8/10/2019 EOR-IOR

1/21 JANUARY 1998 69

E O R / I O R

CO2 foam has been used as an effective

mobility-reducing agent for CO2 flooding

in the oil-recovery process. Recent researchindicates that some CO2 foams can provide

selective mobility reduction (SMR). SMR in

foams reduces CO2 mobility by more in

higher- than in lower-permeability cores in

laboratory experiments. Unlike Darcy flowof ordinary fluids in rocks, where the

mobility is proportional to rock permeabil-ity, the mobility of foam with SMR is less

than proportional to core permeability and

foam flows through higher-permeabilityrocks at a lower rate than would be expect-

ed for the existing pressure gradient. This

allows foam to flow at the same velocity in

high- and low-permeability regions in thereservoir, preserving the uniformity of the

flood front while propagating through

rocks with nonuniform permeability. Use of

a CO2 foam with SMR delays CO2 break-

through and provides a high displacement

efficiency in heterogeneous reservoirs.

FOAM-DURABILITY TEST

For this study a high-pressure foam-dura-

bility test apparatus was constructed and

screening tests were conducted successfullyto select surfactants for field foam applica-

tion. The test determined the foaming abil-

ity of each surfactant, the stability of foam,

and surfactant properties, such as the inter-

facial tension (IFT) between a surfactantand dense CO2 and the critical micelle con-centration (CMC) of a surfactant. The

foam-durability apparatus comprises a CO2source tank, a visual cell made from a trans-parent sapphire tube, a buffer-solution

cylinder, and a pump. The major partof the

system, the CO2 tank and sapphire-tube

high-pressure cell, is contained in a tem-

perature-controlled water bath. The buffer-solution cylinder and the pump are

installed outside of the water bath, and

their temperatures are maintained at the

same temperature as the water bath by a

separate temperature-control system.During operation, the sapphire visual cell

is first filled with the solution to be tested.Once the system reaches the desired pres-

sure, the dense CO2 is introduced through a

needle at the lower end of the cell. The CO2is drawn upward inside the cell. The densi-

ty difference between dense CO2 and the

tested solution causes CO2 bubbles to form

and collect at the upper end of the cell.

These bubbles will either form a layer offoam-like dispersion at the top of the sap-

phire tube or coalesce into a clear layer of

dense CO2, depending on the effectiveness

of the surfactant. After 1.75 cm3 of CO2 has

been introduced into the sapphire tube, thepump is stopped and the length of time that

the formed foam persists is measured.

Surfactant solutions (1 wt% active compo-

nent) were prepared by dissolving the sur-

factant as received from the suppliers into abrine system consisting of 5.6 wt% NaCl

and 1.4 wt% CaCl2. Different concentra-

tions of the surfactant solution were pre-

pared by diluting the batch solution withthe 7 wt% brine. All screening tests were

conducted at 77F and 2,000 psig.

By measuring the time required to form a

bubble at the needle in the sapphire tube

and the number of bubbles formed within acertain time period, the average volume

and radius of each bubble is calculated.Once the average radius of dense CO2 is

known, the IFT between surfactant solu-tion and dense CO2 can be calculated.

Results and Discussion. The IFT decreas-

es with surfactant concentration and levels

off at a region where the IFT no longer

decreases as surfactant concentrationincreases. The concentration at which the

interfacial properties between surfactant

and CO2 show no significant change is the

CMC and can be graphically determined.

The IFT curves and CMC values vary withsurfactant formula. The CMC values for

Surfactants 1 through 5 are 0.04, 0.06, 0.07,

0.07, and 0.35 wt %, respectively.

The foaming ability of a surfactant is

defined as the ease with which a bubble isformed at the needle when the surfactant

CO2-FOAM FLOODS:

FOAM PROPERTIES AND

MOBILITY-REDUCTION EFFECTIVENESS

This article is a synopsis of paper SPE

37221, Assessment of Foam

Properties and Effectiveness in Mobility

Reduction for CO2-Foam Floods, by

Jyun-Syung Tsau, SPE, and Reid B.

Grigg, SPE, New Mexico Petroleum

Recovery Research Center, New

Mexico Inst. of Mining and Technology,

originally presented at the 1997 SPE

International Symposium on OilfieldChemistry, Houston, 1821 February.

Fig. 1Decay of CO2 foam with Surfactant 4.

8/10/2019 EOR-IOR

2/21

70 JANUARY 1998

E O R / I O R

solution contacts the CO2. Durability is

defined as the persistence of foam bubbles

after a standard volume of CO2 has been

introduced. A cathetometer is used to mea-sure the foam height and the weight of the

CO2 to allow calculation of the percentage

of foam inside the sapphire tube and assess-

ment of the persistence of foam. Foam wasfound to form more easily as the surfactantconcentration increases. The foaming abili-

ty of surfactant increases as the IFT between

CO2 and the surfactant solution decreases.

The longest-lasting foams were not neces-

sarily found at the best foaming conditions.For Surfactant 4 (Fig. 1), there is an opti-

mum concentration (0.075 wt%) at which

the foam has the best stability or the longest

durability. The persistence of foam decreas-

es at concentrations either above or belowthis optimum concentration. This trend was

observed with Surfactant 1, with an opti-mum concentration of 0.05 wt%. In both

cases, the optimum concentration is close to

the surfactants CMC (i.e., 0.07 forSurfactant 4 and 0.04 for Surfactant 1). The

bubbles formed by Surfactants 5 and 2 coa-

lesced in less than 1 minute, whereas most

of the bubbles formed by Surfactant 3 lasted

longer than 90 minutes. The optimum con-centration for generating the longest-lasting

foams were not identified for Surfactants 5,

2, and 3. Surfactant 3 generates the most

stable foams, followed by Surfactants 4, 1, 2,and 5 (listed by decreasing level of stability).

FOAM-MOBILITY TEST

Core systems containing well-defined high-

and low-permeability regions were con-

structed to assess the flowing-foam proper-ties and verify the existence of SMR in het-

erogeneous porous media. This experiment

used two well-defined permeability regions

in capillary contact arranged in series. The

series assembly uses two 1/2-in.-diameter

cores approximately 3 in. long. The core-holder is fitted with five equally spaced

pressure taps so that the middle tap is near

the junction of the cores. The abutting endfaces of the cores are carefully cut perpen-dicular to their axes and ground flat before

mounting end to end. The space between

the two core faces is filled with fine sand.

Pressure differences between each pair of

pressure taps is recorded. The fluids flowinginto a foam generator and the composite

core are injected by two pumps, a positive-

displacement pump for the CO2 and a pis-

ton pump for brine or surfactant solution.

The pressure is maintained at an almost

constant level by leading the output fluids

into a backward-running piston pump.When the experimental conditions reach

steady state, pressure drops in each segment

of the core are recorded as functions of time.

The mobility of injected fluid, defined as theratio of Darcy or superficial velocity of the

fluid to the average pressure gradient along

each segment of core, is calculated and com-

pared for different injection conditions.

The foam generator and core sample arepreflushed with synthetic brine for at least 40

pore volumes (PV) before the brine perme-

ability measurements are begun. The hetero-

geneity of the series composite core was

determined by measuring the brine perme-abilities for four different sections along the

core. Following the permeability measure-

ments, dense CO2 and brine were injected

simultaneously into the core sample. The

mobility of the two-phase mixture was mea-

sured for each core section and used as a ref-erence for later comparison. After establish-

ing the baseline, a sequence of foam experi-ments was performed. To satisfy the adsorp-

tion requirement, 50 PV of surfactant solu-

tion was displaced. Then CO2 and surfactantsolution were injected into the core until

steady-state conditions were reached. Foam

mobility was measured. The ratio of volu-

metric flow rate of CO2 to aqueous phase was

maintained at four to one. The total injectionrate was varied from 5.0 to 15.0 cm3/hr cor-

responding to velocities of 3.1 to 9.4 ft/D.

The two composite cores had permeabilities

ranging from 525 to 128 md for Core 1 and

819 to 106 md for Core 2. Surfactant con-centrations of 0.1 wt% were used in Core 1

experiments, while 0.05 wt% surfactant solu-

tions were used in Core 2 experiments.

Results and Discussion. Comparison of themobility data in the first three core sections

indicates that the mobility of CO2/brine is

reduced by the addition of surfactant. Foam

mobilities are significantly lower than the total

mobility of CO2/brine. This mobility reduc-

tion varies with surfactant, surfactant concen-tration, and flow condition. In general, foam

mobility is lower when foam is generated at

higher surfactant concentrations or whenfoam is displaced at a lower injection rate.

When mobility dependence on rock per-

meability is examined, SMR is also found to

depend on the surfactant type, concentra-

tion, and flow rate. When the mobility of

CO2/brine or CO2/foam is plotted vs. the

sectional permeability, the slope of the lineindicates the degree to which the the mobil-

ity of fluid depends on the permeability of

porous media. A slope of one indicates that

the mobility of the fluid is proportional tothe rock permeability as described by

Darcys law. A value of less than one shows afavorable dependence of SMR that will lead

to a more uniform displacement front when

the fluid is flowing through heterogeneous

porous media. In general, the value of theslope decreases when surfactant is added to

the brine as a foaming agent. This suggests

that foam is useful in correcting the nonuni-

form flow of CO2 and brine in a porous sys-

tem containing differing permeabilities. Atlower velocities, the value of the slopedecreases, indicating a more favorable SMR

occurs at a lower displacement rate.

