Surfactant Flooding

Carbonate Reservoirs
Wilton T. Adams, SPE, Texaco Inc.
Vernon H. Schievelbein, SPE, Texaco Inc.

Summary. Some of the most difficult yet most attractive EOR targets are found in the west Texas carbonate reservoirs, where
tertiary recovery is complicated by characteristically very low permeability coupled with high hardness and salinity of the reservoir
brine. This paper presents data obtained in two well-pair surfactant floods undertaken in the San Andres formation, Bob Slaughter
Block, Hockley County, Texas.
To provide the most data in the least time, a close well spacing and relatively large production/injection ratios were used. For
each well pair, a different surfactant system was developed, tailored to resist the local aquifer brine salinity and hardness.
Consequently, neither system required a preflush. Each test consisted of a large slug of surfactant, followed by biopolymer, fresh
water, and formation brine, together with appropriate nonadsorbent tracers. The large slug sizes were intended to allow subsequent
designs to be made by interpolation rather than extrapolation. The data provided by these tests demonstrate that (1) lowpermeability carbonate reservoirs can be chemically flooded, (2) polymer can be injected and propagated in these reservoirs
without any biodegradation, (3) these systems can mobilize tertiary oil effectively at a chemical consumption near that observed in
the laboratory, and (4) two-well tests can provide a quick field evaluation of surfactant systems proposed for large-scale use.
Introduction
The west Texas carbonate reservoirs contain vast amounts of residual hydrocarbons even though most of these reservoirs are in an advanced stage of depletion. The San Andres dolomite reservoir in
the Texaco Bob Slaughter Block lease is a good example of such
a reservoir. This lease was drilled in the late 1930's and initially
produced a large volume of oil by primary production. The lease
has been subjected to waterflooding since the 1960's, and despite
20 years of waterflooding, the lease still contains a large tertiary
oil potential.

Geologic Setting
The field lies on the northeast flank of the Permian Basin. The reservoir rock occurs in the Upper San Andres formation. The underlying Lower San Andres formation was deposited during the maximum
transgression of the Permian Sea and is dominantly a micritic limestone containing little porosity. The Upper San Andres formation
contains intertidal and supratidal sediments deposited during withdrawal of the sea across the broad, shallow shelf. A hot, acid climate combined with the broad, flat shelf resulted in carbonate
deposition, including algal mats, in what geologists call the sabkha
environment.
The carbonates were penecontemporaneously dolomitized by
downward-seeping marine storm waters or terrestrial floodwaters
alternating with times of moving capillary evaporative waters. Anhydrite beds developed, usually above or landward of the dolomites,
and also precipitated as nodules and infillings in the dolomite. As
regression of the sea continued, the anhydrite unit slowly covered
the dolomites, thus sealing them. Several distinct dolomite facies
were formed. These dolomite lithofacies reflect the original carbonate characteristics, such as grain size and origin of particles.
At the Slaughter lease, the San Andres formation is located about
5,000 ft [1500 m] below the surface. It is about 100 ft [30 m] thick,
and has a 109F [43C] temperature. The formation matrix is dolomite with varying amounts of anhydrite. The dolomite has intergranular porosity and is mostly fine-grained without large vugs.
The horizontal permeability is low, ranging from 1 to 25 md and
averaging 5.9 md. The average porosity is 12%, with a range of
8 to 18 %. The reservoir is distinctly stratified, and individual pay
zones correlate well over distances of several hundred feet, but the
correlations are not as good over distances greater than 1,000 ft
[300 m], which is typical of the distances between usual field-pattern
Copyright 1987 Society of Petroleum Engineers

