You are on page 1of 10

TECHNOLOGY FOCUS

Heavy Oil

Recently, I served as cochairperson of the SPE Forum on “Enhanced Oil Tony Kovscek, SPE, is an associate
Recovery—What’s Next?” The consensus of attendees was that while there is some professor of energy-resources engineering
promising technology, no game-changing enhanced-oil-recovery (EOR) technolo- at Stanford University, where he directs
gies appeared to be on the immediate horizon. In the area of heavy hydrocarbons, the heavy-oil and unconventional
on the other hand, it was expressed that in-situ upgrading of resources such as
heavy oil, tar, and oil shale could be a game changer if more-vigorous work was resources group (SUPRI-A). Kovscek
conducted in the area of upgrading. The dawn for surface upgrading of bitumen holds BS and PhD degrees from the
appears to have occurred. University of Washington and University
Various possibilities for in-situ upgrading were discussed in some detail includ- of California at Berkeley, respectively. He
ing electrical heating to create an upgrader and refinery-in-the-ground for oil shale; is the Stanford University SPE Student
solvents tailored to precipitate the asphaltene fractions of crude oil; and in-situ
Chapter faculty adviser and serves on
combustion, in which the heaviest fractions of the crude are consumed selectively
as fuel for combustion, thereby increasing the API gravity of the oil. A common the Continuing Education Committee,
theme is that in-situ upgrading is a complex process needing much more R&D. the JPT Editorial Committee, and the
The conversion of trillions of barrels of oil equivalent to a lighter, sweeter syn- Western Regional Meeting Program
thetic crude oil is an exhilarating thought, especially given the possibility that such Committee. Kovscek is review chairperson
upgrading can be accomplished in a fashion that is close to being environmentally of the SPEREE (RE) Editorial Review
benign if it is carried out in situ with geological sequestration of any carbon dioxide
created during upgrading. Considering the analogous time frame for implementa- Committee. He received the 2005 SPE
tion of the markedly simpler process of steamdrive for EOR, however, provides a Western North America Region Technical
sobering perspective. From the initial engineering report to commencement of a Achievement Award and the 2006 SPE
commercial project at Mt. Poso, a total of 20 years elapsed. Distinguished Achievement Award for
Papers for this feature span many aspects related to heavy hydrocarbons includ- Petroleum Engineering Faculty.
ing in-situ combustion, steam injection, optimization of thermal-recovery pro-
cesses, chemical-based EOR for heavy oil, as well as artificial lift. Many questions
linger for me regarding in-situ upgrading. Is an extended R&D time frame realistic
today? If a significant portion of the technical workforce is poised to retire within
a decade, where/when will the future workforce be trained for such an enormous
challenge and who will be their mentors? Ultimately, could we extract the energy
in the form of clean-burning hydrogen, or preferably methane, because of the sig-
nificant natural-gas infrastructure already in place? JPT

Heavy Oil additional reading available


at the SPE eLibrary: www.spe.org

SPE 110479 • “World’s First Metal PCP SAGD Field Test Shows Promising
Artificial-Lift Technology for Heavy-Oil Hot Production: Joslyn Field Case” by
Jean-Louis Beauquin, Total, et al.
SPE 107949 • “Optimization of Cyclic Steam Stimulation Under Uncertainty” by
K. Revana, Halliburton, et al.
SPE 106908 • “Applicability of Water-Shutoff Treatment for Horizontal Wells in
Heavy-Oil Reservoirs” by Francesco Verre, Imperial College, London, et al.