When results from the foam-durability

tests are compared with the mobility tests,

the stability of foam in the bulk phase canbe correlated with the effectiveness of

mobility reduction of flowing foam in the

porous media. The mobility reduction is

enhanced as foam stability increases. Themobility-reduction factor (MRF), defined as

the ratio of total mobility of CO2/brine to

foam mobility, increases with the foam life.When foams become more stable, more

resistance to flow results in a higher mobili-

ty reduction. On the basis of these observa-tions, the capability of surfactant in stabiliz-

ing the bubble file or lamellae in the porous

media is believed to be the most likely rea-

son for the effectiveness of foam in reducing

the mobility of CO2.Use of a proper CO2 foam could minimize

the mobility contrast between high- and low-

permeability zones in reservoir flow, increas-

ing the efficiency of oil displacement.

Experimental research indicates that the SMRproperty of CO2 foam is real. It is observed in

parallel- and series-core tests with capillary

contact and can be presumed to function

similarly in actual field situations.

CONCLUSIONS

1. The stability of foam in the bulk phase

can be correlated with the performance of

foam flowing in porous media. When com-

paring different surfactants, greater foamstability gives more mobility reduction in

foam displacement.

2. The MRF increases as the reduction

factor of the IFT between CO2 and theaqueous phase increases.

3. An optimum concentration exists atwhich the most stable foam in the bulk phase

is formed. This optimum concentration is

close to the CMC of each surfactant solution.

4. Factors that favor reducing the mobili-ty of CO2/brine also lead to a more favorableSMR when foam flows in a composite core

consisting of differing permeabilities.

Please read the full-length paper for addi-

tional detail, illustrations, and references.

The paper from which the synopsis hasbeen taken has not been peer reviewed.

8/10/2019 EOR-IOR

3/21

JANUARY 1998 71

E O R / I O R

After the 1973 oil embargo, the U.S. gov-

ernment funded several initiatives to pro-

vide additional worldwide resources as a

means of lowering oil prices. One of these

initiatives was Project Deep Steam. Thisproject evolved on the theory that signifi-

cant reserves of heavy oil, worldwide,

were beyond the reach of surface-generat-

ed steam because of heat losses in the tub-ing. Two approaches were taken to achieve

the goal of steam delivery to deep forma-

tions. One was to improve or develop newinjection strings to minimize heat loss,

and the second was to design and demon-strate the operation of a device that would

generate steam in the wellbore at the

depth of the formation. Sandia Natl.

Laboratory was given a U.S. $23 million

budget for a 4-year program beginning in1976. Sandia initiated an in-house effort

for steam-generator development. Both

the insulated-injection-tubing develop-

ment and testing were carried out by

external contract.The outside diameter (OD) of the down-

hole steam generator could not exceed 4.5

in. This OD would fit inside 7-in. casing. To

achieve large-magnitude firing rates, thecombustion process would have to occur at

high pressure. Steam would be generated in

either of two ways. One was by a tradition-

al heat-exchanger surface that separated the

hot flue gases from the water. Adequate sur-face for efficient heat transfer was obtained

by heat-exchanger lengths of up to 100 ft.

The cooled flue gases were exhausted up

the annulus. Combustion did not have to

occur at pressures as high as those required

for injection. In the second method, com-bustion occurred at high pressure and

water was injected directly into the hot flue

gases. Then, both steam and flue gases were

injected into the formation. Injection of theflue gases proved to be either a benefit or a

problem; one benefit was that it was helpfulin the air-quality permitting process.

Operation of both downhole steam genera-

tors could be carried out on the surface,

with injection tubing transporting the

steam downhole. The Sandia developmentand one of the outside contracts were based

on a direct-contact steam generator. The

other outside contract was for an indirect-

contact steam generator.

FIELD OPERATIONSKern River Field. The equipment for this

test was sized for a delivery of 5 million

Btu/hr at a maximum pressure of 400 psi.Along with the steam, nitrogen and carbon

dioxide were generated. The combustion

was carried out with minimum excess air.

Because this was the first operation of a

direct-contact steam generator of Sandiadesign, the generator was operated on the

surface rather than downhole. The OD of

the steam generator was 4.5 in., and liquid

propane was the fuel. The Kern River field

had been steamflooded for a number ofyears before this test, and the injector well

communicated quickly with surrounding

production wells because of this previous

steaming. No particular increase in produc-tion was noted because this was a rather

mature steamflooded field.

Wilmington Field. For this operation two

different generators and support systems

were demonstrated simultaneously. Thefirst was operated downhole, and the sec-

ond was operated on the surface. Both were

fired with air as the oxidizer and diesel as

the fuel.Downhole Air/Fuel Generator. The out-

put for this system was sized for 5 million

Btu/hr at a pressure rating of 1,400 psi.

Installation of the generator was a rather

complex task. The connectors from the

generator to the surface consisted of twojointed tubulars and four continuous tub-

ing lines. The air line was 23/8-in. jointed

tubing and was the load-bearing element

for the installation. Water was conducted

down 1.66-in. jointed tubing that had to be

made up by hand tools after the larger tub-

ing had been made up with power tools.The other lines were continuous coiled tub-

ing that had to be pulled out of the way by

hand to make up the jointed tubulars.

These four lines were attached to the 23/8-

in. tubing by a custom-made cast-alu-

minum clamp. Because diesel contains sul-fur that produces acids when burned, caus-

tic was injected through a 1/4-in. line to

neutralize the generator effluent. Two

additional tubing strings were installed to

verify operation. One was a continuous,fully sheathed, 1/8-in. thermocouple line

and the other a1

/4-in. sample return line.The final line was a 3/8-in. fuel line. A

thermal packer with a short stinger hung

below the generator. A cap on the end ofthe stinger was used to keep annulus flu-

ids isolated from the generator assembly

before operation.

When the installation was complete,

the operation was begun by blowing theplug off the end of the stinger. Then, a

volume of air was injected through the

generator to create a compressible cush-

ion before ignition. Next, water injection

began. Ignition was achieved by use of aslug of pyrophoric fluid that was injected

into the diesel line that had first been

flushed with water. When ignition was

achieved, pressure transients wereobserved at the surface monitoring point.

Usually 1 to 2 minutes were required for

flows to quasistabilize. An increasing

pressure was usually observed for a signif-

icant period of time following initiation.Several problems occurred in the down-

hole equipment. The lines for the fuel, air,

and water had filters and check valves just

above the steam generator to ensure that

critical ports were not plugged and thatfluids from the formation could not flow

back through the generator during shut-

downs. In spite of corrosion inhibitors in

the water supply, large quantities of corro-sion debris sloughed from the water line

and began plugging that filter. The gener-

ator had to be pulled, and the mesh size

and accumulation volume increased.

Improved production was observedduring this test because of the injected

flue gases and their communication with

surrounding production wells. The flue

gases reduced the caloric value of the pro-duced market gas, but the blended pro-

REFLECTIONS ON A DOWNHOLE

STEAM GENERATOR

This article is a synopsis of paper SPE

38276, Reflections on a Downhole

Steam-Generator Program, by A.B.

Donaldson, SPE, New Mexico

Highlands U., originally presented at

the 1997 SPE Western Regional

Meeting, Long Beach, California, 2527June.

8/10/2019 EOR-IOR

4/21

72 JANUARY 1998

E O R / I O R

duction from the rest of the field kept the

value within specifications.

Surface Oxygen/Fuel Generator.

Oxygen was used in one trial to eliminatethe large amount of noncondensable

nitrogen and increase the partial pressure

of carbon dioxide. This also eliminated

compression equipment. While the exper-iment with oxygen did not have the sameoperational problems as the downhole

unit, it had its share of difficulties. The

major problem was survival time of the

combustor can that isolated the combus-tion process from the injected water until

combustion was complete. None of the

conventional stainless steels used had suf-

ficient strength to survive the very high

thermal stresses caused by an oxygen/hydrocarbon flame.

At the end of operations in the Long

Beach,California, area, the apparent scarci-ty of oil disappeared and government fund-

ing was reduced. A number of the projectteam members elected to leave Sandia and

obtain nongovernment funding to attempt

commercialization of the technology.

Enhanced Energy Systems, Inc. (EESI). A

number of deficiencies in the Sandia equip-

ment were obvious after the field experi-

ences. Manpower requirements and sophis-

tication were excessive for oilfield applica-tions. The equipment was redesigned for

commercial applications, with the goal ofone-person operation and the use of field

gas or crude oil for generator fuel. The con-

cept of steamdrive was replaced by huff n

puff operation. Installation of the units

downhole was difficult, and many of thelater operations were conducted with the

generator on the surface.

