Professional Documents
Culture Documents
10
4.9 × 10 −4 ( kkrw A sin ) . . . . . . . . . . . . . . . . . . . . . . . . (4) Viscous flood analogs
qc = Log. (viscous flood analogs)
w ( M − 1) 0
1 10 100 1,000 10,000
(c) Viscosity, cP
Dietz showed that a displacement is “gravity stable” if G > M − 1;
while if G < M − 1, injected water underruns the oil. In Rajasthan, Fig. 1— Oil recovery vs. injection volume (a), well spacing (b),
average water injection rates into the Mangala FM3 member will and oil viscosity (c).
be 10,000 BWPD/well, which results in values of G ≈ 3 and M ≈ 15,
indicating gravity instability. The gravity-stable critical injection
rate to prevent a “Dietz tongue” from underrunning the oil column of scatter but a reasonable correlation coefficient of r2 = 0.55. The
is approximately 1,500 BWPD/well, too small to employ for an plot shows the expected increase in recovery as the amount of
“economic” operation. water injection increases, with an average recovery of 20, 30, and
40% of stock tank oil initially in place (STOIIP), respectively, at
Waterflood Results in Analog Viscous Oil Fields injected water volumes of 1, 2, and 6 HCPVI.
Table 2 contains some performance data and reservoir properties Fig. 1b essentially shows no trend when recovery is plotted
for more than 40 viscous oilfields undergoing waterflood, primarily against well spacing. This probably indicates that operators tend
from western Canada. The oil viscosities range from 5 to more than to infill their viscous oil fields down to a spacing that yields what
1,000 cp, with a median viscosity of 50 cp. These data have come they consider to be an appropriate “economic” level of recovery.
from analyses of publicly available oilfield histories, from the tech- Clearly, one will see an increase in recovery efficiency as well
nical literature, and from the Saskatchewan Research Council. Most spacing decreases in a given field, but when data regarding the
of the viscosity measurements are on dead oil (no gas) samples at current well spacing vs. current recovery are plotted for a number
reservoir temperature. However, some viscosity data from outside of fields, the result is the noncorrelation shown in Fig. 1b.
Canada may represent live oil measurements, in which case those Fig. 1c shows recovery plotted vs. the logarithm of oil viscos-
values should be increased by a factor of 30–50% to make them ity. The plot shows the expected decrease in recovery as viscosity
comparable to the Canadian data. increases, with average oil recovery of 40, 30, and 20% STOIIP,
Fig. 1a shows oil recovery plotted against the hydrocarbon pore respectively, for 20, 200, and 2,000 cp oils. The best-fit trend line
volumes of water injected (HCPVI) for the analog data set. The has a correlation coefficient of r2 = 0.33. The impact of oil viscosity
best-fit (logarithmic) line through the data points has a fair degree is obviously an important factor when it comes to oil recovery by
30
R2= 0.63
20
10
0
1.E−06 1.E−05 1.E−04 1.E−03 1.E−02
HCPVInj / (Viscosity × Spacing), cp-1m-1
Fig. 2 —Recovery vs. a function incorporating the volume of water injected, oil viscosity, and well spacing.
means of waterflooding, but its effect can be mitigated partially by and oil viscosity (cp). However, the correlation (r2 = 0.63) is only
maximizing water injection volumes and minimizing well spacing, marginally better than that from simply plotting water injection
both of which are under the control of the operator. The scatter in vs. oil recovery.
performance data in Fig. 1 is no surprise when one considers the Fig. 3 shows an “efficiency plot,” with recovery plotted vs.
interrelated and synergistic impacts of water injection, well spacing, HCPVI for a few analog Canadian waterfloods where oil viscosi-
and oil viscosity for each field. A high viscosity oil field developed ties range from 13 to 650 cp (median 50 cp). As expected, these
with close well spacing and large injection volumes could show viscous oil waterfloods do not perform identically because they
excellent recovery, but the same field developed by a different opera- have been operated differently and the effects of heterogeneities
tor with lower injection volumes and larger well spacing would show may be exacerbated for unstable floods. However, the plot shows
lower recovery. For example, the 54% of STOIIP recovery efficiency that very reasonable waterflood recoveries can be achieved if suf-
for the 60 cp oil at Gudong may not be unexpected owing to the very ficient volumes of water are injected and if production continues
close well spacing (50 m) used and the high injection throughput to high water cuts. In a perfect world, Fig. 3 would show smooth
(5.1 HCPVI) in this particular case. curves with ever-decreasing slopes; but in the real world, operators
Logically, one might expect that it should be possible to develop try to increase recovery as operations mature (i.e., surveillance,
a better performance correlation for viscous oil waterfloods; for infill drilling, or pattern modifications), causing some of the
example, one that relates better recovery to a combination of higher “irregularities” in the plot. There are similarities between Fig. 1a
water throughput, closer well spacing, and lower oil viscosity. Fig. and Fig. 3, even though Fig. 1a is a performance snapshot at one
2 shows a plot of oil recovery vs. the simple function of water point in time for many fields while Fig. 3 is a more detailed moving
throughput (HCPVI) divided by the product of well spacing (m) picture of performance vs. time for individual fields.