SPE Reservoir Engineering, November 1987

well spacing. The barriers, where present, are anhydrite or poorly

developed, impermeable dolomitic material. The Slaughter lease
has consistently produced crude oil with a low GOR. The reservoir
oil has a l.29-cp [1.29-mPa s] viscosity, 31.4 API [0.87-g/cm 3 ]
gravity, and an FVF of l.14.
In the early 1970's, several commercial compensated neutron
porosity logs (CNL) were made in wells from a previous pilot area,
located about 1,800 ft [550 m] to the southwest of the two-well
tests reported here. They were adjusted to a common horizon, the
widespread anhydrite bed above the reservoir. The detailed porosity
correlation of these logs showed that there is good horizontal continuity of the reservoir in this area, while the close vertical spacing
of the porosity zones suggests poor vertical conformance.
A correlation of one of these logs (Well T -4) with the log of Well
E-4 located about 300 ft [90 m] to the south and east of the twowell tests is shown in Fig. I. Fig. 2 shows the correlation between
the CNL and sonic porosity logs. Fig. 3 shows a generalized core
lithology log superimposed on the porosity log (CNL) for Well E-4.
The porosity log reasonably approximates whole core analyses. For
Well T-4, the pattern of permeability values followed the trend of
the porosity data.
The higher porosity values generally occur in the sucrosic dolomites (Fig. 3). Dolomicrites, including fossiliferous dolomicrites,
generally have lower porosity and the algal dolomicrite exhibits
the lowest porosity. The latter could form an effective vertical barrier or seal for migrating fluids.
Water analysis (Table 1) shows that the original formation water
was extremely saline [220,000 ppm total dissolved solids (TDS)]
and contained high calcium and magnesium ion concentrations. The
formation is being flooded with a mixture of produced water and
Ogallala fresh water. Dilution with fresh water resulted in a salinity and hardness decline, as reflected by water analysis at the producing wells and the central production-handling facility where
commingled produced brine is processed before being reused in
the waterflood. Accordingly, much of the formation has been
flooded with brine of intermediate salinity and hardness, but surfactant flood solutions might contact higher-salinity and -hardness
brine.
The formation has been considered for chemical flooding, the
process with the best potential for significantly reducing residual
oil saturation. A surfactant flood pilot test was conducted in this
reservoir in 1974 with a first-generation salinity-tolerant surfactant formulation. The 5-acre [2-ha], one-quarter five-spot test dem619

POROSITY LOG CORRELATION

T-4
20

10

E-4
0

.,. CNL (I.)

4950'

20

10

E-4
0

POROSITY

.,. CNL (I.)

20

10

20

CNL (I.)

ZONE

10

SONIC (I.)

4950'
kA )4 MD

4950'
5000'

5000'
10

--'-0::;;;:;:-------_
5000'

6
8

10
5050'

5050'

'"

5050'

T.D. ,I
5100'

' - -_ _L -_ _ _ _ _ _ _ _ _ _ _--.J

5100'

Fig. 1-Porosity correlation between two Slaughter wells

separated by 90 m.

Fig. 2-Correlation of CNL and sonic porosity logs for Slaugh

ter We" E4.

onstrated that tertiary oil could be produced from carbonate

reservoirs by surfactant flooding. The pilot also made evident, however, the need for a more effective surfactant formulation and for
more effective procedures to deal with conformance problems associated with the heterogeneous San Andres formation.
Following this pilot, both surfactant-formulation technology and
the economic incentives improved, which encouraged further development work at Slaughter. Two surfactant formulations evolved
that recovered encouraging amounts of oil from Slaughter cores
in laboratory floods. Rather than a second pilot, a pair of two-well
surfactant flood tests were designed to define the best surfactant
formulation for Slaughter conditions and to obtain scale-up and design criteria necessary for later implementation on a larger scale.
Field activities for both these tests are now complete. This paper
gives an account of these field activities and the equipment used.
It presents the chemical analyses and mass balances on which performance results are based.

and 2 in. [5.0S cm] in diameter. These cores were prepared by assembling 2- to 3-in. [5- to S-cm] -long Slaughter core segments.
The brine permeability of clean cores ranged from 10 to 20 md
and the porosity from O.IS to 0.19%. The cores were initially saturated with produced brine and field crude oil thinned to reservoir
viscosity by addition of 25 vol % heptane. The cores were waterflooded with several PV's of produced brine to irreducible oil saturation, usually 0.30 to 0.32%.
The surfactant formulations were then injected into the cores with
no pretreatment, and the surfactant slug displaced with a viscous
polymer solution in Ogallala water. Polymer solution injection was
terminated when oil production ceased, usually before 1.6 PV of
tertiary recovery fluids were injected. During waterflooding and
chemical flooding, temperature was maintained at 109F [43C]
and the fluid displacement rate was 0.1 mLimin for a l-fi/D
[0.3-m/d] frontal advance rate.
Oil production history for a flood with 0.25 PV nonemulsion formulation slug is shown in Fig. 4. The surfactant slug was driven
by 1,000-ppm RhodopolTM polysaccharide polymer in Ogallala
water. In this flood, the oil saturation was reduced from a waterflood residual saturation of 0.31 to 0.07, for a tertiary oil recovery
of 75 %. Oil/water ratios as high as 0.5 were obtained. Oil production history for a O.IS PV emulsion formulation slug is shown in
Fig. 5. Again, this slug was driven with 1,000-ppm polysaccharide polymer in Ogallala water. Oil saturation was reduced from
0.35 to 0.10 for a tertiary oil recovery of 70% and a sustained 0.4
oil/water ratio.
Polymer drive solutions other than I ,OOO-ppm Rhodopol in Ogallala water were also tested. A polymer concentration of 500 to 1,000
ppm was found to be adequate, and tertiary oil recovery did not
increase when more concentrated polymer solutions were used.
More oil was recovered when the polymer solution was prepared
in Ogallala water rather than CBW or CBW /Ogallala water
mixtures.

