Summary. Injecting water into a
low-pressure gas reservoir nearing
abandonment will displace gas and
increase ultimate recovery. This
paper examines the theory and
reports the results of a case history
Water was injected into a southern
Louisiana gas reservoir for more than
10 years, successtully increasing
recovery by 25 Bet {710% 106 m3}
This paper also discusses the im-
portance of considering all possi-
ble producing mechanisms early
ina reservoirs life to predict future
performance properly and shows
that straightline plz performance
does not necessarily indicate that
a gas reservoir has a depletion
drive, as is frequently assumed in
practice.
“Waterflooding a
volumetric gas
reservoir that does not
experience water influx
can increase recovery
from 5 to 16% of OGIP,
depending on initial
and abandonment
902 ———_______—
Waterflooding
Increases Gas Recovery
Lloyd D. Cason Jr., SPE, Exxon Co. U.S.A
Introduction
[Natural gas i a major source of energy pro-
duction and consumption in the U.S. As
reserves decline investigators are studying
the potential of increased supplies from
several unconventional sources! and from
‘waterdrive reservoirs with high production
rates in the gas zones? and the aquifers.
Production at high rates from a waterdrive
reservoir isan atempe to reduce reservoir
pressure to low value, thereby minim
ing the actual gas volume occupying residual
pore space, A reservoir with litle water in
Fux that depletes to low pressure has the op-
posite problem because all the pore space
1s sill filled with gas
‘This paper shows that increased recovery
can be obtained by waterooding a low=
pressure gas reservoir and discusses ana
tal Feld case, The Discorbis 1 reservoir
(Reservoir D-1) in the Duck Lake field of
southern Louisiana (St. Martin Parish) ini=
tilly produced by limited aquifer inlux.
‘Water injection was initiated ater reservoir
pressure had fallen below 1,000 psi [6.9
MPa and continued for 11 years. An in-
cremental recovery of 25 Bef [710% 108
m3] was ateibued to water injection, This
project was primarily a low-cost expansion
‘ofthe fel’s existing saltwater disposal sy
tem, providing a unique opportunity for an
‘economic project at the Ba prices of the
1910's, which were about 15% of current
Prices
Analysis of Reservoir D-1 also provides
the opportunity to show that straight plc
performance does not necessarily mean that
2 gas reservoir hs a depletion drive. The-
‘ory showing that depletion-drive eas reser-
voirs will exhibit straight-line pc plot has
been developed. The corollary that @
Straight-ie plz plot proves the existence of
a depletion drive-—has not heen proven,
hough itis frequently assumed in practice
Theory
1 a gas reservoir produces strictly by pres-
sure depletion, hydrocarbon pore space at
abandonment pressure should be equal to
that at intial pressure. significant amount
of gas, however, can remain in the reser-
voir. Injecting water at abandonment pres
sure displaces 2 fraction of the PV to
Cony 888 Sota of Peseum Engrs
production wells, leaving a residual gas sat
ration that conisins minimal standard
volume because of the low trapping pres
sure. Thus, the good features of pressure
depletion and water displacement are com-
bined in a controlled production mechanism
Craft and Hawkins? showed that recov-
ery for a waterdrive reservoir can be ex:
pressed as
100((1 SF —~Syped
O-SF
ao
and that recovery fora depletion rive reser-
voir can be expressed 2s
p= 100(F g~ Feo: ®
FFor imbibition fluid displacements, Naar
and Henderson® concluded that residual
nonvertng-phase saturation shouldbe about
‘one-half the initial nonsseting-phase satu-
ration, so that
or AS = =Sy) @
‘Substituting Eq, 3 ito Eg. 1 and subtract
ing Eq. 2 yields
AEp=50B (By ®
Eq, 4 presents a simple means to calculate
‘theoretical incremental recovery, a a per-
‘centage of original ga in place (OGIP), that
results from waterflooding a gas reservo
{hat is pressure depleting. This recovery fac-
tors epresenatve of & homogeneous teser-
voir where permeability is uniform areally
and venically, and water coning is ne-
elected
‘To investigate the magnitude of expected
theoretical incremental recovery, values are
calculated with Eq. 4 over a range of initial
and abandonment pressures for a 0,65-gravi-
ty gas. Initial pressure is varied from 3,000
10 6,000 psi [20.7 to 41.4 MPa}, and aban
ddonient pressure is varied from $00 10
1,000 psi [3.4 to 6.9 MPa}. The results
shown in Fig. 1 indicate that incremental
recovery can vary betwoen 5 and 16% of
the OGIP, Fig. 1 also shows that higher
reservoir pressures a the stat of injection
increase the potential incremental recovery
This i possible because a larger volume of
‘October 1960 +
pr226 is left to displace at these higher pres-
sures. Depending on relative costs, water-
flooding at a higher pressure when well
‘cepability i larger may be more economi-
cal than compressing a low abandonment
pressure
‘Values of incremental recovery caleulat-
ed by Eq, 4 could be used to sereen prospec
tive reservoirs. A more detailed analysis
should be made for specifi reservoirs, how
‘ever, because several variables affect rooov-
ery. The simplified assumption of Eq, 3 may
‘ot apply to all reservoirs. Geffen et al.”