84 JPT • MARCH 2008


HEAVY OIL

Optimize Cyclic Steam Stimulation


Though Experimental Design

Peace River Carmon Creek is an ultra- a poorer-quality estuarine zone with Area 6 appears to be the poorest for
heavy-oil lease in northwestern Alberta, an order of magnitude lower vertical-/ CSS application. It is thin, with very
Canada, holding nearly 1.27×109 m3 horizontal-permeability ratio compared high oil viscosity and a very thick BWZ.
of 7°API oil. The Carmon Creek Project to the underlying better-quality del-
targets approximately half of that oil for taic zone. This generalization does not Reference-Case CSS Design
development by cyclic steam stimula- apply to the entire Peace River field. In The reference-case CSS well configu-
tion (CSS). There are plans for a signifi- some parts of the field, the estuarine ration and steaming strategy consists
cant increase in oil production over the zone is of better quality than the deltaic of 500-m horizontal wells drilled on
next 5 years. The purpose of this study zone. Lower portions of the deltaic zone 150-m spacing. There are 16 such wells
was to optimize CSS well configuration typically have higher water saturation, drilled into a standard subsurface “pad”
and steaming strategy. termed “basal water zone” (BWZ). The measuring 1300×1200 m. Steam-injec-
total gross reservoir thickness and the tion rate is 1000 tons/d per well. The
Introduction BWZ thickness vary greatly across the first-cycle steam-injection volume
Various thermal-recovery schemes have field. Bitumen viscosity also is highly (steam slug) is 30 000 tons, and the
been piloted at Peace River. At pres- variable, both areally and vertically, length of the first production cycle is
ent, CSS is used to extract the oil, most throughout the Bluesky reservoir. 247 days. The steam slugs increase
recently with closely spaced multilateral A full-field static reservoir model in subsequent cycles. The production-
horizontal wells drilled from a central was built covering an area of approxi- period lengths after the first cycle are
surface pad. mately 2000 km2 and including more equal to 12.4 days per 1000 tons of
The CSS target is the Bluesky for- than 400 wells. The main objective steam injected in that cycle. The wells
mation, an approximately 30-m-thick was to capture and characterize the are drilled 5 m above the reservoir base
semiconsolidated-sand layer buried regional reservoir differences by use of along their entire horizontal trajectory.
at a depth of approximately 600 m a 250×250-m coarse areal grid with
and characterized by a wide range 0.5-m vertical resolution. CSS-Concept-Selection Workflow
of reservoir properties. These vary- Previous Peace River CSS history-
ing reservoir features are expected to Area Models. The full-field static matching work lead to development
result in different optimum CSS-well model was divided into six distinct of a single, consistent physical model,
configurations and steaming strategies geological areas, and a relatively fine- believed to be applicable to all areas
for each geologically unique portion of grid (25×25 m) sector model was con- of Peace River. On the basis of field
the field. structed for each of these areas. observations of initial geomechanical
Areas 1 and 2, characterized by high reservoir conditions, a steam-induced
Geological Models permeability, good seals, thick pay, fracture is most likely to be initiated
Full-Field Model. The Bluesky reser- high oil concentration, high hot-oil vertically at the heel, transforming to
voir in the Carmon Creek Project area mobility, and either an insignificant or a horizontal fracture when it reaches
generally is classified into two intervals, no BWZ, have the most favorable reser- the top of the reservoir. This physical
voir properties for CSS. model and fracture realization were
This article, written by Assistant Area 3 is poorer than Areas 1 and used in all reservoir models in this
Technology Editor Karen Bybee, con- 2 because of the thicker BWZ and study. A thermal reservoir simulator
tains highlights of paper SPE 109826, somewhat higher oil viscosity (lower was used.
“Peace River Carmon Creek Project— hot-oil mobility). A unique modeling workflow, split
Optimization of Cyclic Steam Stimulation Area 4 can be characterized by thin into the following three phases, was
Through Experimental Design,” by P.F. pay, top-gas presence, and very high oil developed to determine the optimum
Koci, SPE, and J.G. Mohiddin, SPE, viscosity. This area is likely to be very well configuration and steaming strat-
Shell International E&P, prepared for the marginal for CSS application. egy for each geological area.
2007 SPE Annual Technical Conference The estuarine zone in Area 5 is more 1. The initial phase explored the effect
and Exhibition, Anaheim, California, 11– permeable than the deltaic. The oil of grid size and accuracy. Comparison
14 November. The paper has not been viscosity of Area 5 is the highest among between horizontal and vertical CSS
peer reviewed. the six areas of Peace River. wells also was investigated.