Hondo, Texas. The formation is shallow

chalk with poor permeability and the pay

is thin (from 7 to 40 ft). A major oil com-pany had attempted steaming in a shallow

sand zone in the area, but the return was

insufficient to continue the effort. Seven

wells were stimulated. The casings hadsmall diameters, and the steam generatorwas installed at the surface with effluent

conducted to the wells. Neither insulated

injection string nor packer was used. Flue

gases were injected into the formation with

the steam. Because of poor injectivity, theequipment was operated at the lowest out-

put, 2 million Btu/hr, at the maximum

pressure, 1000 psi. The fuel used for the

steam generator was site crude, and the

location water had a high hardness.Production was improved from primary

production but was not sufficient for fur-ther operation in this field.

Hospah, New Mexico. The formation is

a consolidated sand, approximately 1,100

ft deep, with a thickness from 10 to 20 ft.

The wells were not thermally insulated,and a downhole packer was not used. Fuel

for the generator was old diesel. The injec-

tivity of the wells was good, and some flue-

gas production was observed at wells alongthe path of least resistance. The oil pro-

duced was in the form of an emulsion.

Various chemicals were used in an attempt

to break the emulsion, but none was suc-

cessful. Production increases were attrib-uted to creation of a high-pressure pocket

of gas that blocked the cold-water seepage

into the formation.

Long Beach. This customer was the first

to purchase a full line of equipment.

Because of the complexity of equipmentand procedures, EESI personnel were

retained to operate the equipment. Becausefield gas was available at this location, both

the generator and compressors were fueled

by site gas. Expansion cooling of the fuelgas led to icing conditions inside the con-

trol valve, and the flow could not be

bypassed. A desiccant was used to dry the

gas. A major problem was controlling deliv-ery of fuel, air, and water 2,000 ft down-

hole. A thermal packer was used to isolate

the annulus from the reservoir, but filters

and check valves were used sparingly. The

equipment was designed to operate atapproximately 8 million Btu/hr at pressures

up to 1,850 psi. Two wells were steamed

with downhole units. One well was lost in

this operation because of parted casing at

400 ft.As operating experience was gained,

many of the urgent operational bugs were

found and corrected, and some reasonable

periods of steam injection were experi-

enced. The generator had to be pulled onone or two occasions because of deteriora-

tion of the combustion-chamber wall. At

this time, operation of the steam generator

was returned to the surface and no further

installations were made downhole.Two other nearby wells were cyclicly

steamed from this surface location.

Insulated surface-injection piping was used

to transport the steam to the wells. Some ofthe most reliable operations occurred dur-

ing this time. Peak production immediately

after bringing the well back into production

sometimes exceeded 500 BOPD, compared

with a prestimulation production of 20 to40 BOPD. The customer calculated total

production costs of approximately U.S.

$12/bbl, which was the lowest of any of

their production. Although conventionalsteam would probably have been economi-

cally viable at this location, compliance

with air-quality regulations was difficult.

THE COMPLETION

While conventional steam-generation tech-

nology has the advantage of being the

accepted technology for shallow reservoirs

with good injectivity, there are some areaswhere improvements can be made. In insu-lated injection tubing currently available,

heat losses at bare collars, expansion joints,

packers, and subs almost negate the bene-

fits of insulation unless the annulus is com-pletely dry. This conclusion is based on a

water-reflux mechanism that occurs even

when the annulus fluids are initially

expelled. Residual water film or leaks into

the annulus can generally be expected.

CONCLUSIONS

In general, no advantage was evident inplacing the generator downhole, particu-

larly when injectivity of gases was a prob-

lem. It was evident that injection of thecombustion gases was desirable in some

cases and, in one case, necessary.

Formations with poor injectivity are a

major resource awaiting a viable recovery

mechanism. Placement of a direct-contactsteam generator downhole is not a solution

to this problem.

RECOMMENDATIONS

These recommendation focus on whatmodifications can be made to the downhole

steam generator for successful stimulation

of tight formations.

New Concepts. A downhole steam genera-tor that does not produce nitrogen gas

appears to offer potential for low-injectivity

applications. There have been field trials of

radio-frequency heating devices as well as

one that uses electricity for energy. It isuncertain whether any of these have under-

gone extensive field tests. One other possi-

bility is to transpose submerged combus-

tion technology developed for toxic-wastedisposal to oilfield applications. Anotheridea is to use the heat of mixing two fluid

components that can be separated by distil-

lation at the surface. A downhole heat

exchanger would be used to heat feed water

to steam, and the mixture of the two com-ponents would be returned to the surface

for processing.

Please read the full-length paper for

additional detail, illustrations, and ref-

erences. The paper from which the syn-

opsis has been taken has not been peerreviewed.

8/10/2019 EOR-IOR

5/217 4 JANUARY 1998

E O R / I O R

Foam has the potential to relieve several

common problems by better areal sweep,

better vertical sweep, less viscous fingering,

and lower handling costs when compared

with gas-injection improved oil recovery(IOR). Foam may be introduced by contin-

uous coinjection of gas and surfactant solu-

tion or by injection of a slug of surfactant

solution followed by injection of a gas slug,known as a surfactant-alternating-gas

(SAG) process. SAG injection has certain

advantages over continuous foam injectionin foam IOR processes. SAG injection min-

imizes contact between gas and water in theinjection facilities, reducing corrosion, and

can achieve high injectivity and low mobil-

ity at the displacement front. High injectiv-

ity results as foam near the well dries out,

weakens, and collapses, while stronger,wetter foam farther from the well maintains

mobility control. Recent simulation results

show that SAG processes can overcome

gravity override with less increase in injec-

tion-well pressure than is possible withcontinuous foam injection.

SCALEUP OF LABORATORY RESULTS

There are a number of published coreflood

studies where gas is injected into core sam-

ples presaturated with surfactant solution.Some show foam persistence for many pore

volumes (PVs) of gas injection. However,

extrapolating these results directly to field

scale is dangerous. Because of dispersive

effects, the mobility within the shock frontmay exert influence in the length scale of a

coreflood that does not occur in the field.

Processes such as foam generation may

occur slowly on the time scale of a core-flood but be virtually instantaneous on thetime scale of a field project. Unsteady-state

displacements in a coreflood may not be

directly proportional to field scale, and

foam behavior in a reservoir may not be

directly proportional to unsteady-state dis-

placements in a coreflood. Capillary endeffects in a coreflood study may prevent

water transport out of the core, prolonging

foam life beyond that observed in a larger

system without this effect. A more reliable

procedure is to derive the fractional-flowcurve from the coreflood and then scale to

reservoir size with fractional-flow methods.

APPARATUS AND MATERIALS

An unfired Berea core with a permeability of

720 md and porosity of 0.22 was used in theexperiments. It was cut into a 9.5-in.-long

right-circular cylinder. An N2 gas phase

with a 1.0-wt% NaCl/0.01-wt% CaCl2/1.0-

wt% surfactant aqueous phase was used.

The coreholder, positioned downstream ofthe foam generator, held the core in a verti-

cal position. The coreholder was designed

to be lightweight so that the weight of the

core could be measured effectively during

the flood to determine water saturation, Sw.

Flow lines were flexible, transparent nylon,

and baffles were installed to restrict air flow

around the apparatus and minimize exter-

nal forces on the coreholder. Ports divided

the length of the core into three sections of2.6, 2.75, and 4.15 in., respectively, from

inlet to outlet. The weight of the system,

with Sw in the core equal to one and zero,

was measured before the tests, allowing

determination of Sw during the test.

One goal of these tests was to measurecap-illary pressure, Pc. A probe was designed that

used a differential-pressure transducer to

measure the pressure in the gas phase on oneside and pressure in the water phase on theother side of the transducer. A backpressure

regulator maintained a steady backpressure of

147 psi on the entire system during the two-

phase flow. Differential-pressure transducers

measured pressure drop, p, across each ofthe three sections of the core. All transducershad a range of 0 to 80 psi, and were calibrat-

ed before the test. Data were recorded by a

computerized data-acquisition system.

EXPERIMENTAL STRATEGY

Tests were performed at room temperature.The core was vacuum saturated with brine,

then hundreds of PVs of brine were injected

as backpressure was changed between 0 and

100 psi to eliminate gas from the core.

Backpressure was set at 147 psi at the end of

the extended brine injection and held con-stant. Dozens of PVs of surfactant solution

were then injected. Gas was then introduced

with the surfactant solution at a water volu-

metric fraction, fw, of 0.2. fw was incremen-

tally reduced as pressure responses stabi-lized. When a minimum value of fw=0.002

was reached, water injection ceased and anextended period of gas injection began.

EXPERIMENTAL RESULTS

Gas/Water Coinjection. The value of Sw was

derived from measured weight. Two methods

of estimating the amount of water in the dead

volume were used. The first assumes that the

imposed fw equals the water volume fractionin the endcap. The second is based on endcap

weight that changes in a shorttime compared

with changes that occur across the core. Both

methods give qualitatively similar results.

Water relative permeability,krw

, values werecomputed from measured p with a form of

Darcys law and with the assumption of a

water viscosity of 1.0 cp. The measured p in

Section 2 was used in all calculations to avoid

end effects. Pc andp were measured, and Pcof the Berea computed from Sw values, by useof capillary pressure curves for unfired Berea

during primary drainage, at values of N2/

aqueous interfacial tension (33 dynes/cm),

measured at room temperature. The Berea Pcwas computed to compare measured Pc withand without foam under the same conditions.

