
buulpucmfitiu,
morn
x.. .-
AP
. . P+-nfm!e.
w..-.. The number of Liiffer- derived porosity for that interval.
ent rates at which a well is tested, and the length of The final reservoir descriptions showed permeabil-
each injection and falloff period will depend on the ity variations from iayer to iayer, which on the face
, ._ ---VOU
:----1
characteristics of the rescl dllu +h.-:A;.AAII1.1,s11
CllcLllu.v,Uwa, ..-,1. of it, .,ll--.+~ wdntions. Such
. .....yv-----------
.A~~w Ilninlw an implication
Often a two-rate test that can be completed in one is not intended. Once an acceptable match was ob-
working day is sufficient. tained, individual layer properties were varied over a
wide range, and the accepted description was the one
Test Procedures giving the best fit of the data.
In the field from which the examples were taken, test The analytical procedure is aimed at obtaining
gas was produced from one of the original wells and results that are accurate within acceptable engineer-
compresses .._.in d *nn
1 uu-hp
_ .:- I. .+. -.
SUIgIG-3
..,..+ .l.la
LdgG puI LaulG
_o.-h;
lL1ael..lLu
ma
ing tRi~~~IIC~S, ~~d ~h~ f~s~s were .rnodeied SO that
capable of discharging about 1,300 Mcf/D at approx- matched properties could be vaned to test the degree
imately 800 psig. of dependability of the results and the corresponding
After long string is set on a well, mud is displaced effect on predictions. If the best history match is
with water, a permanent packer is set about 50 ft questionable, or if it points to some anomalous con-
above the producing zone, tubing is run, and the tree ditions, a second test, based on the results of the first,
is set. The well is swabbed to the top of the packer, can be designed to investigate the problem.
logged and perforated, and injection is begun. If nec- For the times used, the transient is generally seen
essary, the well is acidized, and it is injected into at to 250 ft and occasionally as far as 500 ft from the
the maximum attainable rate for several days to move wellbore, roughly one-fourth to a maximum of one-
the invaded fluids as far away from the wellbore as half the well spacing. These particular tests, then,
possible. Injection pressures during this initial cleanup were not designed to necessarily yield average drain-
period decline substantially. At such time as injection age radius permeabilities, but rather, to yield essen-
pressure has fairly well leveled off generally after tially point data that are contoured to develop the
3 days to 1 week the well is shut in for 2 hours, permeability distribution for the field. The reservoir
then tested in the following sequence: rock is a clean, well consolidated, fine-to medium-
1. Three-hour injection at low rate, grain sandstone, and permeability variations from
2. Two-hour falloff, well to well are not extreme. The permeability map
3. Three-hour injection at high rate, was entered in a multidimensional reservoir simulator
4. Overnight falloff. to make fieldwide predictions and quantitative esti-
Tubing pressure is deadweighted at regular inter- mates of individual well performance.
vals throughout the test.
It is not uncommon for a well to continue to clean Field Example 1
up during the test, and mnning the test at a low rate The well penetrated 78 ft of productive section, leav-
first permits the cleanup effect to be isolated rather ing an estimated 20 ft between the bottom of the well
than mistaken for a change in turbulence, which could and the gas/water contact. Fig. 1 depicts the final
happen if the test were run at a high rate first. Con- matched reservoir description and completion inter-
versely, an apparent increase in damage during the vals. Other parameters are as follows:
test would be an indication of significant turbulence.
Total depth, ft 5,550
The characteristic of gas wells cleaning up with
Reservoir pressure in vicinity
throughput and the problems this causes in test inter-
of well, psia 628
pretation have been discussed in the literature.] -3
Reservoir temperature, R 590
Test rates selected were 600 to 700 Mcf/D for the
Gas gravity 0.66
low rate, and maximum attainable for the high rate.
Long string 7Y8 in. set in
At these rates, volume injected is negligible com-
978 -in. hole
pared with gas in place in the segment of the reservoir
Tubing ID, in. 3%
seen by the pressure transient, so the test is quite in-
Packer set, depth in ft 5,308
sensitive to gross section porosity, being affected by
permeability variations oniy. ~~e We]j was sw-a~D~~ do-w-n to ~fie p~~~~~, p~fi~-
Six of the original wells and two new storage wells rated, acidized with 3,000 gal of mud acid, and in-
were cored, and the analyses were used to develop jected into for 6 days at an average rate of 1,300
correlations for the range of permeability to be ex- Mcf/D, during which time injection pressure declined
pected for any given value of porosity. from 865 to 729 psia.
MAY, 1974 495
h K $ Pwf Remarks curves presented in this paper are plotted against
~ %1 1 pressure rather than against real gas pseudo pressure.
] 26 *,, ,048ii% I The test as matched did not consider turbulence.
The possible effects of turbulence were tested in three
10 8.1 ,048 ways.
e
42 12,2 .056 < 1. The test was history-matched assuming turbu-
. lence was present, using /3 factors from Katz et al. in
.
. the manner described by Coats et al. 4 A match only
0 TD. of well
slightly inferior to the one shown was obtained using
20 12!2 .056 a 6-percent reduction in skin effect and the same per-
1 meability in the undamaged zone as discussed above,
Fig. lWell 1 resewoir description. 2. The test with the damage removed was simu-
lated presuming (a) no turbulence, and (b) turbu-
lence. Turbulence was found to have a maximum
The well was shut in for 2 hours and a two-rate effect of 4 psi, reached after 30 minutes on the higher
test was mn as follows: injection rate. On the final falloff, pressure increase
1. Three-hour injection at approximately 650 due to turbulence had completely disappeared after
Mcf/D, 7 minutes. The short dashed line on Fig. 4 is the
2. Two-hour falloff, calculated final falloff with turbulence and without
3. Three-hour injection at rate varying from 1,385 damage.
to 1,297 Mcf/D, 3. An isochronal test at injection rates ranging
4. Fourteen-hour failoil. frcxm 1,30(! to ! 4,00!3 -McVI) was simulated with and
Fig. 2 is a printout of the history match of the test. without turbulence. The results are shown on Fig. 5,
This run was based on the description shown on Fig. indicating that the effect of turbulence (assuming the
1, and used a damage equivalent to a skin of 9.2 on correctness of the correlations of Katz et al.) becomes
the initial injection and falloff, cleaning up slightly to significant at rates in excess of 8,000 Mcf/D.
a skin effect of 8.9 on the final injection and falloff. The degree of reliability of the history match was
Figs. 3 and 4 are Horner plots of the final injection checked by changing permeability in the undamaged
and falloff with damage (as matched) and with dam- regions and rematching. Adequate matches, but not
age removed. The effects of afterflow persist for 1A so good as presented here, could be obtained with
hour, and the straight line portion of the test appears permeability variations of &12 percent and appro-
after 1% hours. Over the pressure ranges seen on priate changes in damage.
these tests, nonlinear gas effects are small, so the The matched reservoir description shown on Fig. 1