Surfactant Formulations
Research and development at our Bellaire Research Laboratories
resulted in two new surfactant formulations that showed excellent
coreflood performance under conditions thai duplicate the Slaughter lease as closely as possible. The formulation referred to as the
"nonemulsion formulation" contains 1.5% (wt/vol) Solubilizer A
and 3.5% Witco petroleum sulfonates. The "emulsion formulation"
contains 1.46% Solubilizer B, 3.6% Witco petroleum sulfonates,
0.95% synthetic sulfonate, 4% gas oil, and 4% Slaughter crude.
Solubilizers A and B are alkyl ether sulfates and alkylaryl ether
sulfates, respectively. Slaughter central battery water (CBW) was
used to prepare the surfactant formulations, and both formulations
were designed to tolerate the high salinity and divalent ion environment of the Slaughter reservoir.
The laboratory floods' surfactant formulations were selected on
the basis of flood performance in linear cores 24 in. [60 cm]long
620

SPE Reservoir Engineering, November 1987

E-4

TABLE 1-WATER ANALYSIS

POROSITY I. (CNU

20

15

10

LITHOLOGY

SUCROSIC DOLOMITE
FOS. SILIFEROUS DOLOMICRITE
DOLOMICRITE
rALGAL DOLOMICRITE

4950'

U
U

Concentration (mg/L)
Original
Central
Ogallala
Produced Water Battery Water Water
TDS
220,000
106,000
811
7.3
pH
6.5
Calcium as Ca + +
15,000
8,600
70
Magnesium as Mg + +
4,200'
2,400
60
Sodium
63,806
28,000
80
Chloride
136,000
64,000
122
Sulfate
1,800
140
690
Bicarbonate
293
695
340

1.0

,..

0.5

0.6

z z

/'

"'0

~ ~

t!:~

5000'

,..

---

0.4

RECOVERY
EFFICIENCY

0.6

0.3

..

I-

:::>

"'I-

..J

0.2 (5

0.4

~ 0
u

'"

0.2

D::

0.1

OIL
_~~~URATION

o ......
o

............._ i _ _ _ _ ' 0

_"'-..IL..L-_--'--_....J.._~--'

5050'

5090'~-----=~--------~

0.5

1.0
1.5
VOLUME INJECTED, Vp

2.0

Fig. 4-Slaughter coreflood with 0.25 PV of the nonemulsion

system.

,....

1.0

0.5

0.8

0.4

Fig. 3-Slaughter Well E-4: core lithology on CNL porosity log.

z z

w Q

"'~ECOVERY

~ ~ 0.6

'"

...
... :::>

..
ffi

/'

0.3

,..

Two-Well Test Objectives

The concept of using a two-well configuration for field trials of
these two surfactant formulations was supported by several considerations. It was felt that some field trial was necessary to assess
the formulation potentials under actual reservoir conditions. Excellent oil recovery had been demonstrated in laboratory corefloods
that simulated field conditions as closely as possible. But a great
number of uncertainties still existed that could not be resolved with
corefloods. In addition, displacement of the complex surfactant formulations into a low-permeability dolomite reservoir was a daring
extension for surfactant flooding. Therefore, it was desirable to restrict the risk to as few wells and as little reservoir area as possible. To assess full potentials, however, large slug sizes were needed,
and this would be practical only with a compact test pattern. The
two-well configuration, with its close well spacing, reduced the time
and expense normally required; of the four wells required, two were
existing wells.
The primary objective of the two-well test was to define better
the oil recovery potential of chemical floods in a carbonate matrix.
Naturally, there were many subsidiary objectives associated with
a primary objective of this magnitude. Notable ones included (1)
evaluating the two-well test pattern for simplified field chemical
flood evaluation, (2) evaluating a biopolymer product that was new
to the industry and offered injectivity improvement, (3) evaluating
isopropyl alcohol (lPA) as a tracer, and (4) evaluating the concept
of preblending surfactant components to avoid problems associated with mixing components.
SPE Reservoir Engineering, November 1987

V)

>

8
w
Q:

l-

:::>

EFFICIENCY

W I-

..J

0.4

0.2

..J

0.2

1.0

-OIL SATURATION
O~--~~~--~--~~~--~--~--~ 0

0.5

1.0
1.5
VOLUME INJECTED, Vp

2.0

Fig. 5-Slaughter coreflood with 0.18 PV of the emulsion

system.