50%
120
13 40
80
50
24
40%
21
50
Oil Recovery, %OOIP
55
650
Fig. 3 —Recovery vs. water injected for a set of Canadian analog reservoirs (labeled with viscosity).
50
10
0
0
0
1960 65 70 75 80 85 90 95 00 05
Time, years 2008
Cal-Day Avg OIL Bbl/day Cum OIL 102.2MM/Bbl Percent: WTR Cut %
Cum WTR 620.0MM/Bbl Well Count count
50
10
0
0
0
90
3000
30
60
1000 2000
20
30
10
0
0
0
1973 77 81 85 89 93 97 01 05 2008
Time, years
Cal-Day Avg OIL Bbl/day Cum OIL 42,804,486.8 Bbl Percent: WTR Cut %
Cum WTR 474.0MM/Bbl Well Count count
90
30
2000
60
20
1000
30
10
0
0
0
0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000
Cumulative OIL, M/Bbl
Cal-Day Avg OIL Bbl/day Percent: WTR Cut %
Cum OIL 42,804,486.8 Bbl Well Count count
Cum WTR 474.0MM/Bbl
150 175
10000 12000 14000
60 70
100 125
40 50
8000
75
30
6000
4000
50
20
2000
25
10
0
0
0
1991 93 95 97 99 01 03 05 07
Time, years 2008
Cal-Day Avg OIL Bbl/day Percent: WTR Cut %
Cum OIL 23,393,456.6 Bbl Well Count count
Cum WTR 860.3MM/Bbl
75
30
6000
4000
50
20
2000
25
10
0
0
0
0 2500 5000 7500 10000 12500 15000 17500 20000 22500 25000
Cumulative OIL, M/Bbl
Cal-Day Avg OIL Bbl/day Percent: WTR Cut %
Cum OIL 23,393,456.6 Bbl Well Count count
Cum WTR 860.3MM/Bbl
60
30
2000
40
20
1000
20
10
0
0
0
1954 59 64 69 74 79 84 89 94 99 04 2006
Time, years
Cal-Day Avg OIL Bbl/day Cum OIL 101.5MM/Bbl Percent: WTR Cut %
Cum WTR 488.4MM/Bbl Well Count count
60
30
2000
40
20
1000
20
10
0
0
0
100
10
540
90
9
480
80
8
420
70
7
360
60
6
300
50
5
240
40
4
180
30
3
120
20
2
60
10
1
0
0
0
1987 89 91 93 95 97 99 01 03 05 07
2008
Time, years
Cal-Day Avg OIL Bbl/day Cum OIL 1,076,255.7 Bbl Percent: WTR Cut %
Cum WTR 12,675,063.6 Bbl Well Count count
100
10
540
90
9
480
80
8
420
70
7
360
60
6
50
300
5
40
4
240
180
30
3
20
120
2
10
60
01
0
0
0 125 250 375 500 625 750 875 1000 1125 1250
Cumulative OIL, MM/Bbl
Cal-Day Avg OIL Bbl/day Cum OIL 1,076,255.7 Bbl Percent: WTR Cut %
Cum WTR 12,672,063.6 Bbl Well Count count
100
50
4500
45
90
4000
40
80
3500
35
70
3000
30
60
2500
25
50
2000
20
40
1500
15
30
1000
10
20
500
10
5
0
0
0
1954 59 64 69 74 79 84 89 94 99 04
2008
Time, years
Cal-Day Avg OIL Bbl/day Cum OIL 30,890,360.2 Bbl Percent: WTR Cut %
Cum WTR 336.7MM/Bbl Well Count count
100
50
4500
90
45
4000
80
40
3500
70
35
3000
60
30
2500
50
25
40
20
2000
1500
30
15
1000
20
10
10
500
5
0
0
0
0 3500 7000 10500 14000 17500 21000 24500 28000 31500 35000
Cumulative OIL, MM/Bbl
Cal-Day Avg OIL Bbl/day Cum OIL 30,890,360.2 Bbl Percent: WTR Cut %
Cum WTR 336.7MM/Bbl Well Count count
100
50
45
90
40
80
35
70
30
60
25
50
7500
20
40
6000
15
30
4500
10
20
3000
10
5
1500
0
0
0
1952 58 64 70 76 82 88 94 00 06
Time, years 2008
Cal-Day Avg OIL Bbl/day Cum OIL 64,994,974.5 Bbl Percent: WTR Cut %
Cum WTR 579.2MM/Bbl Well Count count
40
20
4500
30
15
3000
20
10
1500
10
5
0
0
0
0 7500 15000 22500 30000 37500 45000 52500 60000 67500 75000
Cumulative OIL, MM/Bbl
Cal-Day Avg OIL Bbl/day Cum OIL 64,994,974.5 Bbl Percent: WTR Cut %
Cum WTR 579.2MM/Bbl Well Count count
30 30
50
40 R2 = 0.66
Discussion
The data in this paper can be used in a benchmarking sense, to
30
get a feel for what other operators have achieved in waterfloods of
their viscous oil fields. They can also be used to show what may
20 be “best practices” in these mature viscous oil waterfloods. Some
recoveries may appear quite high to those readers unaccustomed to
10 the waterflooding of viscous oils, but they can be useful in setting
Viscous flood analogs incl. Wilmington
Log. (viscous flood analogs incl. Wilmington) best-in-class performance targets.