Foam strength is characterized by its

resistance factor (RF), the ratio of mobilitywithout foam to that with foam at the sameSw. An RF value of 1 indicates no foam,while an RF value of 100 means that the

presence of foam has reduced gas mobility

100 times more than gas mobility without

foam at the same Sw. Computation of RF

requires knowledge of gas relative perme-ability, krg, without foam. Because krg is

high, and changed very little with the low

values of Sw used in these tests, the value of

krg was assumed to be 1.0.

Results.When gas and water are injectedtogether, the foam is strong (RF2,000)

COREFLOOD STUDY OF SURFACTANT-

ALTERNATING-GAS FOAM PROCESSES

This article is a synopsis of paper SPE

38318,Coreflood Study of Surfactant-

Alternating-Gas Foam Processes:

Implications for Field Design, byK.R.

Kibodeaux, SPE (now with Texaco E&P

Technology), and W.R. Rossen, SPE, U.

of Texas, originally presented at the

1997 SPE Western Regional Meeting,Long Beach, California, 2527 June.

8/10/2019 EOR-IOR

6/21

beforefw=0.008, then there is catastrophic

weakening of the foam between fw=0.02

andfw=0.008, followed by a gradual weak-ening asfw decreases. Foam strength is still

appreciable (RF=25) after breakage, and

foam collapse is not complete. When the

foam breaks, fluid is absorbed and swellingresults (imbibition), accompanied by adecrease in Pc. As Sw increases, krw also

increases as Pc drops. Measured values of Pcwere high, compared with those expected

without foam.

DISCUSSION

In these tests, Pc was measured during foam

flow in consolidated porous media for the

first time. Foam is strong asPc increases to

some limiting value. There is a critical Pcabove which foam lamellae rupture, and a

limiting Pc value is expected to exist inporous media at which a strong foam weak-

ens drastically. In these tests, the value of the

critical Pc was approximately 12 psi. Otherinvestigators have reported a foam film sur-

viving at Pc>17 psi. The Pc measurements,

although unexpectedly high, were valid

indications of the actual values in the core.

One goal of this work was to apply frac-

tional-flow methods and a coreflood-derived

fw curve to scale up laboratory coreflood

results. A time/distance diagram for an SAGprocess was generated from the fractional-

flow curve. Mobility is high at the initial con-

ditions behind the shock front. At the shock

front, mobility is low, with spreading waves ofgently increasing mobility behind it. A foamof very low mobility exists in a thin, moving

front composing the shock. Within this zone,

foam forms, strengthens, exceeds critical Pc,

and weakens. Fractional-flow theory states

that all the points between initial conditionsand the shock on the fractional-flow curve,

including the strong foam near fw=0.02, lie

inside the the shock. This extremely strong

foam within the narrow shock front improves

the mobility ratio of the displacement with noadverse effect on injectivity.

CONCLUSIONS

1. Pc was measured during foam flow in

consolidated porous media for the firsttime. Unexpectedly high Pc were measured

in the presence of strong foam.

2. Sw declined and Pc increased as injec-

tion fw was reduced in steps to a point

where foam abruptly weakened. This point

is defined as the limiting Pc. The shape of

the fractional-flow curve was similar to

those proposed previously.3. When the foam weakened, imbibition

occurred, with a increase in Sw and a

decrease in Pc. This imbibition event, as

well as multiple steady states of foam and amultivalued fractional-flow function, maycomplicate the use of fractional-flow meth-

ods for foam-performance prediction.

4. Coreflood results cannot be scaled direct-

ly to field performance because both evapora-

tion and capillary end effects alter water trans-port late in the flood. Using laboratory core-

floods to derive fw as a function of Sw, then

scaling up with fractional-flow methods or

computer simulation, provides more reliable

scaleup from the laboratory to the field.5. Results suggest moderate mobility

reduction in a broad region behind theshock front, while weakened foam near the

well allows good injectivity.

Please read the full-length paper for addi-

tional detail, illustrations, and references.

The paper from which the synopsis has

been taken has not been peer reviewed.

JANUARY 1998 7 7

E O R / I O R

8/10/2019 EOR-IOR

7/21

78 JANUARY 1998

E O R / I O R

On 17 May 1996, Total Minatome Corp.

initiated an enhanced-oil-recovery (EOR)

high-pressure air-injection project at the

Horse Creek field in North Dakota. This

project is the third high-pressure air-injec-tion program completed in the Ordovician

Red River formation in the Williston basin.

The Horse Creek field is 7 miles east of the

Cedar Creek anticline, in the south centralportion of the Williston basin. The field was

discovered in 1972 and comprised 15 pro-

ducing oil wells. In 1993, geological, labo-ratory, and reservoir-modeling studies were

conducted to evaluate the EOR potential ofthe field. On the basis of the results from

these studies, an EOR unit for this field was

formed in 1995. There are currently 11 pro-

ducing wells, three air-injection wells, one

monitor well, and one water-disposal wellwithin the unit area. Air is being injected

into the reservoir at a rate of approximately

8,500 Mscf/D at 4,700 psi discharge pres-

sure. Nine months after injection began,

the reservoir pressure continues to rise andproduction has increased from 293 to 400

BOPD.

RESERVOIR CHARACTERIZATION

Geology. The Horse Creek field is a strati-

graphic oil accumulation within theOrdovician Red River formation. The Red

River formation in this portion of the

Williston basin is subdivided into four

porosity zones. These porosity zones are

referred to as Zones A through D (Fig. 1).Air injection and production in the Horse

Creek field has been limited to Zone D.

The upper portion of the Red River for-

mation consists of a series of briningupward, cyclic carbonates deposited in asubtidal to supratidal environment on a

restricted shelf during the Red River marine

transgression. Rock types include lime-

stones, dolomites, and anhydrites.

Lithology varies from laminated mudstones

to heavily bioturbated packstones, wacke-

stones, and mudstones. Dolomitization of

the heavily bioturbated units has resultedin a secondary porosity within Zone D,

from which the Horse Creek field produces.

Stratigraphic cross sections, core data, and

petrophysical-facies maps indicate that

Zone D comprises two separate lobes thatare discontinuous across the field. The

HORSE CREEK AIR-INJECTION

PROJECT: AN OVERVIEW

This article is a synopsis of paper SPE

38359, The Horse Creek Air-Injection

Project: An Overview, by B.C. Watts,

T.F. Hall, SPE, and D.J. Petri, SPE,

Total Minatome Corp., originally pre-

sented at the 1997 SPE Rocky

Mountain Regional Meeting, Casper,Wyoming, 1821 May.

Fig. 1Horse Creek field type log.

MASSIVEANHYDRITE

GAMMA RAY

8900

9000

9100

9100

GAMMA RAY

DENSITY @

CORE 2

CORE 1

C ZONE

D ZONE

A ZONE

TOP RED RIVER

B ZONE

North Dakota

DENSITY POROSITY30 10

LITHOLOGY

CORE DESCRIPTION

DEPOSITIONALENVIRONMENT

LAMINATEDMEMBER

BURROWEDMEMBER

SUPRATIDAL

SUPRATIDAL

TO

UPPER

INTERTIDAL

(RESTRICTED

FAUNA)

LAMINATED

DOLOMITIC

MUDSTONE

HEAVILYBURROWED

PACKST/WAKESTDOLOMITE

HEAVILYBURROWEDDOLOMITICMUDSTONE

TIGHTLY BURROWED

LIME MUDSTONE

SUBTIDAL(NORMALMARINE)

SUBTIDAL TOLOWER INTERTIDAL

(NORMALMARINE)

8/10/2019 EOR-IOR

8/21

lower unit of Zone D is one continuous

unit that occurs across the entire field area.

Porosity and Permeability. Porosity andpermeability in Red River Zone D are

related to several stages of dolomitization.

The highly burrowed subtidal to lower-

intertidal lime muds in Zone D, whichcontain mainly macroporous pore throats(2 to 20 m), contain most of the reser-

voir-quality porosity and permeability.

Burrowing appears to have prepared these

rocks for dolomitization and porosity

development as compared with the sur-rounding nonburrowed, laminated

supratidal to upper-intertidal lime muds,

which, although porous, contain mainly

microporous (0.2 to 0.5 m) pore throatsand are nonproductive. On the basis ofcore data, thin-section descriptions, and

log analysis, reservoir-quality porosity isintercrystalline and ranges from 8 to 20%

and averages 16%. Reservoir thickness

ranges from 0 to 45 ft and averages 20 ft.Permeability ranges from 1 to 97 m, aver-

aging 10 to 20 md.

Net Pay. Net-pay thickness for Zone D was

determined by log analysis. The log datawere calibrated with core data, and petro-

physical values were calculated for all wells

from digital data. Net-pay cutoffs deter-

mined by porosity, permeability, produc-

tion, and relative oil/water permeabilitydata were established at 12% porosity and

water saturation < 50% for the primary

(moveable) oil reserves.