Test Site
The two-well tests were conducted in a portion of the Slaughter
lease that has been extensively waterflooded and researched. The
site for the two-well test was selected after review of waterflood
and pressure response data from both the 1974 pilot and 1980 tests.
The test site is 1,350 ft [415 m] from an existing mixing plant that
could easily be adapted for two-well tests. In 1981, Well Cl was
drilled and completed as an injector and existing Well E2 was completed as its producer. These two wells are perforated over the most
permeable interval of the San Andres formation. The bottomhole
spacing is 86 ft [26 m]. Injector G 1 was also drilled and completed
in 1981 at a bottomho1e spacing of 101 ft [31 m] from Injector El,
which was converted to a producer for this test.
621

400r------------------------------------,5000

START FRESH WATER

~ 300

START SURFACTANT
UJ

::,
0:

4000

START POLYMER

3000

'"

.~
UJ
0:

200

::>

U)
U)

2000 ~

>-

<.>

CL

"5
;;;

100
1000

JJASONDJFMAMJJASONDJFMAM
1981
1982
1983
TIME. months

Fig. 8-Well G1 injection history.

9.900 bbl [1570 m 3 ] for the first well pair (Wells CI and E2) and
about 12,000 bbl [1900 m 3 ] for the second (Wells GI and EI).
Because pressure was to be held constant during the field tests, it
was felt that these numbers would serve as initial estimates of PV.

100'
L----.J

Fig. 6-Potential flow model of the test site.

Field Operations
300 r------------------------------------, 5000
START SURFACTANT
START POLYMER

'"

.~

3000

::,

0:

::>

0:

U)
U)

2000 ~

>-

CL

"5
z
1000

J J A SON 0 J F M A M J J A SON 0 J F M A M
1983
1981
1982
TIME. months

Fig. 7-Well C1 injection history.

The bottomhole distances and a 2 : 1 production/injection ratio

were selected to minimize the swept volume, which minimizes the
amount of chemicals and time needed for the test. With well spacings this close, fractures in the target volume had to be avoided.
The new wells were drilled to avoid flow parallel to the reservoir
fracture trend. To avoid creating new fractures, injection pressure
was not to exceed 1,500 psi [10.3 Mpa].
The 30-ft [9-m] interval into which the injection and production
wells were completed is the most receptive interval within the 100
ft [30 m] of the San Andres reservoir. The permeability of this interval is approximately 25 md. As mentioned previously, anhydrite
and poorly developed dolomite barriers exist between productive
intervals and restrict fluid movement between intervals, but the barriers are fragile and may not be completely sealing. Fluid tracer
response data were used to determine whether these barriers were
in fact sealing.
Estimates of PV for these two patterns were obtained from an
in-house potential flow model that provides streamlines, arrival
times, and associated PV's assuming constant flow rates, uniform
pay thickness, unit mobility, and no fluid movement to or from
other intervals. This model (Fig. 6) indicated a target PV of about

622

First Well-Pair Test (Wells Cl and E2). Bob-Slaughter-Block

brine injection into Well CI began on April 24, 1981. On June 27,
addition of 340 ppm of thiocyanate tracer was begun to define the
swept volume before surfactant injection began. Approximately 600
Ibm [270 kg] in 5,000 bbl [800 m 3 ] were injected over a 20-day
period, after which brine injection continued until about 40,000 bbl
[6400 m 3 ] of brine had been injected. A preliminary analysis of
the tracer response history indicated that fluid sweep was fairly well
confined to the interval of interest and was confined to the pattern,
as predicted by the model (Fig. 6). After a final inspection of the
mixing plant facilities. the emulsion formulation injection commenced on Aug. 26.
As each surfactant batch was mixed, its viscosity and activity were
checked in a field laboratory and later confirmed at Bellaire. The
low viscosity of the premixed components made mixing easy, and
the mixing equipment functioned as expected. Several of the later
batches showed abnormally high viscosity; this was corrected by
adding small amounts of additional solubilizer.
This surfactant system remains a single phase over a wide temperature range. During the winter, however, it became clear that
cooling sustained during passage from the mixing plant to the wellhead was probably causing surfactant solution phase separation.
Quick laboratory tests showed that the phases recombined on warming more readily than they separated on cooling. Heat transfer calculations indicated that the surfactant was being warmed by its
passage downhole, so there is reason to believe that the formulation entered the formation as a single phase.
Nevertheless, there was a decline in injection rate with the constant I,SOO-psi flO.3-MPaj injection pressure (Fig. 7). This slow
decline persisted throughout surfactant injection. A hot-water flush
in Oct. 1981. followed by a short shut-in and a backflow. recovered no material that would explain this decrease. and on resumption of injection. unusually high injection rates persisted for several
hours. This indicated that there was no wellbore damage. This, coupled with some radial flow calculations, led to the conclusion that
the decline in injection rates was caused by an imbalance between
the higher viscosity (8 to 20 cp [8 to 20 mPa' s]) of the surfactant
solution and the increase in water relative permeability because of
oil mobilization. By Feb. 13. 1982. 12.846 bbl [2042 m 3 ] of surfactant had been injected, an average of about 75 BID [12 m 3 /dl
over the 171-day period of surfactant injection.
When fluids were analyzed for tracer, it became clear that the
iodide tracer used in the surfactant was being masked by one of
the surfactant components. The solubilizer contained a large amount
SPE Reservoir Engineering, November 1987