reported that residual gas saturations can
range from 1S to 50%. Each 10% ditt
fence in the ratio S,,/S,, esuls in & 20%
change in incremental recovery. Laboratory
‘measurements to determine imbibition
residual saturations should be used where
possible. Also a less-than-ideal pattern oF
vertical sweep efliciency ina nonhomogene-
‘ous reservoir would cause average residual
saturation to be larger than that measured
inthe laboratory
‘An actual reservoir producing mechanism
will also affect incremental recovery. Pres-
sure depletion may be assumed from early
time production history when water influx
is actually occurring. Any influx reduces the
remaining hydrocarbon PV and incremen-
tal recovery attributable to water injection
“,, .the good features
of pressure depletion
and water displacement
are combined in a
controlled production
mechanism to
maximize recovery.”
Duck Lake Reservoir
D-1 Development
Discovery and Geology. The Duck Lake
field is 75 miles (121 km] west of New
(Orleans inthe Atchafalaya River basin. The
field was discovered in 1949 and contains
about 30 productive horizons ranging in
pt for 8,000 1 13,000 ft [2400 to 4000
‘] and in ag from Mile Miocene to Low-
er Miocene. Formations within the field are
Tow-relief, domical anticline structures
trending in a norteast/southwest direction.
Reservoir D-L isthe largest gas reservoir.
Fig. 2 illustrates th anticlnal nature ofthe
formation. The reservoir is hounded by a
fault on the northwest and a gas/water con-
tact at 12,574 ft [3833 ml] subsea on the re
maining Sides. Fig, 3 shows that water
initially did noe underlie the reservoir and
that influx was from a edge drive
‘Rock and Fluid Properties. Table 1 shows
basic rock and fluid properties for Reser
voir D-1. Rock properties ae based on logs
and sidewall core, Porosity averages 25%
and initial water saturation averages 18%
lover the 6,100-acre [24.7% 108-m?] area,
while permeability averages 1,750 tnd, Net
sand thickness ranges from 40 1090 f (12
027 mj and net pay thickness averages $0
(15 ml. The reservoir was not abnormal
|y pressured but was deep enough tobe con-
solidated and to have normal formation
compressibility over is producing life
Hydrocarbon properties were determined
from laboratory analysis of fluid samples
taken early in the life of the reservoir. At
initial conditions of $,800 psi [40 MPa] and
240°F [116°C], the low-yield gas-
Condensate fluid was at its dewpoint. The
Yield, however, slow enough that phase be-
havior changes should not affect pressure
performance data
Production History. Reservoir D-1 was
produced under unit operations with éril-
‘ng nits that were approved before the pro-
duction initiation in 1954. voluntary
reservoir-wide unit was later approved in
1968. Fig. 4 shows production history of
Reservoir D-1. Gas rate appears to have
been controlled mainly by market cond
tions. Sisteen wells were intially complet
ced in Reservoir D-, and production peaked
at 120 MMef/D [3.4%10® m3/d) in 1957,
‘The rate steadily declined to @ low of 62
MMef/D [1.8108 m3/d] in 1961, but in-
creased 10 130 MMefID [3.7108 m3/d]
in 1966 because of increased market demand
ad added completions. From 1967 to deple-
tion in 1981, the gas rate decreased continy-
ally asthe reservoir pressure decline reduced
capability and wells were lost to water pro-
duction. The first wells watered out in 1960
Fig. 1—Theoretical incremental recovery.