Original Oil in Place(OOIP).Well controlwithin the field is adequate to give a reliable

estimation of effective OOIP. A porosity

cutoff of 12%, average water saturation of

35%, and a formation volume factor of

1.205 yield an effective OOIP of 45.7 mil-lion bbl.

PRODUCTION PERFORMANCE

AND EOR POTENTIAL

The primary producing mechanism for

Zone D is liquid and rock expansion. A200- to 300-ft oil column has developed

within the macroporous petrofacies.

Production has included a water cut of

approximately 60%, which has remainedconstant since the initial field production.

The constant water cut and steady decline

in reservoir pressure indicate that the trap-

ping mechanism is stratigraphic and sug-

gest that there is no active aquifer. Remain-ing primary reserves of 1.25 million bbl

were calculated by decline-curve analysis.

The primary recovery factor for the field is

estimated to be 9.9% of OOIP.

EOR Potential. Several different EOR

methods were considered for the Horse

Creek field. Waterflooding was ruled out

because of oil and water relative permeabil-

ity differences which, coupled with the cur-rent water saturation, impeded the ability ofthe oil to form a bank, reducing recovery.

Gas (N2 and CO2) injection was eliminated

because of excessively high costs.

The process finally selected was high-

pressure air injection. This method provid-ed all the benefits of waterflooding and gas

injection. Technical parameters contribut-

ing to this decision include the following.

1. Reservoir temperature of 220F, indi-

cating that in-situ oxidation would occurspontaneously without downhole igniters.

2. Oil with a gravity of 32 API and agood affinity for oxidation.

3. An in-situ oxidation process with ahigh oxygen-utilization rate that indicated

good oil recovery.

4. Rapid field repressurization that

would restore reservoir pressure and

enhance early oil recovery.The decision to use the air-injection

method was influenced by the favorable

comparison of reservoir and fluid parame-

ters from the Horse Creek field with the

successful air-injection programs at twoother units in the same field.

RESERVOIR AND

LABORATORY STUDIES

A series of reservoir models, simulations,

and laboratory studies was conducted toevaluate the EOR potential and to ensure

that the most efficient and economical

recovery process was used.

Pressure/Volume/Temperature (PVT)

Analysis.A PVT analysis of the oil was per-

formed and included differential liberation

and N2 swelling tests.

Waterflood Study. A waterflood studyindicated that the field was not a good

waterflood candidate because of unfavor-

able water/oil relative permeabilities and a

low incremental-recovery factor. On the

basis of this study, waterflooding was ruledout as a possible EOR process.

Black-Oil Models. The first phase of the

reservoir modeling consisted of a black-oil

model of the southern portion of the field.

The objective of this model was to confirm

the volumetric assumption in this portionof the field and to estimate the incremental

recovery from gas injection. Results from

this model were encouraging and led to the

decision to construct a full-field model.The objectives of the full-field black-oil

model were (1) to acquire a full-field his-

tory match and confirm the volumetric

OOIP, (2) to verify the lack of water influx

into the reservoir, (3) to examine the reser-voir response to gas injection, and (4) to

provide a reservoir model for subsequent

simulations. Several different production

scenarios were tested with various injec-

tion rates and injection-well locations.Model results indicated that the best pro-

duction scenario included three injection

wells with an initial injection rate of 10

MMscf/D. The model estimated incremen-tal oil recovery of 7.6 million bbl. Fast and

reliable results from the black-oil model

enabled an initial economic projection to

be made for the project.

Accelerating Rate Calorimeter (ARC)

Tests. ARC tests were conducted to assessthe oil oxidation kinetic parameters under

quasiadiabatic conditions. Results from

these tests are qualitative but help to deter-

mine the range of conditions under which

the oil will react with air. Two main oxida-tion reactions were detected by the ARC

tests. The first reaction began at 279 to

315F and was described as a low-tempera-

ture oxidation (LTO) reaction that pro-

duced polar compounds. The second reac-tion, which began at 404 to 441F, is the

reaction that would produce significant

quantities of CO and CO2. Results from the

ARC test indicated that the oil would react

under low-temperature conditions.

Combustion-Tube Burns. Two combus-

tion-tube burns were performed at the U.

of Calgary to quantify the combustion

characteristics of the rock, oil, and brine.Both burns were conducted in a 6-ft-long

vessel with a 4-in.-diameter core from the

Horse Creek field. The experiments used

different air-flux rates and ignition tem-peratures to assess the displacementprocess. The results indicated that a pro-

gressing temperature front is established at

approximately 600F. This temperature

range is indicative of LTO. The steady

thermal-front displacement, low airrequirement, low fuel load, and high per-

centage of oil recovered indicated an effi-

cient recovery process.

Thermal Model. To obtain parameters to

describe the oil oxidation kinetics, a ther-

mal simulation of the combustion-tube-burn experiments was performed. A match

JANUARY 1998 79

E O R / I O R

8/10/2019 EOR-IOR

9/21

80 JANUARY 1998

E O R / I O R

of the combustion-tube results was

obtained by use of a simple, complete oxy-

gen combustion with no coke formation.

Air-Injection Radial Model. An air-injec-

tion radial model was built with a thermal

simulator to study the conditions around

an air-injection well. This was importantbecause air-injection startup is the mostcritical phase of project operations. The

radial model used the reservoir properties

from the full-field black-oil model and the

kinetic parameters from thermal modelingof the combustion-tube experiments. The

following are some important results from

the thermal model.

1. Temperature Profile. The temperature

of the oxidation front was from 500 to700F. Following 15 years of injection, the

heated zone was estimated to be a cylinder

with a diameter of approximately 900 ft.2. Oxygen Behavior in the Reservoir.

Oxygen was consumed in a rapid and sta-ble reaction. After 15 years of injection,

oxygen did not move beyond a radius of

500 ft from the injection well. The model

predicted that no oxygen would reach a fic-

titious producer located 3,500 ft from theinjection well.

3. Hetrogeneities. Sensitivity cases with

vertical permeabilities from 1 to 300 md

showed no oxygen breakthrough and indi-cated that the oxygen did not go beyond a

radius of 700 ft from the injection well.

Full-Field Thermal Simulation. The final

phase of reservoir study included a full-field model with a thermal simulator. The

model used the reservoir properties fromthe full-field black-oil model along with a

compositional description of the oil and

introduction of chemical reactions that

enable oil oxidation and CO2 formation.

Predictive-model runs were performed for a20-year period, with air-injection rates

averaging 10 MMscf/D. Several different

injection scenarios were examined to deter-

mine the optimum location and number ofair-injection wells. The main results fromthe thermal simulation follow.

1. Incremental oil recovery was estimat-

ed at 7.2 to 7.9 million bbl.

2. The oxidization front did not go

beyond 3,500 ft from the injection well.3. Oxygen breakthrough did not occur in

the producing wells.

4. The incremental recovery with air

injection was approximately 10% more

than with nitrogen injection.5. Gas sweeping and reservoir repressur-

ization are the main mechanisms influenc-ing incremental recovery in this light-oil

reservoir. Viscosity reduction is not a sig-

nificant factor.The results of the full-field thermal sim-

ulation confirmed the economic feasibility

of the project. Data from thermal simula-

tion helped to define a field production-

monitoring program that was implemented

in the field.

PRODUCTION FACILITIES

Engineering for the project facilities began

in March 1995. Detailed design of the com-

pressors was completed and contracts for

their construction were awarded in April1995. Detailed engineering for all other

phases of the compression facility were

completed, and construction on the plant

foundations and installation of the 13-mile

fuel-gas pipeline were initiated. The com-pressors were delivered in December 1995,

and construction of the remaining facilities,including piping, instrumentation, well

conversions, and installation of high-pres-sure air lines, was completed during the

first quarter of 1996.

The injection plant for the Horse Creek

project was designed (1) to build safety

and redundancy into all systems, (2) toprovide computer control and call-out

warning systems, and (3) to meter and

control rate and pressure accurately to

each injection well.

Compressors for the project were select-ed on the basis of field-proven perfor-

mance and ability to supply the project

requirement of 10 MMscf/D at 5,000 psi.

Two Integral 8 throw, seven-stage conpres-sors, driven by two V-16 turbo-charged

natural gas engines, are used to compress

the air.

Before air injection could begin, two pro-

ducing wells had to be converted to air-injection wells and one new injection well

had to be drilled and completed. All injec-

tion wells were fitted with high-pressure

wellheads and permanent air-injection

packers with 23/8-in. tubing with premium

two-step connections, and 20,000 ft of 23/8-in. 5,000-psi air-injection lines was

installed from the compression facility tothe injection wells. All production wells in

the unit are equipped with 23/8-in.

anchored tubing and have beam pumps.

Each producing well has its own produc-

tion facility.

Safety and Environment Factors. Special

precautions must be implemented to pre-

vent explosions of hydrocarbons when

using high-pressure air injection. To reduce

this risk, the following equipment, materi-als, and processes were employed.

1. Special packers, two-step tubing con-

nections, and innovative pressure testing to

minimize the potential for downhole leaks.2. Special high-temperature lubricants

for the compressors, wellheads, and tubing

to minimize the danger of explosion.