TABLE 2-FIRST WELL PAIR CHEMICAL RECOVERY

Amount Injected

Amount Recovered

Ibm

Presurfactant tracer (SCN - )

Surfactant
Surfactant tracer (ETOH)
Polymer (Flocon 4800)
Polymer tracer (IPA)
Posllest tracer (SCN - )

Ibm
- - - _k_9_ - - 610.6
277.1
218.0
274,849 124669 80,786
16,622
7539
7,283
2,399
1088
594.1
3,238
1469
860.7
117
53.0
22.1

of ethanol (which is substantially cheaper), so ethanol was designated the surfactant tracer and analyses for it were performed.
The biopolymer supplier and laboratory and field personnel were
on site during polymer injection startup to ensure that the injection
solution was meeting the quality control standards. No problems
were encountered with solutions prepared from Flocon 4800
broth. In March, when solution filterability became very poor, the
problem was traced to the source water, not the polymer; filter cartridges were replaced, and the problem did not recur.
Two incidents of bacterial contamination at the surface occurred,
both coincident with an interruption of biocide addition. An injection solution batch in March and part of a batch at the end of May
were discarded, and the injector was backflushed, then dosed with
biocide. No bacterial contamination was found in either injector
or producer wellbores. In late April, the only serious problem during
polymer injection occurred. The injection rate had declined to only
40 BID [6.4 m 3 /d]. At the wellhead, viscosity was found to be
depressed and iron was detected. No iron was present at the mixing plant and the problem was traced to the injection pump's bypass
line, where the coating on a coupling had failed, allowing it to rust
severely. The coupling was replaced and the injector backflushed;
several pounds of gelled polymer and high concentrations of polymer in the backflushed fluids were recovered. Injection resumed
at a somewhat improved flow rate.
Polymer injection concluded on July 16, having averaged 40 BID
[6.4 m 3 /d] over 146 days. Injection was switched to fresh water,
but injection rates did not increase as much as expected. Consequently, a workover was performed in mid-August. The wellbore
was treated with nonemulsifying, iron-sequestering (NEFE) hydrochloric acid and sodium hypochlorite. This improved injectivity
somewhat. On Dec. 5, injection of a postpolymer tracer slug of
thiocyanate was begun. A little later, in mid-December, the injector was again treated in greater depth with hypochlorite and acid.
This seemed to remove the last of the rust -gelled polymer because
the injection rate rose to its preflood levels.
Postpolymer tracer injection was concluded after 41 days. The
injection rate averaged 117 BID [18.6 m 3 /d]. Freshwater injec-

Second Well-Pair Test (Wells GI and EI). Brine injection into

Well G I began on April 21, 1981. Pretest thiocyanate tracer injection began that July, and continued for 24 days at an average rate
of 131 BID [20.8 m 3 /d]. A long period of water injection followed, while the first well-pair test field activities were completed. During this 350-day period, about 68,500 bbl [10 900 m 3 ] of
brine were injected at an average rate of about 200 BID [32 m 3 /d].
Injection before the start of surfactant totaled almost 85,000 bbl
[13 500 m 3 ]. Again, tracer response history indicated reasonable
fluid confinement to the interval of interest and horizontal movement consistent with the streamline model.
Injection of the nonemulsion surfactant system began on July 30,
1982. Because the first well pair had responded so well to surfactant injection, it was decided to try a substantially smaller slug size
for the second test. The slug was therefore limited to about 5,000
bbl [800 m 3 ] on technical grounds as well as reasons of economy
and timing. During surfactant injection, flow rate again dropped,
as expected. No injectivity problems were encountered. Surfactant
injection was concluded in late September, after injecting 5,058
bbl [804 m 3 ] over 61 days at an average rate of 83 ~ID [13 m 3 /d].
Polymer injection began on Oct. I, 1982, accompanied again by
isopropanol as a tracer. Injection proceeded smoothly over the
45-day injection period, during which 3,700 bbl [590 m 3 ] of polymer solution were injected at an average 71 BID [11.3 m 3 /d]. The
freshwater filter cartridges were replaced once during this period.
Freshwater injection continued after the end of the polymer slug
until Jan. 1983. During this period, the injection rate returned to
its former level. On Jan. 6, a 3,800-bbl [600-m 3 ] slug of
thiocyanate tracer was started. This took 19 days at an average rate
of 200 BID [32 m 3 /d] and was followed by fresh water, which
continued until November, when injection was switched to brine.
More than 50,000 bbl [8000 m 3 ] of water have been injected into
Well G-I since the end of polymer injection (Fig. 8).