0 From a reservoir engineering perspective, everything that
0 1 2 3 4 5 6 7 8 9 affects conventional waterfloods (i.e., heterogeneity, wettability,
(b) Hydrocarbon Pore Volumes Water Injected, HCPVI relative permeability, and appropriate voidage replacement) also
affects viscous oil waterfloods. As expected, waterflood recoveries
Fig. 25 — Oil recovery vs. water injected, including data points in viscous oil reservoirs increase with larger volumes of injected
from Wilmington. and produced water and with closer well spacing, both of which
are under our control. Viscous oil waterfloods may have a few
unique flow-assurance issues; however, basic reservoir engineering
8. Wilmington: The giant Wilmington field in California has principles still apply, except with a strong focus on a very different
produced more than two billion bbl of oil since the 1930s. It part of the fractional flow curve.
consists of deep marine turbidite sands in multiple stacked sand Although it may seem obvious, waterflooding viscous oil fields
sequences. It is a very good analog with similar thickness to the requires a definite change in operational mindset compared to
Rajasthan discoveries (much thicker than most of the examples conventional light oil waterfloods. For example, 50% or more of
discussed earlier). Waterflooding began in the 1950s, primarily the EUR may be recovered at very high water cuts, perhaps 90%
to mitigate subsidence occurring in the Long Beach harbor area. or even higher. Individual wells may operate to water cuts of 99%
A description of waterflood performance in the Ranger sand or even higher, depending on local economics.
(14–21°API oil with viscosity of 30–70 cp) is given in Wooding Facility designs and ongoing operations need to focus on produc-
et al. (1993). The reservoir was divided into segments for analysis, ing and reinjecting large volumes of water, as cheaply as possible.
with one of the key diagnostic plots being EUR vs. water injected. In fact, the most appropriate way of thinking about viscous oil
Fig. 25 compares results for the Ranger FO, FX, and X-HX1 sands waterfloods is to design and operate a large “oily-water” facility, with
with the analog data presented in this paper. The Wilmington the valuable “product” (oil) skimmed off the top of the water tanks,
waterflood data overlay the analog data nicely (Fig. 25a); if the as opposed to a classic oilfield operation that handles water as an
Wilmington data are included as part of this statistical analysis of unfortunate byproduct. This perspective is clear from Fig. 27, which
EUR vs. water injected (Fig. 25b), the correlation coefficient from shows another production plot from the Provost Glauconitic “A” field
Fig. 1 improves from to r2 = 0.55 to r2 = 0.66. discussed earlier in this paper (Example 3). Here, both total fluid and
oil volumes are plotted on the same scale, which gives a much better
Benchmarking the Rajasthan Discoveries idea of the “real” skimming operation that is ongoing. This mindset of
The Mangala, Aishwariya, and Bhagyam fields are the largest of handling huge volumes of water may be foreign to many companies at
more than 20 discoveries made since 2004 in Rajasthan (Rathore et the beginning of a viscous oil waterflood, and getting operational staff
al. 2006). These fields contain light, paraffinic crude oils with wax in particular to support the concept is a critical success factor.
appearance temperatures some 5°C less than reservoir temperature
and in situ viscosities of 8 to 250 cp. Development plans are based Summary and Conclusions
on waterflooding, with at least the initial volumes of injected water 1. Waterflooding viscous oil reservoirs can be an effective recovery
heated to 80–85°C to prevent problems with in situ wax deposition process, with typical EUR of 20–40% of STOIIP or even higher
(Wheaton 2007). if the appropriate operations are conducted. Simple waterflood
Given that there are no long-term production data to history operations for viscous oil reservoirs should be the base process
match and that the Rajasthan oils have some properties that may be for improved oil recovery, to which other more exotic recovery
considered potentially troublesome for waterflooding, the water- processes are compared.
flood benchmarks presented earlier were important reality checks 2. To maximize waterflood oil recovery from a viscous oil reser-
for expected ultimate recovery and optimal project design. Fig. 26 voir, it is important to inject large volumes of water and to handle
compares results from several Rajasthan waterflood simulations large volumes of produced water along with the oil. Normally,
with the analog data; the simulated Indian results follow the trend 50% or more of the ultimate oil recovery is produced at water
line very well. cuts of 90% or greater.
40 50 60
150
150
120
120
75
30
90
90
50
20
60
60
25
10
30
30
0
0
0
0
1991 93 95 97 99 01 03 05 07 2008
Time, years
Cal-Day Avg OIL MBbl/day Percent: WTR Cut %
Cal-Day Avg FLD MBbl/day Cum OIL 23,393,456.6 Bbl Well Count count
Cum WTR 860.3MM/Bbl
Fig. 27— Provost Glauconitic “A” production history showing oil and total fluid volumes.