3. Passivation of all tubulars to reduce

internal rust, corrosion, and scale.4. Wellhead controls at the injection wells

to prevent backflow and overpresssuring.

5. Dual compressors to add redundancy

and maintain constant injection to the

reservoir.6. A nitrogen-deployment system to

ensure continuous positive pressure to the

reservoir in the event that both compres-

sors malfunctioned at the same time.7. An emergency generator to provide

automatic electrical backup in the event of

a power failure.

PROJECT PERFORMANCE

The average gas/oil ratio has decreased

with repressurization of the reservior. Theaverage bottomhole pressure has increased

approximately 550 psi. Average daily pro-

duction rate has increased from 293 BOPD

to approximately 400 BOPD. All these

changes were predicted by the reservoir-simulation studies. On the basis of the lim-

ited production history, the production

goal of 1,100 to 1,300 BOPD probably will

be met.

CONCLUSIONS

1. The EOR potential of the Horse Creek

field was determined by an integrated team

of geologists and engineers.

2. A black-oil model provided fast andreliable predictions of incremental recovery

and production required to make initial

economic projections and business deci-

sions for the project.

3. ARC and combustion-tube experi-ments provided the kinetic parameters for

thermal modeling and indicated that air

injection would be an efficient recovery

process.4. Thermal simulation of the project pro-

vided the ability to examine different pro-

duction scenarios to test injector locations

and production results.

5. Production results to date indicatethat the project is proceeding as predicted

by the reservoir models and simulation

studies.

Please read the full-length paper for addi-

tional detail, illustrations, and references.

The paper from which the synopsis hasbeen taken has not been peer reviewed.

8/10/2019 EOR-IOR

10/21

JANUARY 1998 81

E O R / I O R

Polymer flooding is an enhanced-oil-

recovery method with great potential in

the Daqing oil field. Many new problems

will be encountered producing crude oil

mixed with polymer solution. During thewaterflood development phase, the

oil/water mixture is treated as a

Newtonian fluid. Analysis of oilfield fluids

indicates that fluids produced by polymerflooding are non-Newtonian and that con-

ventional methods of calculating inflow-

performance-relationship (IPR) curvescannot be used. This paper considers the

variation of rheological properties of fluidproduced from polymer-flooded reservoirs

on the basis of core-displacement and rhe-

ological-property experiments.

RHEOLOGICAL PROPERTIES

OF IN-SITU FLUID

The in-situ fluid in polymer-flooded reser-

voirs is a mixture of aqueous polymer

solution and crude oil. Many studies have

been conducted to determine the rheolog-

ical properties of polymer solutions, butno studies have been published on their

rheological-property variation in the direc-

tion of flow. Rheological properties of

polymer solutions change continuously

along the flow direction as a result ofadsorption, retention, and shear degrada-

tion of the solution. In laboratory dis-

placement tests, the viscosity of a polymer

solution was measured at different outlet

points in an artificial column core.Apparent viscosity decreases in the direc-

tion of flow. Assuming a power-law fluid,

the consistency coefficient, k, decreases

sharply near the inlet of the core and con-tinues to decrease slowly in the directionof flow, and the power-law exponent, n,

increases in the direction of flow. This

indicates that larger molecules of the poly-

mer are degraded and the non-Newtonian

behavior of the polymer solution is

reduced in the direction of flow.

FLOW BEHAVIOR OF POLYMER-

SOLUTION/OIL MIXTURE

Core-displacement experiments were per-

formed in the laboratory to compare thebehavior of polymer-solution/oil mixtures

with different volume ratios with that ofsingle-phase polymer solution. When pres-

sure drop is plotted vs. equivalent shear

rate on a log-log graph, the polymer-solu-tion/oil mixture plots are identical to that

for polymer alone at shear rates greater

than 10 seconds-1. This indicates that a sin-

gle-phase fluid model can be used to calcu-

late the pressure drop of polymer-solu-tion/oil mixtures.When flow rate and pres-

sure drop through a core are plotted, the

pressure drop varies linearly with the nth

power of the flow rate. The relationship of

pressure drop and flow rate exhibits thebehavior of a non-Newtonian power-law

fluid in the porous media. At high flow

rates, the pressure drop increases, deviating

from the power-law model, indicating that

the rheological properties of the polymersolution in porous media are not the same

as in a capillary tube.

IPR CURVES FOR OIL WELLS

The apparent viscosity of polymer solution

varies greatly during the time the solutionis mixed at the surface, injected, and finally

produced with the oil. In the Daqing oil

field, the apparent viscosity decreasesapproximately 30% from the time it ismixed to the time it reaches the injection

wells. Apparent viscosity decreases 60%

when it reaches an observation well 30 m

from the injection well and 70% at a pro-

duction well 106 m away. This indicatesthat the loss of apparent viscosity is large

near injection wells. Severe shear degrada-

tion of large polymer clumps usually occurs

near the bottom of injection wells, decreas-

ing the non-Newtonian behavior of thepolymer solution. When the shear rate is

greater than the critical shear rate, the fluidexhibits viscoelastic behavior in the vicini-

ty of the wellbore and power-law fluid

behavior in the formation. The full-length

paper contains a differential equation forinflow performance. This equation com-

bines equations of continuity, motion, and

state. Boundary conditions are defined.

Calculations. An IPR curve was calculatedwith the derived equation. When calculat-

ed flowing bottomhole pressures were com-pared with measured values, the maximum

absolute error was 230 kPa and the maxi-mum relative error was 5.82%. The close

agreement of measured and predicted val-

ues indicates that variable rheological para-meters must be used to predict IPRs for oil

wells in polymer-flooded reservoirs.

Analysis of IPR curves of oil wells shows

that, when the rheological parameters offluid vary in injection and production

wells, the pressure drop in the formation

increases as the k of the produced fluid

increases and n decreases. Pressure drops

decrease ask decreases and n increases.

Case History and Influence Factors.

Polymer-solution rheological parameters

were measured for an oil well in a poly-

mer-flooded reservoir in the Daqing oil

field with a permeability of 475 md, and aporosity of 0.267. The concentration of

the injected polymer solution was 1000

mg/L, k = 0.0809 Pas, and n = 0.5637.The value of k for the fluid produced from

an oil well in this field was 0.0145 Pasand n was 0.809.

CONCLUSIONS

1. Rheological properties of formationfluids in polymer-flooded reservoirs change

in the direction of flow.2. Calculations of IPR curves for oil wells

in polymer-flooded reservoirs need to

include variations of rheological properties as

the fluids flow through the reservoir.

Please read the full-length paper for

additional detail, illustrations, and ref-

erences. The paper from which the syn-

opsis has been taken has not been peerreviewed.

CALCULATION OF IPR CURVES FOR OIL

WELLS IN POLYMER-FLOOD RESERVOIRS

This article is a synopsis of paper SPE

38936, Calculation of IPR Curves of

Oil Wells for Polymer-Flooding Reser-

voirs, by Yue XiangAn, SPE, Xia

Huifen, Yunxiang Zhang, and Li

Jingyuan, Daqing Petroleum Inst., orig-

inal ly presented at the 1997 SPE

Annual Technical Conference and

Exhibition, San Antonio, Texas, 58October.

8/10/2019 EOR-IOR

11/21

82 JANUARY 1998

E O R / I O R

Field testing has confirmed that a newly

discovered, modified hot-lime process

(MHLP) is a significant improvement overexisting precipitation-softening options. A

Permian Basin produced oilfield water

containing 2,000 ppm hardness, 500 ppm

sulfides, 10,000 total dissolved solids

(TDS), and 200 ppm oil is being convert-ed successfully to steam-generator-quality

feedwater. Alkali consumption and sludgeproduction have been reduced by 50%

compared with the conventional process.In addition, alkali consumption by

entrained CO2 is eliminated. Many hot-

lime softeners (HLSs) currently in service

can be inexpensively converted to this

more efficient process.

The lack of economical water treatmentis one of the most critical obstacles to

achieving a successful steam-injection pro-

ject. Because of strict environmental regula-

tions and lack of available fresh water, pro-

duced oilfield water is typically used assource water for steam generation. For

most steamfloods, quality of the produced

water is fair and sufficient treating can be

accomplished with typical oil-removal andsoftening techniques. In many west Texas

oil fields, however, produced-water quality

is much worse than that currently used for

steam generation. Hardness and sulfide lev-

els are 10 times the average for Californiasteamflood source waters.

Identification of the significant reserves

and economic potential of a thermal project

in a large west Texas oil field led Marathon

Oil Co. to embark on an operational steamfield test in 1995. The test consisted of

installation of a 5,000-BWPD oilfield

water purification facility and three sin-

gle-pass waste-heat steam generators and

drilling of one steam-injection well.

WATER QUALITY

Steam for most oilfield steamflooding is

produced in conventional steam genera-tors. These generators are fired with nat-

ural gas or waste heat and use a single-

pass tube arrangement to produce 80-

quality steam. Saturated steam (including

the 20% liquid phase) is injected into theoil formation with steam-injection wells.