100

2000

800

80

300

600

:::<7

1500

~ 200

1'"

1000

lt
0:

ti

g:

60

'"

400

UJ

;;;;
>...J

'"~

'"

40

<l.

:::>

UJ

Recovery
(%)
78.7
64.8
96.6
54.6
58.6
41.7

tion continued until Nov. 8, 1983, when injection was switched

to field brine. At this point, about 100,000 bbl [16000 m 3 ] of
water had been injected since the end of the polymer slug.

2500

400

k9
480.5
178 102
16056
1310
1898
48.7

100
500

J J A SON D J
1981

F MA M J J AS 0 N D J F M A M J

1982

1983

TIME, months

200
20

o
J J

0
A SON D J F M A M J J A SON D J F M A M J

1981

1982

1983

TIME, months

Fig. 9-Well E2 surfactant and ethanol response.

SPE Reservoir Engineering, November 1987

Fig. 10-Well E2 polymer and isopropanol response.

623

0.06,.----------------------,

0.04
t-

::0

U
...J

tW

0.02

SURFACTANT POLYMER

~~~-r~-r~_r~~~~~~~~._rT~

JJASONDJFMAMJJASONDJFMAMJJASO
1981

1982

1983

TIME, months

Fig. 12-Swept-area oil saturation at end of emulsion-system

test.

Fig. 11-Well E2 oil production.

TABLE 3-01L RECOVERY ALLOCATION

Total oil produced, STB [m 3)

Waterflood oil background,
STB [m3)
Recoverable oil in injected
surfactant, STB [m 3)
Incremental oil, STB [m 3 )
Residual oil saturation to
chemical flooding, %
Recovery efficiency, %

First Test
(Emulsion)
7,932 [1261)

Second Test
(Nonemulsion)
13,658 [2171)

4,942 [786)

11,936 [1898)

918 [146)
2,072 [329)

42 [6.7)
1,680 [267)

7.5
77

18
43

Results
First Well-Pair Test. The mass balance on tracer recovery reflects
a general but not total confinement to the interval of interest (Table 2). The pretracer recovery was good (78.7%), which demonstrates good reservoir interval isolation and good pattern
containment. The declining tracer recovery progression (with the
exception of the ethanol tracer) reflects some change in fluid displacement pattern. The change might be the result of filling the reservoir with increasingly more viscous fluids coupled with injection
rate changes. But even in the worst case, a very substantial portion
of the injected solution was produced. Tracer analysis procedures
were carefully reviewed to ensure that no masking or other effects
were present.
Ethanol appears to have functioned entirely satisfactorily as the
tracer for the emulsion surfactant system. The analysis indicates
that essentially all of the tracer was recovered, which is taken to
mean that all the injected surfactant entered the appropriate layer.
This volume amounted, as will be seen later, to about 1 \13 PV. Of
this, about 65% of the surfactant was recovered, while the remainder
was retained by various mechanisms, including adsorption and partitioning into the oil unrecovered from the swept volume (Fig. 9).
Retention of the sulfonates in the surfactant system amounted to
a little under 1.3 mg/g rock. From the point of view of field operations, preblending the surfactant components proved to be a big
advantage both in handling high-viscosity fluids and generally in
simplifying mixing procedures.
Polymer recovery was high, 55% of the mass injected (Fig. 10).
Recovered polymer samples did not show any evidence ofbiological, oxidative, or shear degradation. The liquid concentrate obtained
from the manufacturer was easy to process, and as noted above,
the filterability of injection solutions remained acceptable.
Isopropanol seems to be a good tracer for the polymer soiution.
Isopropanol recovery was accepted as the indicator of polymer
624

sweep efficiency. On this basis, polymer retention was 0.0014 mg/g

rock, a low value that confirmed polysaccharide's reputation as not
being substantially adsorbed.
Oil recovery from the swept volume was calculated on the basis
of production before and after the flood (Fig. 11). During the period preceding surfactant injection, oil cut at Producer E2 persisted
at an average just under 1.3 %. No decline in oil cut could be detected, as would be expected if the brine injection were- depleting
a swept volume. The response to surfactant at Producer E2 was
prompt; within 1 week the oil cut had risen and surfactant and tracer
were detectable in produced fluids. At its peak, the oil cuts averaged around 5 %. After giving due consideration to the overproduction ratio, oil cuts from the two-well pattern probably attained and
may have exceeded 40%.
Several adjustments to the produced oil-cut data were made to
isolate better the fraction that is strictly attributable to the surfactantflooding process, A constant 1.3% waterflood oil cut was assumed,
and the corresponding volume of oil was subtracted from the daily
production. In fact, a decline in background production should have