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

JPT • MARCH 2008 85


2. The CSS-optimization phase was a Initial screening of a few row-steam- • Simultaneous development of the
parametric study aimed at selecting the ing strategies did not indicate marked estuarine and deltaic zones in most
optimum well-design and -position and differences in performance; hence, the areas except for areas with very thick
steaming-strategy combination for each reference-case assumption of “double- BWZ or top gas.
of the geological areas identified. row overlap” also was not altered.
3. In the third phase, sensitivities Uncertainty Analysis
with respect to reservoir properties Cumulative Discounted Net Oil Once the optimum well configuration
were investigated for each of the geo- (CDNO). The optimization process and steaming strategy for each of the six
logical areas. This exercise also gener- was based primarily on maximizing the geological areas were chosen, the last
ated the uncertainty range in produc- CDNO obtained from the standard sub- phase involved quantifying the effect of
tion forecasts for each area. surface CSS pad. This output parameter key uncertainties and developing deter-
can be considered as an approximation to ministic P15, P50, and P85 realizations.
Initial Modeling net present value before taxes (NPVBT),
Model Size. From the beginning, it was if well costs or facilities costs or number Uncertainties and Ranges. All static
anticipated that many simulation runs of wells are not changing. It was chosen and dynamic parameters were varied
would have to be made. Therefore, it over recovery factor (RF) and cumula- simultaneously to prevent probabilistic
was highly desired to produce the fast- tive oil/steam ratio (COSR) because nei- “double-dipping.” This can occur when
est (and hence smallest) possible model ther of these output measures account the RF is a function of the oil-in-place
without compromising the results. for the early acceleration component. (OIP) volume. The RF-vs.-original-OIP
Various “elements-of-symmetry” mod- relationship is not very strong in typi-
els were carved from the standard-subsur- Experimental Design. To optimize cal black-oil models but can be very
face-pad area. The “quarter-pad,” “eighth- these multiple input parameters simul- strong in thermally enhanced oil-recov-
pad,” “three half-well,” and upscaled taneously with a minimum number of ery projects.
“three half-well” element-of-symmetry simulation runs, experimental-design The ED software was used again to
models provided approximately the same (ED) software was used to design the provide ED for analyzing the uncertain-
results, but the latter model ran signif- input-parameter combinations to be sim- ties. Linear ED models with parameter
icantly faster. Therefore, the optimum ulated. Second-order ED models with interactions were used. The effect of
model chosen was the three half-well input-parameter interactions were select- bitumen-viscosity uncertainty over-
element-of-symmetry model, precisely ed because nonlinear relationships and whelms that of all remaining param-
3/32 of the standard subsurface pad. interactions among the various parame- eters. This observation is consistent
ters were expected. The input parameters across all six areas, indicating the need
Fracture-Grid Dimensions. The size and the simulated CDNO responses were for additional viscosity-data collection
of grid in which a fracture is located entered into the ED software, and the to narrow this wide uncertainty range.
can have a large effect on simulation standard least-squares multiple-regres-
results. Extensive fracture-grid sensi- sion analysis was applied to obtain the Probabilistic Forecasts. The uncer-
tivity was performed for fracture-grid CDNO-response-surface coefficients. tainty-analysis CDNO-response-surface
thickness, ranging from 0.1 to 15 m for coefficients were used in Monte Carlo
a vertical fracture and from 0.1 to 1.5 m Well-Spacing Optimization. Well cost simulations to obtain the full range
for a horizontal fracture. The optimum and oil price have to be included if well- of possible realizations. The resulting
solution appeared to be a combination spacing optimization is desired. This individual realizations were sorted in
of a 5-m vertical-fracture-grid thick- was done in an approximate manner ascending CDNO order and plotted.
ness and 1.5-m horizontal-fracture-grid by calculating CDNO for a standard- The scatter in RF and COSR plots
thickness. This selection made run subsurface-pad area, multiplying it by implies that a number of individual
times manageable. an oil price, and subtracting the cost Monte Carlo runs, with radically differ-
of wells located in the same subsurface ent input-parameter combinations, RFs,
CSS-Design Optimization area to obtain approximate NPVBT. and COSRs, can yield the same CDNO.
CSS-Design Parameters. Following In addition, because well spacing The process of picking deterministic
the exclusion of vertical wells, two well implies partial acceleration of oil produc- P15, P50, and P85 cases (on the basis
types were considered, horizontal and tion, a range of discount rates was used to of CDNO probability distribution func-
the upward-trending deviated well. The reach a proper decision on optimum well tions) for each area consisted of choos-
upward-trending deviated well was de- spacing. This entire process was repeated ing those Monte-Carlo-run configura-
signed and piloted to overcome poor ver- for each of the six geological areas. tions that had both RF and COSR close
tical permeability inhibiting steam from to a 20-run average near the P15, P50,
rising toward the top of the reservoir. Optimum CSS Design. An extrapola- and P85 CDNO values.
The five CSS-design parameters tion of the various simulation results Once the deterministic P15, P50, and
considered in the experimental design indicated that the optimum develop- P85 cases were selected for each area,
were: (1) steam-slug-size factor, which ment concept was actual simulations were performed and
was the volume of steam injected dur- • Horizontal wells in all areas. the results plotted in the form of RF vs.
ing the first cycle, (2) pore volume (PV) • Well spacing of 75 m per zone in all PV of steam injected. Good agreement
injected, (3) horizontal-well elevation areas, except for the poorest area with is observed between the simulation
above base reservoir, (4) well spacing, thin pay and thick BWZ where 100-m results and existing empirical forecasts
and (5) production-cycle length. spacing appears to be better. for the reference-case design. JPT

86 JPT • MARCH 2008


087_BHCentri_jpt.qxd 2/13/08 14:08 Page 21
HEAVY OIL

Combustion-Tube Experiments Show Heavy-Oil-Production


Increase With Addition of Catalyst

A catalyst was evaluated to aid heavy-


oil combustion. The underlying objec-
tive is to increase the oil mobility inside
the reservoir by use of the catalyst
during combustion. The catalyst was
evaluated in a combustion tube with
12.5°API heavy oil from the Gulf of
Mexico (GOM). Use of this organo-
metallic catalyst at low concentrations
potentially can upgrade oil properties
in situ, avoiding the cost of surface
facilities to achieve the same goal.

Introduction
Reservoir exploitation is moving toward
large reserves of heavy crude oil.
Producing, transporting, and marketing
this heavy oil presents many problems,
including the inability of most refiner-
ies to accept heavy crude oils. Heavy
crude oils could be more acceptable if
they can be upgraded before being sent
to refineries. In-situ combustion is one
technique used to reduce oil viscosity.
When combustion is combined with
the use of a catalyst, the process is called Fig. 1—Cumulative oil production in combustion-tube experiments.
in-situ catalytic upgrading.
Comprehensive studies of enhancing
the combustion process by addition of crude oils there is a favorable modifi- to determine if this organometallic
metallic salts have shown that for some cation of the reaction kinetics. Salts, catalyst at low concentrations is able to
such as tin chloride or ferric nitrate, improve heavy-oil recovery.
This article, written by Assistant Tech- promote combustion of light oils,
nology Editor Karen Bybee, contains producing more-uniform combustion Experimental Method
highlights of paper SPE 107946, that occurs at a higher temperature. The experimental setup comprised six
“Increase Heavy-Oil Production in However, in another study, it was not main parts: fluid-injection system, com-
Combustion-Tube Experiments Through possible to obtain sustained combus- bustion tube, fluid-production system,
the Use of Catalyst,” by M.A. Ramirez- tion of light oil without the addition gas chromatograph, wet-test-meter sys-
Garnica, Instituto Mexicano del of metallic salts. The full-length paper tem, and a data-recording system.
Petróleo; D.D. Mamora, SPE, Texas describes a combustion experiment
A&M U.; H.R. Nares and P. Schacht- using a combination of organometallic Fluid-Injection System. The fluid-
Hernández, Instituto Mexicano del catalyst derived from either acetylace- injection system consists of two parts,
Petróleo; A.A. Mohammad, Texas tonate or alkylhexanoate compounds nitrogen injection and air injection.
A&M U.; and M.C. Cabrera-Reyes, and an air-injection system in a com- Both parts are independent (through
Instituto Mexicano del Petróleo, pre- bustion tube. The hydrocarbon used 1/4-in. tubing) and are opened or closed
pared for the 2007 SPE Latin American was a 12.5°API heavy crude oil from to the system with valves in the control
and Caribbean Petroleum Engineering the GOM. The catalyst was mixed panel. The injected-nitrogen or -air rate
Conference, Buenos Aires, 15–18 April. with the heavy oil before the experi- is controlled by a mass-flow controller,
The paper has not been peer reviewed. ment. The objective of this study was installed before the injection pressure