Many impurities in the steam-generatorfeedwater can be tolerated because the

20% liquid phase provides a place forthem to concentrate and still remain solu-

ble. However, according to typical steam-

generator-manufacturer guidelines, all oil,

sulfides, hardness, and suspended solids

must be removed to prevent damage tosteam-generator tubes. Oil in the feedwa-

ter contributes to film formation and cok-

ing in the generator tubes, resulting in

their eventual failure. Sulfides are

believed to be corrosive. Hardness deposi-tion creates steam-generator scaling,

eventually leading to hot spots and tube

failures. Suspended solids must be

removed because they contribute to for-

mation of steam-generator sludge. Typicalsteamflood source waters contain 200 ppm

hardness and essentially no sulfides. The

extremely high amounts of hardness and

sulfides in the source water in this westTexas field made their removal the major

challenge for a successful project.

OVERALL PROCESS FLOW

Produced oilfield water first enters a hydro-

cyclone for rough-cut oil and water separa-tion. Oil content is reduced from 200 to 40ppm. Next, single-media filters are used to

reduce oil content further to 10 to 20 ppm.

Water then enters a packed column, where

nitrogen at 4 scf/gal is used to reduce sul-

fide levels from 500 to< 200 ppm. An oxy-gen scavenger is added to reduce oxygen

content from 20 to approximately 0 ppb.

Water is then pumped to the modified hot-

lime softener (MHLS), where heat, lime,

and caustic are added. Hardness is reducedfrom 2000 to 4 ppm here. Anthracite filters

are used next to remove any suspended cal-cium carbonate (CaCO3) and magnesium

hydroxide [Mg(OH)2] precipitates. Weak

acid cation vessels are then used to polish

hardness to

8/10/2019 EOR-IOR

12/21

JANUARY 1998 83

E O R / I O R

water and collected in the lower cone or

solid/liquid separation chamber. Here, pre-

cipitated material is separated from the soft-

ened water in the form of sludge. Periodic

blowdown is performed to remove thismaterial from the vessel.

MHLP. After additional laboratory and field

testing, a new, more efficient method was

discovered and commercially demonstrated.The process is a variation of the standard

HLP. Fig. 2 shows a schematic of the modi-

fied vessel. Instead of adding heat, lime

and/or caustic simultaneously to precipitatehardness, as in the existing HLP, the MHLP

precipitates hardness in two separate steps.In the first step, cold, hard water enters the

vessel and is sprayed into a steam atmos-

phere and heated to near boiling, just as in

the standard HLP. Soda ash can be added atthis point to supply the necessary bicarbon-

ate ion for hardness precipitation if the ion

is not present in sufficient quantities natu-

rally. However, instead of adding lime or

caustic soda at this point, the MHLP allowsretention time for precipitation reactions

resulting from only the addition of heat totake place. Laboratory and field testing has

shown that 10 to 20 minutes is required for

completion of thermal softening reactions at

vessel operating temperatures. If less reten-tion time is allotted (as with the HLP), ther-

mal reactions will not take place. Instead,

the preference for reactions with lime and

caustic overrides thermal reactions. The

amount of thermal softening that takesplace can be significant. In the field test,

produced oilfield water was softened from

2,000 to < 1,000 ppm in this step, with-

out any chemical addition.The second step of the MHLP consists of

adding lime and/or caustic soda to precipi-tate hardness not removed during thermal

softening in the first step. Chemical reac-

tions that take place are identical to those in

the HLP. Resulting reactions from the addi-

tion of heat and chemicals form CaCO3 andMg(OH)2 precipitates. The amount of pre-

cipitates generated by the MHLP is signifi-cantly less than that of the HLP. Only 1

mole of CaCO3 is created per mole of hard-

ness during Step 1 (thermal softening)compared with 2 moles /mole of hardness

generated in Step 2 (lime softening). This

translates to a significant reduction in the

amount of precipitated solids. Typically

these solids are hauled to a landfill, whichcan prove costly. The MHLP has several

additional benefits.

1. Improved Removal of H2S. H2S is diffi-

cult to remove from water at elevated pH.

Because the typical HLP quickly elevateswater pH by introducing lime early, gaseous

H2S is converted to ionic sulfide, which

cannot be removed. The MHLP does not

elevate pH significantly for several minutes,

so H2S is removed by steam deaerationmuch more effectively.

2. Elimination of Lime Consumption by

CO2. The standard HLP introduces lime

into the vessel in the steam/vapor space

where CO2 can exist. CO2 dissolved in the

produced water will react with added limeand create CaCO3. Therefore, the presence

of CO2 increases alkali demand and

increases the amount of precipitated solids,both of which are unfavorable. The MHLPeliminates these problems because the alka-

li is introduced below the water level in the

vessel. CO2 does not exist in water at HLP

temperatures (it is driven off by Step 1 heat-

ing) and therefore cannot consume lime.

Existing HLS can be inexpensively con-verted to this more efficient process. Most

vessel designs currently in service afford

necessary retention time for thermal reac-

tions to occur in the reaction chamber ofthe vessel. The primary modification

required for conversion is lowering thealkali feed point from the top of the reac-

tion chamber to the top of the reaction

chamber downcomer.

USE OF TWO ALKALIS TO

IMPROVE PERFORMANCE OF THE

PRECIPITATION PROCESS

When lime alone is used, pH must be main-

tained in the very tight pH range of 9.3 to9.6 to achieve acceptable effluent hardnesslevels. If the pH levels fall outside this

range, hardness levels climb quickly. To

overcome this chemistry problem, both

lime and caustic are fed to precipitate

hardness in the HLS. After thermal soft-ening takes place in the upper portion of

the reaction chamber, hydrated lime is

added to boost the pH to near 9.0.

Hardness is reduced to approximately 100ppm. The result is precipitation of large

amounts of hardness with cheap lime. A

relatively small amount of caustic soda isthen added to precipitate the majority of

remaining hardness. Because caustic sodais free of calcium ions, hardness does not

increase when overfed. While caustic

could be used exclusively and accomplish

the same result, the relatively high cost of

the chemical makes this economicallyunattractive for high-hardness waters.

Fig. 1HLS vessel.

Solid/Liquid

SeparationChamber

Fig. 2MHLS vessel.

8/10/2019 EOR-IOR

13/21

84 JANUARY 1998

E O R / I O R

In addition to producing CaCO3 and

Mg(OH)2, sodium carbonate (Na2CO3) or

soda ash is also formed by caustic reactions.

Presence of these excess carbonates allowsthe process to achieve extremely low hard-

ness levels. Effluent hardness levels less

than 4 ppm have been routinely achieved in

the precipitation softener.

TEST PROJECT OPERATING

CHALLENGES

Plugging. The MHLS has been in service

for more than 1 year. After design refine-

ments were complete, operating efficiencyand operability have surpassed expecta-

tions in many respects. However, the

major operational challenge still facing

the vessel is periodic plugging of the reac-

tion chamber and downcomer with pre-cipitated hardness. Field testing estab-

lished that standard HLSs require signifi-cant modification to accommodate the

solids generated from precipitation soft-ening of west Texas produced oilfield

water. With the standard design HLS,

operation would be halted after 2 weeks

because of a plugged reaction-chamber

downcomer. Through a series of improve-ments, periods of run time for the MHLS

extended significantly. Those modifica-

tions include the following.

1. Enlargement of the reaction-chamber

downcomer. Typically, reaction-chamberdowncomers are sized to achieve a down-

ward water velocity of 1 ft/sec. Field expe-

rience indicates that a more appropriate

design criterion for extremely hard waters

is 0.5 ft/sec.2. Modification of the downcomer

splashplate to a witchs hat design.

Standard HLSs use a flat plate at the bottom

of the reaction-chamber downcomer to pre-vent disturbance of the sludge below. Field

testing operations suggested that the stan-

dard splashplate design allowed precipitat-

ed solids to stack, eventually plugging the

downcomer. The current witchs hat design

allows solids to slough off much moreeffectively.

3. Creation of a high-rate cyclonic

motion in the 10-ft-diameter reaction

chamber with tangential nozzles. To keepprecipitated solids in suspension as water

moves down the downcomer, 600 gal/min

of water (from the reaction chamber) is

recirculated.4. Use of a steeper-sloped reaction-cham-

ber cone. Standard HLS design calls for a

45 from horizontal cone in the reaction

chamber, even though the sludge repose

angle is greater than 50. Future vesseldesigns will use a 60 from horizontal cone

in an attempt to alleviate the plugging ten-

dencies further.

Sulfide Removal. Sulfide removal from the

source water was the second significant

technical challenge for project engineers.

Removal of an extremely large amount of

sulfides (500 ppm) in steam-generatorfeedwater is very unusual. Use of conven-tional methods to accomplish this would

adversely affect water-treating costs.