occurred, according to usual oil-depletion mechanisms. This decline
curve could not be defined; consequently the assumption was made
that the background oil cut was constant. The injected surfactant
formulation also contained crude oil and gas oil; therefore, injected volume of hydrocarbon was also subtracted from the gross pro~
duction. Accordingly, 7,932 bbl [1261 m 3 ] of oil were produced
from Well E2. Of this amount, 4,942 bbl [785.7 m 3 ] are attributed to background production, and 918 bbl [145.9 m 3 ] to the injection, leaving 2,072 bbl (329.4 m 3 ] to be credited to surfactant
flood recovery from the pattern (Table 3).
Calculation of recovery efficiency, sweep, and retention depend
on the estimate of PV used. Initial estimates of PV were made with
the potential flow model mentioned earlier. Later refinements were
made on computer simulations of the test results with our version
of the Intercomp CFTE chemical flooding simulator. This is a finitedifference, gridblock model that allows for phase behavior and
crossflow. With an acceptable match to the preflood thiocyanate
tracer and to the oil recovery information, those gridblocks showing a decline in oil saturation at the conclusion of field operations
were identified. The calculated PV was 9,651 bbl [1534 m 3 ]. This
number was considered the best estimate of the first well-pair test's
PV and was used to calculate values of chemical retention and recovery efficiency.
Because there was no decline in oil cut during the presurfactant
period, in which 3.20 PV's of brine were injected, it was assumed
that the two-well test pattern was completely watered out. On the
basis of laboratory coreflood and previous experience in this area
of the field, the oil saturation was assumed to be 32 %. This suggests a target reservoir oil volume of 3,088 bbl [491 m 3 ] (Fig. 11).
SPE Reservoir Engineering, November 1987

TABLE 4-SECOND WELL PAIR CHEMICAL RECOVERY

Amount Injected

Amount Recovered

Ibm

Presurfactant tracer (SCN - )

Surfactant
Surfactant tracer (I - )
Polymer (Flocon 4800)
Polymer tracer (IPA)
Posttest tracer (SCN - )

Ibm
- - - _k_9_ - - 1,221
89,342
220
1,603
1,548
105

554
40525
99.9
727
702
47.7

150.--------------------------------------,

_k_9_

Recovery
(%)

692
30644
182
372
896
39.7

56.7
34.3
82.9
23.2
57.9
37.7

314
13,900
82.8
168
407
18.0

r-------------------------------------~300

.,..

200

::: 100
POLYMER

z
~
u

'":::J
i2

....

f-"

.
~

100

50

'"

MAMJJASONDJ

MAMJJASONDJFMAMJJASON

FMAMJJASON

1982

TIME, months

TIME, months

Fig. 13-Well E1 surfactant and iodide production.

Fig. 14-Well E1 polymer and isopropanol production.

Accordingly, the recovery efficiency of 77 % and an average residual

oil saturation to chemical flooding of 7.5 % was obtained from the
two-well pattern (Fig. 12).
Second Well-Pair Test. As in the first well-pair test, tracer recoveries were good but not complete (Table 4). Surfactant was traced
with iodide ion, which was not masked by any of the nonemulsionsystem components. Produced iodide concentrations were corrected for a background level that dropped during the flood. The drop
was ascribed to production of fresh water from the polymer slug,
and a logarithmic decline was used to represent the iodide background levels between the start of polymer injection and the end
of freshwater injection. Because background concentration is critical to calculating correct tracer recoveries, and consequently surfactant adsorption and retention, a special effort was made to obtain
an accurate production value.
The combined surfactant and Tracer E-I production records indicate that about 41 % of the surfactant entering the swept volume
was recovered (Fig. 13). On this basis, surfactant retention was
0.44 mg/g rock. Surfactant, tracer, and oil response at Well E-I
were delayed and more gradual compared with the emulsion-system
test. This could be the result of larger PV in the nonemulsion test
pattern and poorer confinement to the interval.
The polymer production history reflects results that are very much
in keeping with the first test. Collected samples again showed that
the polymer was still in good condition and that the polymer tracer
recovery was practically the same-58% (Fig. 14). The mass
balances indicate a retained level of 0.0055 mg/g rock, which is
higher than the first test but still quite low. In view of the lower
injection rates and other events attending the first test's polymer
injection, the second test's polymer retention level is probably better for design purposes.
Oil recovery calculations were handled the same way as before.