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

88 JPT • MARCH 2008


transducer. The 1/4-in. tubing line is
connected with a fitting to a 1/8-in.
tubing line, which is the gas inlet to the
combustion tube.
Nitrogen Injection. Nitrogen is used
to flush the system before any com-
bustion run. Nitrogen also is used to
pressurize the combustion tube. At the
end of the combustion run, nitrogen is
injected into the system to flush and
cool the tube.
Air Injection. Air refers to the nor-
mal mixture of oxygen and nitrogen
in the atmosphere. Air is injected at
a constant 3-L/min rate throughout
the combustion run. A cylinder con-
taining air is connected to the injec-
tion system. When the temperature at
the clean sand/mixture pack interface
reaches approximately 570°F, air is
allowed to flow into the combustion
tube to begin ignition and to maintain Fig. 2—Produced-oil volume vs. time.
combustion. Air injection will continue
until the front reaches the bottom of
the combustion tube (no more liquids to display or register its signal to the trations of 88.87, 5.38, and 5.75%,
are produced). At this time, injection is data logger and to the control panel and respectively. For the experiment with
switched to nitrogen. to the PC monitor. catalyst, the catalyst was mixed homo-
geneously with the heavy crude oil
Combustion Tube. The combustion Fluid-Production System. A back- before the experiment.
tube is a stainless-steel cylinder with a pressure regulator maintains the outlet Approximately 200 g of mixture
3-in. external diameter and a 401/8-in. pressure of the combustion tube at a was introduced into the tube once the
length. Sharp-edged flanges seal the constant predetermined level during combustion tube was fastened safely
ends of the cell to copper gaskets. the experiment. The liquids leaving the in a vertical position. A heavy metal
The tube has an assembly that allows combustion tube pass through a two- plunger that passed through the ther-
the introduction of two thermowells. stage separation where they are col- mowells was used to pack the sample
Another tube was soldered off-center lected at the production outlet. Gases into the tube. The process of adding
on the top flange to allow air injection pass through a condenser kept at low sample and packing was repeated until
into the combustion tube through a temperature to recover any volume the tube was filled to approximately
reduction fitting to an inlet. There is of liquid in this stream. Gases flow- 4 in. from the top. Approximately
tubing soldered to the bottom flange ing toward the gas chromatograph are 5 mL of linseed oil was placed on the
of the combustion tube to allow the scrubbed of acid before entering the top of the sample to begin the igni-
collection of fluids in the produc- next system. tion. The combustion tube then was
tion system. filled to the top with clean sand. The
The combustion tube is placed inside Gas Chromatograph and Wet-Test- mixture remaining in the large mix-
a vacuum jacket. A connection installed Meter System. A small fraction of ing bowl was weighed to determine
at the top flange of the jacket provides produced gas is injected into a gas the amount of mixture placed in the
electric current to the resistance igniter. chromatograph where the gas is ana- combustion tube.
The vacuum jacket is isolated from the lyzed for carbon dioxide (CO2), oxygen The top flange of the combustion
combustion cell with Teflon ferrules (O2), nitrogen, and carbon monoxide tube was installed, and the flange bolts
installed in both flange ends. The exte- (CO) every 5 minutes. A wet-test meter were fastened. Nitrogen was intro-
rior of the vacuum jacket is an alumi- installed before the gas chromatograph duced at the injection inlet and with
num cover with aluminum end caps. allows measurement of the produced the outlet of the combustion tube
The center of the jacket is connected combustion gases, which is recorded plugged, the cell was pressure tested
to a swivel that allows jacket rotation in a PC. for leaks at 400 psi for 2 hours. Once
from horizontal to vertical position. the pressure test was performed suc-
One set of eight fixed J-type thermo- Experimental Procedure. Dolomitized cessfully, the outlet plug of the com-
couples runs through the thermowell carbonate previously triturated at a bustion cell was slowly opened and
end, and a set of six movable J-type particle size of 600 to 850 μm was the pressure in the tube decreased
thermocouples runs through the other used as a porous medium. A mixture to atmospheric.
end. All thermocouples used were of this triturated rock, heavy crude The injection assembly was removed,
0.002 in. thick. Each one of the ther- oil, and distilled water was prepared and an electric igniter was placed and
mocouples is connected to a terminal in a large mixing bowl with concen- tightened at the exterior of the com-