X CALCULATEDVALUES
800: O OBSERVED VALUES
: ~ COINCIDENTVALUES
;
:
:
;
::
:.
; .-
:-
:
:
. .
:
.: :
n. :
:
: .
w-
a :
a :
m :
u-) ~

:
:
w
:
o 0 0000 00000 Om 000 0 00000 It-1 Olmomooooo
. o Nrooo -Curooo -- moo twrmooo o - NNIO 0000
.. .. .. . . .. . . . . . . .. . . .. . . .. . . .. . . . ..-. . .. . . .. . . . . . . . . . . .- . . . . . .
:
: 0 0 Oo-m WCNalrnln mm u-jwr- r- *r-mmo o oooooNln-
-- . -
500: . . ....: ~.-.~
TIME
Fig. 2Well 1 history match.

496 JOURNAL OF PETROLEUM TECHNOLOGY


is a result of distributing the total permeability feet
>/- ---
em . in accordance with porosity/permeability correlations
--- previously mentioned. The effect or significance of
MO -
/
stratification was tested in the following manner.
1. The permeability in the undrilled 20 ft was
varied, and it was found that changes of *25 percent
:
780
did not rnateria!ly affect the histoxy match.
7W

I:
; ,: 2. The 802 md-ft in the section drilled were dis-
tributed in various manners. Neither history match
nor predicted performance was materially altered by
combinations ranging from equal permeabilities (10
md) in all three layers to one in which permeability
in the higher-porosity section was three times that in
the lower-porosity section (15 md vs 5 red). In both
cases, slight changes in skin effect were required to
Fig. 3-Well 1 final injection.
adequately match history. Variation in predicted
injectivity was less than 2 percent.
The simulated isochronal tests at rates from 1,300
to 14,000 Mcf/D were used to analyze the effect of
770
damage on injectivity and the effect of additional
\, perforating. Fig. 6 shows the results of the simulated
7W -
\, tests using 31A-hour injection periods, 25-hour falloff,
\\ and including turbulence.
Some poirits are in order here. A 3%-hour injec-
tion period moves the transient a distance of 500 ft,
representing nearly half the development spacing, or
the average radius of drainage. For minor variations
in permeability within this area, the simulated test
fairly well represents the wells injectivity. The quali-
tative value of the simulated test i.e., the order oi
magnitude of effect on the wells performance exerted
by damage removal and additional perforating
would be valid up to a fairly large permeability varia-
tion beyond 500 ft. It was also found that after a
3 Vi-hour injection period, a 25-hour shut-in was
Fig. 4-Well 1 final falloff.
necessary to completely restore the well to stabilized

Curve A: with damage and turbulence


Wtih turbulence, no damage
Curve B: with turbulence, no damaga
Ma turbulence or damage Curve C: with turbulence, no damage
dcifkxd w ps;fc?a:d
/ Bc

/ //

/ / -

1000 /

/ ,7/

/
/
/ -

/ .//
Ioc!
/
/

10 I I
1000 Iopoo to 1 I
1000 10,000
-Q, MCFID - Q,MCF)D
Fig. 5Well 1 simulated isochronal test. Fig. 6-Well 1 simulated isochronal test.

MAY, 1974 497


conditions. Even after 12-hour shut-in periods, pres- shut in for 2 hours, and another two-rate injection-
sure transients were great enough to affect isochronal falloff test was run in the following sequence:
test results. 1. Three-hour injection at approximately 680
Curve A of Fig. 6 shows the predicted response of Mcf/D,
the well as is. Curve B is the predicted response with 2. Two-hour falloff,
the damage removed, and indicates injectivity im- 3. Three-hour injection at approximately 1,325
provement on the order of 60 percent or more, The Mcf/D,
~fi--
,1 m-1 ..
, w ASwu..j tn;okt-l;n~ +m r- nf
.-.&.. ----- --- -- GM-W A. kdkates 4. Fourteen-hour falloff.
that damage effects are far more pronounced than The results of this test indicated that deep damage
turbulence, even at high rates. Curve C is the pre- had been removed, and only a residual skin remained.
dicted performance of the well with damage removed Apparently, significant amounts of mud solids drop-
and with the addhional 10-ft interval perforated. Pre- ped out close to the wellbore and materially affected
dicted increase in injectivity from additional perforat- the pressure response on the first test. These solids
ing is on the order of 11 to 13 percent. appear to have been removed by the acid, abetted by
Th. ffillmwr;no
~ AJQ ~-~u .. A ~
nnnclllcis-sn~
.... * . . . . ..v..
c~~ & &~WrS frOlll th13 the rather long [12-day) cleanup injection period.
test: Fig. 8 is a printout of the r-z model history match
1. Wellbore damage is equivalent to a skin effect of the test. The match was obtained using the de-
of 9.2 to 8.9, scription shown on Fig. 7 and a skin effect set initially
2. The capacity of the undamaged reservoir in the at 3.5 and gradually reduced to 1.7, indicating a con-
vicinity of the well is 1,042 md ft, *12 percent. tinuing cleaning up. Possible cleanup on the final
3. Turbulence is negligible at the tested rates, but falloff was checked by setting the skin to O after the
--,,1A ~.a,;.-mifipamtat
QUu,uw..&,..lw..... -. -----
rzstec
--
iq ~XC~SS of ~jOOO Mcf/D. second time step (5 minutes after shut-in); no change
This must be considered in calculated compressor was observed between this description and the one
discharge pressures and horsepower requirements. with a skin of 1.7 through the end of the test. The
4. The effect of stratification is minor. final skin effect, then, is 1.7 or less, and the impact
5. Injectivity capability can be substantially im- of skin on the definition of the falloff is established
proved by removing the damage. Perforating the addi- in the first 5 minutes.
tional 10 ft interval should improve injectivity by
11 to 13 percent.
6. The bag effect shown on Fig. 4 is predomi- h K 0 Pet+. Remarks

nantly the result of damage, secondarily the effect of 8 .0054 .025


afterflow. 14 ,126 0039 :