Before surfactant injection, the oil cut averaged about 2.9% over
a period in which 6 PV's of brine were injected. This was the level
above which oil production (corrected for surfactant oil, water content, etc.) was credited to the nonemulsion surfactant system. As
noted, this response was more gradual than in the first test. By the
time oil cut returned to the pretest levels, this oil amounted to 1,680
SPE Reservoir Engineering, November 1987

1983

1982

1983

0.08 ,----------------------------------------,

0.06

....

:::J
U

g 0.04
....

'"z
0.02
SURFACTANT POLYMER

HH

O~~~~~~~-r~_r~_r~_r~_r~rr~

JJASONDJFMAMJJASONDJFMAMJJASO

1981

1982

1983

TI ME, months

Fig. 15-Well E1 oil production.

bbl [267 m 3 ] out of the total 13,658 bbl [2171 m 3 ] produced (Fig.
15). Again no allowance was made for depletion in production from
nonpattern areas. This could be particularly significant in this pattern because production was at a higher level and the tertiary production occurred for a longer period of time.
To estimate PV, the same procedure and assumptions as for the
first test were used. Because of the more gradual response to injected chemicals of this pattern, the fit to the data is not as good
at this time as for the first test. Consequently, the calculated PV
does not carry as much confidence. Again, however, it does agree
with all previous, less precise approximations and is believed to
reflect fairly the results of the test. For the second test, the PV was
a little under 14,000 bbl [2225 m 3 ]. Following this very conser625

vative line of reasoning, the residual oil saturation to chemical flooding is allocated to be 18 % and recovery efficiency 43 %.
Determining which of these two surfactant systems has the greater
potential in the San Andres formation will depend on more refined
values of the second well-pair PV, as discussed above, and of its
pattern oil recovery because this figure is a small difference between large numbers. Both systems clearly demonstrated that they
were propagated through the formation while mobilizing significant amounts of residual oil, and both systems are under further
development.

Conclusions
1. Both surfactant formulations recovered very significant
amounts of oil from the dolomite reservoir. Low permeability reservoir brine with high salinity and hardness and a carbonate rock
matrix can be tolerated with the tested formulations.
2. The concept of using well pairs for surfactant-system field tests
to gain performance and scale-up data was tested. The results support the concept: two well tests appear simpler, cheaper, and faster
than other field trial schemes discussed in the literature. In particular, less than total confinement does not preclude useful results
when a properly executed analytic evaluation can indicate confinement's true extent.
3. Tests with multiple-component surfactant formulations are
greatly facilitated and made more reliable if component concentrates are preblended at chemical blending plants before shipment
and final dilution in the field.
4. A commercial biosaccharide was an effective mobility control agent in this low-permeability carbonate matrix. When biocides
were used properly, biological problems were controlled. Injec-

626

tivity was not a problem, provided good-quality water was used

to prepare the solutions and iron corrosion products were not allowed to contaminate the solution. Polymer retention on the matrix
was low.
5. Isopropanol and ethanol are good tracers. They are both readily
available, inexpensive, easily handled, do not adsorb or partition
into oil, and are readily detected.

Acknowledgments
We thank Bellaire Laboratories chemists and engineers for their
support, particularly J.M. Shugart, T.D. Vu, L.A. Leu, B.J. Franklin, J. Maddox, J.L. Sides, and the analytic chemists. We also thank
the Texaco Midland Div. for providing the wells, reservoir equipment, and chemicals for the test, and the division and field personnel, especially K.A. Rabenaldt, G.D. Laird, M.R. Reinhold, and
W.B. McGee. We are especially indebted to H.M. Dahl and J .W.
Wood for their help and advice in providing and preparing the geologic data. Finally, we thank Texaco Inc. for permission to publish this paper.

SI Metric Conversion Factors

bbl x 1.589873
ft x 3.048*
psi x 6.894757
* Conversion

factor is exact.

E-Ol
E-Ol
E+OO
SPERE

Original SPE manuscript received for review April 19. 1984. Paper accepted for publication March 20. 1985. Revised manuscript received Sept. 10, 1986. Paper (SPE 12686) first
presented at the 1984 SPE/DOE Enhanced Oil Recovery Symposium held in Tulsa, April

15-18.

SPE Reservoir Engineering, November 1987