JPT • MARCH 2008 89


bustion tube at the depth where the occurred when oil production ceased;
SPE SERVICE DIRECTORY linseed oil was placed. The tube then in other words, the porous medium
was placed inside the vacuum jacket. was burned to the bottom flange of
The electric igniter was connected to the combustion tube.
SPE Online www.spe.org the ignition terminals of the top flange
Awards Program of the vacuum jacket. The injection Experimental Results
Contact: Tom Whipple
Phone: 1.972.952.9452 • Email: twhipple@spe.org assembly was replaced in its position, Combustion Run Without Catalyst.
Regional/Sectional Awards Program and the fixed and movable thermo- The combustion gas composition
Contact: Brian Wiggins couple sheaths were inserted in their during this run was variable, which
Phone: 1.972.952.9451 • Email: bwiggins@spe.org respective thermowells. indicated that the combustion was
Books, Papers, Software Sales The vacuum jacket was placed in a not very stable. During the run the
Phone: 1.800.456.6863 or 1.972.952.9393
Email: books@spe.org vertical position and the outlet of the concentrations of CO2, O2, and CO
Conference Programs combustion tube fastened to the pro- were determined with the gas chro-
Contact: Wayne Spence duction section. The vacuum jacket matograph. The average combustion
Phone: 1.972.952.9449 • Email: wspence@spe.org
was tested for 1 hour. The injection temperature was 716°F, and the aver-
Distinguished Lecturer Program
Contact: Donna Neukum
line was connected to the assembly, age combustion-front velocity was
Phone: 1.972.952.9454 • Email: dneukum@spe.org and the vacuum-jacket heater was set 0.2684 ft/hr. The initial oil produc-
Dues, Membership Information, Address Changes, to approximately 140°F and left for tion occurred at 7.91 hours. Fig. 1
and Copyright Permission 8 hours to allow the temperature of shows the cumulative volumes of oil
Phone: 1.800.456.6863 or 1.972.952.9393
Email: service@spe.org
the sand mix to stabilize. produced, with an initial water pro-
Exhibit Space Sales
Before beginning the experimental duction occurring at 1.0 hour. Fig. 1
Contact: Joan Payne run, the mass-flow controller and the shows an oil recovery of 77.67 % of
Phone: 1.972.952.9356 • Email: jpayne@spe.org gas chromatograph were calibrated, original oil in the tube.
Contact: Jim Klingele
Phone: 1.713.779.9595, ext. 612 • Email: jklingele@spe.org
the bottom of the movable thermocou-
Insurance, Credit Card Programs
ple sheath was raised to the linseed-oil Combustion Run With Catalyst.
Contact: Alyssa Moreno depth, and the package was pressur- Concentrations of CO2, O2, and CO
Phone: 1.972.952.9439 • Email: amoreno@spe.org ized to 300 psi with nitrogen. Electric were determined by use of the gas
JPT Employment/Professional Services current was introduced gradually into chromatograph. Average combustion
Contact: Mary Jane Touchstone
Phone: 1.972.952.9351 • Email: mtouchstone@spe.org the igniter by use of a variable-power temperature was similar to the case
JPT/JPT Online Advertising Sales transformer. Approximately 90 min- without catalyst, the only difference
Contact: Jim Klingele utes later, the temperature in the being that at the beginning, the com-
Phone: 1.713.779.9595, ext. 612 • Email: jklingele@spe.org combustion tube at the igniter level bustion temperature is higher than in
Journal of Petroleum Technology reached approximately 570°F and air the first run, close to 932°F.
Contact: John Donnelly
Phone: 1.713.779.9595, ext. 616 • Email: jdonnelly@spe.org injection was initiated at 3 L/min. Initial water production occurred
Meeting Registration A backpressure regulator was adjust- at 0.9833 hour, with an oil recovery
Phone: 1.800.456.6863 or 1.972.952.9393 ed to maintain a production pressure of 86.17% of the original oil. Because
Email: registration@spe.org of 300 psi. The movable thermocouple of the catalyst effect, not only does
Peer Review reading in the instrument panel was oil production begin sooner than in
Contact: Stacie Hughes
Phone: 1.972.952.9343 • Email: shughes@spe.org observed to increase rapidly to approx- the experiment without catalyst, but
Periodicals, Book Publication imately 970°F, a clear indication that the amount of oil produced with the
Contact: Glenda Smith ignition occurred inside the combus- use of catalyst is larger. Fig. 2 shows
Phone: 1.972.952.9470 • Email: gsmith@spe.org tion tube. Combustion gas composi- produced-oil volume vs. time.
Professional Development Services tion was measured every 5 minutes.
Contact: Tom Whipple
Phone: 1.972.952.9452 Temperature profiles were recorded Conclusions
Email: twhipple@spe.org approximately every 6 in., and pro- Use of a low concentration of cata-
Section Service duction liquids were measured every lyst in the heavy oil could make this
Contact: Brian Wiggins
Phone: 1.972.952.9451 • Email: sections@spe.org
15 to 20 minutes. Accurate readings process commercially attractive. This
SPE Website
of temperature profiles were taken catalyst could reduce the amount of
Contact: Doug Lauver with the set of six movable thermo- air injected in a traditional air-injec-
Phone: 1.972.952.9519 • Email: dlauver@spe.org couples, spaced approximately 0.2 in. tion process, increase oil production,
Subscriptions from each other, which allowed the and reduce the time at which the oil
Phone: 1.800.456.6863 or 1.972.952.9393
Email: service@spe.org
recording of six entries just behind production begins, which could rep-
Technical Interest Groups
and ahead of the combustion front. resent faster revenue. Summarizing,
Contact: Shasta Stephenson These entries were made by pressing the experiments in combustion tubes
Phone: 1.972.952.9514 • Email: sstephenson@spe.org an assigned key on the PC. with catalysts have shown the follow-
Young Member Programs Initial water and oil production ing important effects:
Contact: Melissa Schultea
Phone: +44.207.299.3300
varied, depending on the initial com- • Production acceleration
Email: mschultea@spe.org position of the mixture, with and • Initial temperature of combustion
without catalyst. Liquids were col- relatively sustained
lected in graduated sample bottles • Production increases
that were capped for subsequent anal- • Higher combustion efficiency
ysis. The end of the combustion run • Faster combustion front JPT