Field Example 2 32 ,054 .039


The well penetrated 136 ft of net section, ieaving an shale
estimated additional 8 ft between the bottom of the
hole and the gas/water contact. Completion intervals 22 6s52 ,075
1 I
and the best match of reservoir properties are shown t I
on Fig. 7. This description was arrived at in the man- 20 .027 ,052
shale
ner described earlier. Other parameters are as follows:
rgum
Total depth, ft
Reservoir pressure in vicinity
5,524 40 7,13 ,076 II2
;

TO. of Well
;0
of well, psia 675
8 7.13 .076
Reservoir temperature, R 590
Gas specific gravity 0,66 Fig. 7Well 2 reservoir description.
Long string 7% in. OD set
in 97A-in. hole X WCWA7EDWl~

Tubing ID, in. 31~ o C93ESW0 L!dLIEs


icOXOENTW

Packer set, depth in ft 5,268


After long string was set, the well was swabbed to
the top of the packer, perforated, and injection-tested. :.
The results of the test indicated a zone of very low .:
permeability less than 1 md in all layers close :.. .
to the wellbore. The history match was unaffected by
large changes in permeability beyond a 15-ft radial ..
distance, indicating that during the test period pres- .. :.
sure transients were not seen beyond the damaged
zone.
The well was injected into for 5 days, acidized
with 3,250 gal of mud acid, which was displaced with
gas, and injected into at an average of 1,264 Mcf/D
for another 7 days, during which time injection pres-
sure declined from 855 to 759 psia. The well was then Fig. 8-Well 2 History Match No. 1

498 JOURNAL OF PETROLEUM TECHNOLOGY


Figs. 9 and 10 are Horner plots of the final injec-
tion and final falloff plotting (1) BHP as matched,
840 -
and (2) calculated BHP response with skin removed
from the outset. The figures show that the straight-
820 -
line portions of the curves are well defined, and are
reached in some 45 minutes on the tinal injection and
2 hours or slightly less on the final falloff. The calcu-
Soo - lated response with no skin shows that afterflow effects
.
z
. persist for some 45 minutes on injection, and in
& excess of 1 hour on the falloff, and account for far
780 - more of the pressure change than does the skin effect.
The degree of accuracy of the match was checked
using increased and reduced permeabilities in the
760 -
undamaged zone. Fig. 11 is a printout of the best
_ 4S matched with
shin
match using a 10 percent increase in permeability.
* 4
The match is acceptable, but over all not so good as

I /// /
740 - skin ,*movGd
the match without the permeability increase. The
match is obtained using an initial skin of 5.0, declin-
1/ 1
,1
1
I
1
10
ing over the period of the test to 2.5. A similar match
At ,hr was obtained using a 10-percent reduction in permea-
Fig. 9-Well 2 final injection. bility, and smaller skins.
The effect of turbulence was checked using the ~
factors from Katz et al. A match ahnost identical
with the one shown on Fig. 8 was obtained, the only
difference in reservoir description being the use of a
final skin of 1.6 instead of 1.7.
It was found that relatively small changes in per-
\ meability to a radial distance of 240 ft had a notice-
[ \\ ohl.a aff(mt nn (-alcnbtm+ VZI1lM?Q ~D-~ ]~r~e Cbl@
8C.3 . aulw *U. v,. -... .- . ...- ,. .. . . .

l\
\ ,
\ beyond a radial distance of 490 ft had no effect on
\ calculated response. The test thus saw the reservoir
\
to a distance of 250 ft from the wellbore, and no more
g
/ .
than 490 ft.