Initially, complete sulfide removal, as

recommended by steam-generator-manu-facturers guidelines, was thought to be

necessary. Experience from other steam-

flood operators, however, indicated that

this was not necessarily true. Steamflood

operators in California have operated foryears with source waters containing 40

ppm sulfides and have reported no signif-

icant problems. With this information, atwo-pronged strategy to the sulfide issue

was adopted. First, an uncommon sulfide-removal technique would be engineered

and installed that would reduce sulfide

levels to less than 40 ppm. Second, field-

testing facilities would be used to deter-mine whether higher sulfide levels were

tolerable. For example, if the upper sul-

fide level limit could be extended to 200

ppm, water-treating costs would be

reduced, thereby improving project eco-nomics. Also, field experience found that

sulfide levels below 200 ppm prevented

interference with hardness reactions in

the MHLS. Sulfides under certain condi-

tions were found to tie up available car-bonate alkalinity needed for the process,

thereby creating permanent hardness in

the water.

After more than 1 year of steam injection,

no significant corrosion is evident in thewater plant or steam system with boiler

feedwater containing 200 ppm sulfides.

Therefore, 200 ppm sulfides will be the

newly established upper sulfide target forfuture water-plant designs. Nitrogen with-

out any sort of pH adjustment will be usedto accomplish this level of reduction.

Conventional Sulfide Removal. Sulfides

exist in both ionic and gaseous forms at pHlevels of typical oilfield waters. To remove

all sulfides, water pH must be lowered to

below 5 to convert ionic sulfide to H2S.

Hydrochloric acid is the agent used typical-ly for pH reduction. H2S can then be

removed effectively with conventional

stripping techniques. The problem with

using conventional methods before a pre-

cipitation process is that added acid signifi-

cantly reduces the amount of alkalinityneeded for hardness precipitation. Lost

alkalinity must therefore be replaced in the

form of Na2CO3. This approach of elimi-

nating alkalinity with acid to remove sul-

fides, then replenishing that lost alkalinitywith an additional chemical, is expensive

and unnecessary.

Uncommon Sulfide Removal Method.

The original sulfide removal techniqueused CO2 as both the stripping gas andagent for pH reduction. CO2 forms carbon-

ic acid, bicarbonate, and carbonate when

dissolved in water. The free hydrogen-ions

released by dissociation of carbonic acid

reduce the pH of the water. Laboratory andfield testing indicate the water pH can be

reduced to less than 5 by dissolved CO2.

Because the reactions are reversible, no

adverse effect on alkalinity occurs.

Nitrogen is not as effective as CO2 becausethe water pH actually increases as a result of

stripping of entrained CO2 with the sul-fides. Blending even small amounts of CO2into the nitrogen stream significantly

improves packed-column performance.Adding only 10% CO2 to the nitrogen strip-

ping gas stream removes an additional 90

ppm of sulfides. Any CO2 dissolved in the

water is effectively removed by heating in

the downstream HLS.

CONCLUSIONS

1. A new process for converting hard oil-

field water to boiler-quality feedwater has

been developed and commercially demon-strated. Alkali consumption and sludge

production have been reduced by 50%,

compared with the standard HLP.

2. The addition of small amounts of caus-

tic soda to the MHLP enhances the stabilityand efficiency of softening in high-hardness

west Texas source waters.

3. CO2 is effective at removing sulfides

from water without adversely affecting

water chemistry in regard to softening.However, piping and vessels exposed to

dissolved CO2 must be protected to pre-

vent corrosion caused by carbonic acid.

4. Sulfide levels up to 200 ppm in steam-generator feedwater have resulted in no sig-nificant corrosion to steam generation and

distribution equipment after 1 year of oper-

ation. At a typical produced-water pH, this

level of reduction can be accomplished

with nitrogen.

Please read the full-length paper for

additional detail, illustrations, and ref-

erences. The paper from which the syn-

opsis has been taken has not been peerreviewed.

8/10/2019 EOR-IOR

14/21

JANUARY 1998 85

E O R / I O R

The Simonette Beaverhill Lake A and B

pools were discovered in September 1993.

Enhanced recovery (EOR) by water-alter-

nating-gas (WAG) miscible flooding was

initiated in May 1995. This short, 20-

month cycle time was achieved through

careful planning, multifunctional team

work, and close cooperation between the

partners in all critical decisions. The chal-

lenges to develop Simonette, including uni-

tizing, well spacing and scheduling, con-

serving gas, and selecting the type of floodwere identified in a development plan cre-

ated before discovery. Keys to economic

success were also identified in the planning

stage. When the discovery was made, a

multidisciplinary development team was

formed. Much effort was spent on under-

standing and meeting the needs of all inter-

ested parties, including partners and regu-

latory agencies.

GEOLOGY

Oil is trapped in the Simonette Beaverhill

Lake A and B pools in the updip culmina-tion of a Swan Hills reef, a part of the

Beaverhill Lake group. The Beaverhill Lake

group comprises the Swan Hills formation,

consisting of reefal carbonates, and the

Waterways formation, composed of basin-

filling marlstones and carbonate mud-

stones. The A pool is separated from the B

pool by a normal fault with approximately

30 m of throw. The reefs were deposited in

a series of eight reef stages that grew in

response to rising sea levels. Development

and management of the reservoir is driven

by the properties and relationships of thesestages. Stage porosities within a well range

from 7 to 12%, and permeabilities can

range from 1 to 400 md.

PLANNING

Exploration Economics. The Simonette

exploration prospect was characterized as

having a geologic risk of one in five and

recoverable oil from 795 to 5500103 m3.

The area ranged from 24 to 40106 m2.

Net pay was expected to be from 7.5 to

12.0 m. The original development plan

called for a waterflood with 65-ha spac-ing, the same as most Beaverhill Lake

reservoirs in Alberta, and a nominal five-

spot pattern. With the estimated

exploratory risk of one in five, the

prospect was considered uneconomical.

The performance of other Beaverhill

Lake reservoirs showed that 80% of the

recovery was obtained from 20% of the

wells. Flood-front mapping of other

waterfloods showed that water injectors

could sweep oil as much as 4.83 km. The

performance of these reservoirs caused

the development plan to be revised to use

130-ha spacing.

Improved Oil Recovery (IOR). Implicit in

the economic evaluation was the need for

EOR. Expected primary recovery factors

were 10 to 20%, which would not support

the well investment and facilities cost.

Waterflooding was considered the mini-

mum depletion mechanism. In addition,

the Alberta Energy and Utilities Board

(AEUB) limits production from a reservoir

by a maximum rate limitation (MRL) that istypically 9000 m3 per month/1106 m3 of

recoverable oil.

Development Plan Goals. A Gantt Chart

for the development plan was presented

to management. If this extremely aggres-

sive timeline could be met and the pool

developed on 130-ha spacing, the explo-

ration economics were very attractive.

The following were major challenges in

the timeline.

Designing and selecting a flood with

only three to four wells out of an ulti-mate pool development of 19 wells.

Obtaining AEUB approval in 3 months

with limited data when 6 to 12 months

is typical.

Unitizing the pool within 2 years, withthe same data limitations.

After management was assured that the

timeline would be met, approval was

given for drilling the discovery well in

June 1993. After discovery, water injec-tion was achieved 6 months ahead of plan

and gas injection began 4 months ahead

of the injection target date.

LABORATORY STUDIES

Reservoir fluid studies and laboratorycorefloods are critical sources of data for

designing and selecting a flood scheme for

a reservoir. Representative samples from

the dominant flow facies were selected for

waterflood tests, relative permeabilitydetermination, and (combined with fluid

studies) miscible flooding potential. A set

of capillary pressure curves, covering a

range of porosities and permeabilities, was

determined. These data were gathered tocreate a permeability transform to popu-

late a reservoir-simulation model. The

composition analysis revealed that solu-

tion gas from Simonette could be used as amiscible solvent for the oil. This finding

was instrumental in the decision to use a

miscible flood.

REGULATORY ENGAGEMENT

The AEUB must approve all recovery

schemes as part of their mandate to con-serve the energy resources of Alberta. AEUB

approval can take up to 1 year if they iden-

tify issues with a proposed scheme thatrequire additional laboratory or simulationstudies. Because approval was required

within 3 months of application with little

data, a meeting was held to explain the

nature of the reservoir, development plans,

and the need for a short cycle time. Thestaff of the AEUB raised several questions.

The objectives of the simulation studies

were set to address these. The AEUB staff

was kept informed of the findings and con-

clusions of various studies. The final appli-cation was submitted in October 1994 with

data from only three wells. The applicationwas updated with information from an

A MODERN EXAMPLE OF SHORT-CYCLE-

TIME DEVELOPMENT

This article is a synopsis of paper SPE

38824, Simonette Beaverhil l Lake

A&B Pools: A Modern Example of

Short-Cycle-Time Development, by

T.J. Moynihan, SPE, Chevron Canada

Resources; J.C. Fryters, Chevron

Petroleum Technology Co.; and P.

Chernik, SPE, Shell Canada Ltd. origi-

nal ly presented at the 1997 SPE

Annual Technical Conference and

Exhibition, San Antonio, Texas, 58October.

8/10/2019 EOR-IOR

15/21

86 JANUARY 1998

E O R / I O R

additional five wells in January 1995, and

approval of miscible flood for the Simonette

A pool was obtained in March 1995.

EARLY PERFORMANCE

MONITORING

As each well was drilled, a temporary bat-

tery was installed at the wellsite. T