90 JPT • MARCH 2008


HEAVY OIL

Enhanced Heavy-Oil Recovery by Alkali/Surfactant Flooding

This study presents the results of labo- 1


ratory core studies investigating the
recovery mechanisms of alkali/surfac-
tant (A/S) flooding in heavy-oil reser- Equilibrium IFT, mN/m
voirs. Specifically, mixtures of water 0 ppm NaCl
and A/S systems have been injected
into cores containing heavy oil. Salinity 20 000 ppm NaCl
is varied to generate oil-in-water (O/
W) vs. water-in-oil (W/O) emulsion 0.1
systems, and the effects of generating
different emulsions were compared.
It was demonstrated that in heavy-oil
systems, emulsion formation was nec-
essary to produce the heavy oil.

Introduction 0.01
Several countries, including Canada and 0 5 000 10 000 15 000 20 000
Venezuela, contain massive resources
of unconventional heavy oil and bitu- Alkali Content, ppm
men. These oil sands are character-
ized as unconsolidated, high-porosity, Fig. 1—Oil/water IFT with varying alkali content.
and high-permeability reservoirs. The
single biggest obstacle to successful oil to the production wells. However, that in actual reservoir applications, sev-
recovery from the oil sands is the high mobility-ratio concerns dominate dis- eral processes are working together to
oil viscosity. Heavy-oil reservoirs are a placement of viscous oil, and most improve oil recovery. This work consid-
special subset of the oil sands, whereby enhanced-oil-recovery processes focus ers the possible mechanisms that could
the oil viscosity at reservoir tempera- on reducing oil viscosity or improving be responsible for enhanced recovery of
ture and pressure varies on the order the mobility ratio. Unfortunately, many high-viscosity heavy oil.
of 50 to 50 000 mPa·s. While this oil of the heavy-oil reservoirs in Canada
is very viscous, it does have limited are relatively small and thin, making Materials and Experiments
mobility at reservoir conditions. As them poor candidates for expensive The heavy oil used in these experi-
much as 20% of the oil may be recov- thermal processes. This work investi- ments was from a field in Saskatchewan,
ered by solution-gas drive, but in many gates the potential of A/S flooding for Alberta, currently in primary produc-
cases, the recovery is much lower. enhanced heavy-oil recovery. tion. At ambient temperature (23°C),
To recover additional heavy oil, a fluid Surfactants are a special class of mol- the oil has a viscosity of approxi-
usually must be injected to displace ecule that is both hydrophobic and mately 11 000 mPa·s and a density of
hydrophilic; thus, the most stable con- 982 kg/m3. The heavy oil used in some
This article, written by Technology figuration for these molecules is at the of the core floods was more viscous—
Editor Dennis Denney, contains high- oil/water interface. In surfactant flood- approximately 15 000 mPa·s. This oil
lights of paper SPE 110738, “Enhanced ing, these molecules generally are inject- was from the same field; however,
Heavy-Oil Recovery by Alkali/Surfactant ed along with water to reduce the oil/ differences in handling or possibly geo-
Flooding,” by J. Bryan, SPE, and A. water interfacial tension (IFT), which logical heterogeneities were responsible
Kantzas, SPE, University of Calgary reduces capillary forces that may trap oil for the different measured viscosity
and TIPM Laboratory, prepared for the in rock pores. Alkali flooding is a special values. The aqueous phase was either
2007 SPE Annual Technical Conference subset of chemical flooding whereby the brine containing 20 000 ppm NaCl or
and Exhibition, Anaheim, California, surfactant is generated in situ through deionized (DI) water.
11–14 November. The paper has not the reaction between the injected alkali The preformed surfactant was a com-
been peer reviewed. and the acidic crude oil. It is likely mercial anionic surfactant. This surfac-

For a limited time, the full-length paper is available free to SPE members at www.spe.org/jpt.