.y,
.,
;
j
g ... \ In summary:
7,0
t
A~ .\ .,cMd w .,,.
1. The resuits of the first test indicated extensive
damage, and a need for remedial work and retesting.
c- . SM. r.md
2. The bulk of the damage was removed by the
:\
1
.a

mo \. combination of a small acid job and a total of 12 days


of injection.
3. The well continued to cleanup through the sec-
.~,
J, r 10
L ond test.
Fig. 10-Well 2 final falloff.
4. The effect of turbulence, at the tested rates, is
n-ti ;m;hla ,
uti~ 1.&W

5. The undamaged capacity in the vicinity of the


well is 492 md-ft, &10 percent.
Applications
Results of injection-falloff tests were incorporated in
over-all field studies and a field development program
ac fnllnwc.
La .s. ., .

1. Earlier studies predicted that 300 MMcf/D


could be initially injected at 1,400 psi compressor
discharge, requiring some 13,000 hp, or two ma-
chines. A later two-dimensional areal study, utilizing
permeability and damage data from injection-falloff
tests, predicted an initial discharge pressure of about
----- ~7j~(_)~ hpj or au three
.w-l.. ...~ .cnme
1,825 p-i, renllirinu
t machines. Installation schedules were adjusted to be
certain all machines were ready for initial injection.
The actual initial discharge pressure was 1,790 psi
at 300 MMcf/D.
2. The tests indicated (and performance bore out)
Fig. n-Well 2 History Match No. 2, IO-percent increase
that the majority of wells would clean up on their
in permeability. own. In those cases where treatments were deemed
MAY, 1974 499
necessary, test data were used in sizing jobs and scribed by Coats et al.4 is more flexible in that it per-
in estimating the required hydraulic horsepower. mits the engineer to (a) test the possible effects of
Treated wells on the average showed 5:1 initial in- radial and vertical heterogeneity; (b) easily check the
creases in injectivity, stabilizing at about 3.4:1. extent to which injectivity and deliverability are
3. An early field study indicated that 35 new wells affected by damage removal, additional perforating,
would be required to meet design deliverability cri- etc.; (c) simulate isochronal or other types of tests
teria. The later studv referred to in Point 1 (above) with a considerable saving of time and costs; and (d)
confirmed that requ~rements could be met with thi~ determine whether a well will probably clean up on
number of wells. Field performance through two its own or require treatment.
injection cycles and one withdrawal cycle has shown
no noticeable deviation from that predicted. References
1. Graham, J. R., and Boyd, W. E.: An Analysis of Chang-
Conclusions ing Backpressure Test Curves From Some Gulf Coast
Area Gas Wells, J. Pet. Tech. (Dec. 1967) 1541-1546.
1. Iniection-falloff tests can txovide individual well 2. Chen, T. K., Kennedy, J. H., and McElhiney, J. E.: Re-
descrip&e parameters and ~eservoir permeability covery Improvement Study of the Gray Sand Reservoir;
data in pressure-depleted fields undergoing conversion paper SPE 4596 presented at SPE-AIME 48th Annual
Fall Meeting, Las Vegas, Nev., Sept. 30-Ott. 3, 1973.
to gas storage in cases where such data might not
3. Jordan, W. V.: personal communication.
otherwise Ix-obtainable. 4. Coats, K. H., Dempsey, J. R., Ancell, K. L., 8nd Gibbs,
2. Tests can be interpreted analytically, but their D. E.: Analysis and Prediction of Gas Well Perform-
analysis in single-well models, such as the one de- ance paper SPE 3474 presented at SPE-AIME 46th
Annual Fall Meeting, New Orleans, Oct. 3-6, 1971.
Paper (SPE 3869) waa presented at the SPE.AIME Northern
5. Katz, D. L., Cornell, D., Kobayashi, R., Poettmann, F.
Plains Section Regional Meeting, held in Omaha, Nebr., May H., Vary, J. A., Elenbaas, J. R,, and Weinal]g, C. F.:
18-19, 1972. @ Copyright 1974 American Institute of Mining, Handbook of Natural Cm Engineering, McGraw-Hill
Metallurgical, and Petrolaum Engineara, Inc. Book Co., Inc., New York ( 1959). J_PT

500 JOURNAL OF PETROLEUM TECHNOLOGY

Вам также может понравиться