IPT + October 1989
———-—— 0sFig. 3—Cross section,
Fig, 4—Pressure and production history,
‘TABLE 1—RESERVOIR ROCK
‘AND FLUID DATA
afer 200 Bet [5.7% 10? m3] was produced.
‘This may have been the fis indication that
water influx was occurring. Performance
Productive area, aces 6.100
Porosity, 86 25
Pormeabity, me 1,750
Init water saturation, % 18
Init pressure, pst 5.300
“Temperature, *F 40
Original gas FVF, RBMsct 0.67
‘Wet gas gravity (air= 10) 0.85
Inia yield, DbNMect 14
‘Average timation cip, degrees 2.5
Aquterreserirraaivs rato 8
analysis could have indicated water influx
‘much earlier. Influx rose to a higher struc=
{ure evel oa the west than on the east, but
its effect on production was minimized by
structure and sand distribution. Figs. 2nd
3 show how water owed parallel 10 the
‘orthorn fault and across the central saddle.
‘The spill point was reached before 1966.
‘This allowed influx from the westside 10
reach the eas side ofthe reservoir without
‘sweeping across the structural erst, prema-
turely watering out wells
Reservoir pressure declined continually
‘over the producing life and reached 910 psi
[6.3 MPa} at the end of 1970, when water
injection was initiated. A cumulative of 130,
million bbl [20.7108 m3] was injeted by
‘Aug. 1981, an average of 3,000 B’D [5250
1m), Pressure decine leveled off during
water injection withthe low withdrawal
rates. Reservoir pressure reached a mini
‘mum of 585 psi [4 MPa] atthe end of 1979.
Performance Analysis. The typical per
formance analysis ofa pss reservoir inclodes
‘construction of plot of pz vs. cumulative
2s production. Ite poins plot as straight
Tine, the reservoir is sometimes assumed (0
be either volumetric oF producing by sim-
ple pressure depletion without water influx.
‘A-coneave, upward pz curve indicates an
active waterdrive, The interpretation that a
straight-line plz plot represents a volume
rie reservoir isnot stated but is frequently
assumed t0 bea corollary to proven theory.
ig. § shows the pz plot for Reservoir
D-l. A straight line can be fit through the
eatly points to about 45% of actual gas in
place (GIP) and extrapolated 10 show an
OGIP of 850 Bef [24.1109 m?). Later
data, however, plot concave downward
“Material balance, modified to account for
prossure gradiem across the invaded region
‘with unsteady-sate aquifer influx, as devel-
‘oped by Lutes er al.,? was coupled with the
regresion approach developed by Rossen®
to™match the pressure and production data.
"The match calculates an OGIP of 681 Bet
[19.3109 m3] and an aquiereservoir-
radius ratio of 6:1. The solid line in Fi,
‘shows the mateh between the model and
the actual data pots.
‘The possibilty of a concave-downward
plc plot fr a waterdrve was reported previ
‘ously, Bruns er al.? simulated production
fof a theoretical reservoie for 38 combina
tions of aquifer size and influx properties.
Several cases showed a concave-downward
performance with some remeining straight
{throughout the production life. Once the
aquifer size and influx constants become
large enough, the typically concave-upward
‘curve is exhibited. Dumoré® reports on a
fees reservoir in Germany for’ which a
cconcave-downward plz plot is predicted.
‘Run 20 of Bruns eal. and the prediction
‘of Dumoré are combined with performance
fof Reservoir D-I in Fig. 6. Early-time
Straight-line p/ ean persist uni 40 10 50%
Ne
Sa ea
uN
Fig. 6—plz performance comparison.
October 1989 + PT