JPT • MARCH 2008 91


tant is a sodium alkane sulfonate and 0.3
has been used in heavy-oil studies in the

Total Oil RF/PV Injected


past. The alkali used was sodium carbon- 0.25
ate, Na2CO3, which was chosen over
other alkali agents because of its buff- 0.2
ering capabilities and because it is less
sensitive to variations in the water chem- 0.15
istry than is NaOH. The alkali reacts
with organic acids in the oil and should 0.1
generate monovalent anionic surfactants.
In this manner, both the preformed and 0.05
in-situ surfactants are anionic.
The results of the experiments were 0
divided into findings from the bulk- 0.01 0.1 1 10 100
liquid study and from corefloods. The
data collected from the corefloods were Minimum Effective-Mobility Ratio
produced-fluid volumes and pressure.
Therefore, the actual recovery mecha- Secondary A/S floods Entrainment
nisms occurring at the pore scale can Secondary brine A/S Tertiary brine A/S
only be inferred on the basis of an
understanding of how the systems
behave in the absence of porous media. Fig. 2—A/S flooding, post-waterflood recovery efficiency. RF=recovery
factor, PV=pore volume.
Bulk-Liquid Study
The energy required for emulsification 5000-ppm alkali, and additional alkali of the wetting nature of the glass. The
is proportional to the IFT between the increases the IFT. reason that solids wettability is impor-
two fluids. For dissimilar fluids, such Other researchers have noted the tant is that, in the presence of solids,
as oil and water, the required energy salinity effect observed in Fig. 1. Salt the fluid that wets the solid will be the
can be significant. Shear or agitation is tends to shift the salinity minimum to continuous phase more easily in an
required to overcome the Gibb’s free lower alkali concentrations, and the emulsion system.
energy of formation and, more impor- presence of additional alkali then acts Previous work with thermal heavy-
tantly, the Laplace equation of capillary like increased salinity. Therefore, if oil operations and injection into heavy-
pressure in a droplet. Therefore, two injecting brine, the IFT in the system oil cores at very high rates has provided
requirements must be met to form an seems less stable than for DI water. evidence for the in-situ formation and
O/W or W/O emulsion: The oil/water However, in either case, it can be seen flow of W/O emulsions. Because of the
IFT must be reduced, and shear must that the addition of alkali and surfac- high viscosity of the oil phase, these
be added to the system. The addition of tant to the system reduced the IFT emulsion types will be much more
alkali and surfactant is used to reduce between oil and water by two to three common than O/W emulsions. For all
the oil/water IFT such that emulsifi- orders of magnitude. of these studies, however, the oil/water
cation can occur with a minimum of IFT was high, and solids were deemed
applied shear. Extension of Bulk-Liquid Results to to have become oil-wet. In the case of
Fig. 1 shows the measured oil/water Porous Media. One of the major differ- A/S injection, where interfacial forces
IFT values as a function of alkali con- ences between bulk-liquid experiments are much smaller and the solids wet-
tent, with the surfactant fixed at 0.1 wt% and the behavior of the fluids in porous tability is less clear, it becomes very
(1000 ppm). Because of the similar oil media is that the fluids are in much difficult to determine which emulsions
and water densities, when IFT is higher smaller confined spaces while in the will form and how the solids will be
than 1 mN/m, the apparatus cannot reservoir. Therefore, although at low wetted by oil and water.
measure the IFT, even at its maximum IFT the solid appears to be neutrally wet From the bulk-liquid study, it can
rotational speed. Considering the DI (i.e., fluids contact the solids indepen- be inferred that at low IFT, the dis-
water data, in the absence of any alkali, dent of the wettability), this may not be tribution of fluids on solid surfaces is
the surfactant alone cannot reduce the the case in porous media. Additionally, unclear and cannot be explained easily
IFT to measurable values. Even with although the fluids appear to be neu- by contact-angle/IFT measurements. It
0.1 % alkali (1000 ppm) added, the IFT trally wet for both the DI/A/S and brine/ appears that for this chemical system,
is too high to be measured. By adding A/S solutions, the oil adheres much in the absence of salt, O/W emulsions
0.2% alkali to the aqueous phase, the more strongly to the glass in the pres- can form and be stabilized by electro-
IFT values plateau at approximately ence of salt; thus, the glass becomes less static repulsion between droplets. In
0.02 mN/m for the entire range of alkali hydrophilic even if this is not expressed this state, glass (and sand) appears to
contents tested. For the brine system, by the measured contact angles. Other be at least partially wetted by water.
even in the absence of alkali, it appears researchers have reported this effect In the presence of salt in the aqueous
that the surfactant alone can reduce IFT in higher-IFT systems (>1 mN/m). phase, only W/O emulsions form and
to measurable values. In the presence Therefore, phase-behavior experiments the glass becomes much more hydro-
of salt, the IFT has a minimum value at still enable a qualitative understanding phobic (i.e., oil-wet). Although it is not

92 JPT • MARCH 2008


certain if these relationships will hold these emulsions being generated in the this droplet through a pore constriction
in small rock pores, these assumptions bulk-liquid tests. However, it is certain is small. This state is that of a primary
were used to interpret the pressure and that the chemical caused some change A/S flood after breakthrough of the A/S
production data from the corefloods. in the system because pressure built up solution. Alternatively, if the emulsified
in the core, accompanied by a decrease droplet is contacted by a low-IFT fluid
A/S-Recovery Potential in the produced-water cut. The pres- from one direction but by a high-IFT
Through Coreflooding sure and water-cut response are key fluid from the opposite direction, as is
In bulk-liquid experiments, the systems to understanding what is happening: the case for A/S flooding after water-
were subjected to significant shear dur- Emulsion droplets are forming because flooding, the Jamin pressure drop is
ing agitation. The coreflooding experi- of the low IFT and flow shear, but significantly higher. Therefore, trapping
ments were designed to test if emulsions these droplets are not entrained in the of emulsion droplets will be more pro-
could form as a result of the shear forces flowing aqueous phase. Rather, they nounced, leading to improved sweep.
of flow through rock pores. At high tem- must be plugging off the water chan- For waterflooding and primary DI/A/S
peratures, emulsions can form in porous nels, which in turn leads to improved flooding, the most effective oil-recovery
media as a result of water condensing sweep efficiency and a reduced effec- period occurs before breakthrough of
into very small droplets. Likewise, at tive-mobility ratio. the water or A/S solution. Additional
flow under very high shear (i.e., 25- In systems with high shear, O/W pressure buildup in the system leads
to 250-m/d velocities), emulsification emulsions were produced under DI/A/S to improved flow rates on the basis of
can occur. At lower-shear conditions flooding. In these systems, if flooded Darcy’s law, which explains the weak
and ambient temperature, however, with brine/A/S solution after DI/A/S trend showing that at higher dimension-
researchers have not found evidence flooding, the pressure in the core built less shear rate, the recovery efficiency
of significant emulsion generation and up, but now the produced fluid appeared is slightly higher for these systems. For
flow when viscous oil and water mix. to be a W/O emulsion. Therefore, secondary A/S flooding, however, the
The key question was, under conditions although in many cases emulsions were trend is opposite. The best recovery
of low IFT because of the presence of not observed in the produced fluids, is achieved under conditions of low-
A/S in the water, could emulsions form the bulk-liquid experiments were still est dimensionless shear rate, which
at the levels of shear from flow at nor- useful to determine which type of emul- indicates that entrapment is the main
mal reservoir rates. Therefore, a suite of sion should be forming in the porous recovery mechanism for these DI/A/S
waterfloods and primary and second- medium. It is possible that, even under systems. At lower shear rate, emulsified
ary A/S floods was carried out and is DI/A/S conditions, multiple emulsion oil droplets are formed but the shear is
detailed in the full-length paper. types may be forming in the core, not high enough to force these droplets
For a heavy-oil waterflood at a con- but the results from low-permeability through the pore throats. At elevated
stant injection rate, at early times, water experiments show that the bulk-liquid shear, the trapping of the oil droplets is
is injected into the core. However, the emulsion systems can be used at least reduced, and emulsification and entrain-
viscous oil cannot flow out at the same as an indication of what should be hap- ment are the extreme forms of these
rate; thus, pressure builds up in the pening in reservoir applications. high-shear conditions.
core. During this time of high pressure Fig. 2 shows the total recovery effi- Comparing the recovery efficiency
gradient, the oil recovery is high. After ciency with respect to the minimum for secondary DI/A/S floods and sec-
water breaks through, the pressure in value of effective-mobility ratio for the ondary brine/A/S floods, both as a func-
the system declines sharply because secondary DI/ and brine/A/S floods, the tion of dimensionless shear rate, both
injected water now can flow along entrainment secondary DI/A/S floods, systems follow the same trend, which
paths of low resistance, and the oil- and the tertiary brine/A/S floods. In all indicates that the recovery mechanism
recovery rate drops accordingly. Once cases, the total recovery efficiency is is the same. Whether emulsified oil or
water breaks through, the remaining oil directly correlated to the minimum value emulsified water droplets cause trap-
is produced at very high water cuts. of effective-mobility ratio, showing that ping, the recovery mechanism is that
Comparing the pressure response for improved recovery is the result of an of plugging the pore throats, leading to
the waterflood with that of a primary improved mobility ratio of the system. improved sweep efficiency and reduced
brine/A/S flood, when A/S was inject- It can be said that waterflooding at effective-mobility ratios.
ed with water, the pressure maximum the same rate is not affected by the One last result of recovery by entrap-
occurred at later times, indicating a salinity of the injected fluid. However, ment of droplets is the indication that
delay in the water breakthrough. Even when comparing waterflooding with emulsion droplets are on the same scale
after water breakthrough, pressure primary A/S flooding, the results differ as pore sizes. In this situation, it is not
declined more gradually in the chemical by a larger amount. This large differ- correct to think of the production mech-
flood, indicating a more stable displace- ence indicates that primary A/S floods anism being the flow of low-viscosity
ment front. Because of the more stable are more effective than waterfloods for O/W emulsions. This concept applies
displacement during primary A/S flood- both DI/A/S and brine/A/S systems. only to a situation in which the emul-
ing, the recovery efficiency appeared to Secondary A/S floods are much more sion droplets are much smaller than the
be higher than for waterflooding. efficient than primary A/S floods. This pore sizes, meaning that the emulsion
can be understood through a simple can flow as a pseudosingle-phase fluid.
Coreflooding Results. One important Jamin pressure-drop calculation. If a This type of flow was not observed in
point is that no O/W emulsion was low-IFT fluid surrounds an emulsion the corefloods that showed significant
observed in the produced fluids, despite droplet then the pressure drop to force improvement in recovery. JPT

JPT • MARCH 2008 93

You might also like