CO2 EOR in practice

Carbon dioxide flooding is the most widely applied light oil EOR method in North America, with a history that goes back to the SACROC unit in west Texas, where CO2 injection began in 1972. Its attractions are clear from Lesson 6: at moderate pressure CO2 develops miscibility with most light oils, it swells the oil and cuts its viscosity, and it is far cheaper per reservoir volume than hydrocarbon solvent. Its difficulties are just as clear: it is very mobile, it must be bought and transported, much of it comes back with the produced oil and must be recycled, and it is corrosive in the presence of water. This lesson is about how operators deal with each of these in practice, and about the storage of CO2 that accompanies every CO2 flood.

Carbon dioxide as an injectant

CO2 has a critical temperature of 31.1 °C (88 °F) and a critical pressure of 7.38 MPa (1,071 psia). At the temperatures and pressures of most oil reservoirs it is a supercritical fluid with a density of 500 to 800 kg/m³, liquid-like enough to extract intermediate hydrocarbons efficiently, but with a viscosity of only 0.04 to 0.08 mPa·s. It is far more soluble in oil than methane is. At 14,000 kPa (2,000 psia) and 60 °C (140 °F), a light oil can dissolve enough CO2 to swell by 20 to 40 percent and to drop its viscosity by half or more, even when miscibility is not achieved. CO2 also dissolves in water, forming carbonic acid, which corrodes carbon steel and can dissolve carbonate minerals near injectors.

Mobility control: water alternating gas

Continuous CO2 injection gives early breakthrough and poor sweep, for the reasons covered in Lesson 3. Almost all CO2 floods therefore use water alternating gas (WAG) injection: slugs of CO2 alternate with slugs of water. The water lowers the total mobility of the injected fluids, because when both phases flow each has a reduced relative permeability, and it moves preferentially into the lower part of the section, offsetting gravity override. A typical design injects a total CO2 volume of 0.4 to 1.0 hydrocarbon pore volumes (HCPV) over the life of the flood, in slugs of 1 to 3 percent HCPV, with a WAG ratio of water to CO2 reservoir volumes near 1:1. Many projects begin with a larger CO2 slug to build an oil bank and then taper: the WAG ratio is raised over time to save CO2 once the main oil bank has been produced.

WAG has costs as well as benefits. Water injected after CO2 can shield oil from contact with the gas in water-wet rock, which reduces the microscopic efficiency. Injectivity often falls when water follows CO2, because three-phase relative permeabilities are lower than two-phase ones, and the injection pressure limit can then cap the injection rate. The best WAG ratio and cycle length are found by compositional simulation and refined in the field by watching pattern response. Foams, made by injecting surfactant with the CO2, and conformance gels in thief zones are used where WAG alone does not control mobility.

CO2 supply

A CO2 flood needs large volumes of affordable CO2, delivered by pipeline at high pressure. In the United States the industry grew on natural CO2 from geological sources such as the McElmo Dome and Doe Canyon fields in Colorado, Sheep Mountain in Colorado and Bravo Dome in New Mexico, which feed the Permian Basin through pipelines hundreds of kilometres long, and Jackson Dome in Mississippi, which supplies the Gulf Coast. Anthropogenic CO2 from natural gas processing, ethanol and fertilizer plants, coal gasification and power plants has become increasingly important as capture technology and tax credits have improved its economics.

Canada's CO2 floods rely on captured CO2. The Weyburn unit in southeast Saskatchewan began CO2 injection in October 2000 and the neighbouring Midale unit in 2005, both supplied through a pipeline of about 320 km from the Great Plains Synfuels Plant at Beulah, North Dakota, which gasifies lignite. SaskPower's Boundary Dam power station has captured CO2 since 2014 and supplies part of it to the region's EOR operations. In Alberta, the Alberta Carbon Trunk Line, in service since 2020, carries CO2 captured at a fertilizer plant and a refinery near Edmonton to the Clive field for EOR. The Weyburn operation has changed hands over the years and is now operated by Whitecap Resources.

Supply shapes project design. A flood that buys CO2 under a fixed take-or-pay contract must plan for a steady purchase rate for many years, so it develops patterns in phases, moving purchased CO2 to new patterns as the early patterns' recycle volumes grow.

Recycling and the CO2 balance

Between 30 and 70 percent of the injected CO2 eventually returns to the surface with the produced oil and water, the share rising as the flood matures. Because the CO2 is valuable, it is separated from the produced gas, dehydrated, compressed and reinjected. A mature CO2 flood is as much a gas processing operation as an oil field: produced gas may contain 80 to 90 percent CO2 together with methane and natural gas liquids, which are recovered as a by-product. The CO2 balance distinguishes two utilization ratios. Gross utilization is the total CO2 injected, purchased plus recycled, per barrel of incremental oil; net utilization counts only purchased CO2. Permian Basin floods have typically reported gross utilization of about 10 Mcf per barrel and net utilization of 4 to 8 Mcf per barrel, roughly 0.2 to 0.4 tonnes of purchased CO2 per barrel.

# CO2 balance of a tertiary flood: purchased, recycled and stored CO2, utilization and credit value.
import math
SI = True      # True: metric units, False: oilfield units
MCF_PER_T = 18.9                   # Mcf of CO2 per tonne
gross = 180e3                      # CO2 injected per year, t (purchased + recycled)
target = 0.68e6                    # incremental oil over 15 years, m3
credit = 40.0 if SI else 85.0      # value per tonne stored, dollars (assumed / 45Q for EOR)

years = range(1, 16)
shape = [y * math.exp(-y / 4.5) for y in years]
oil = [target * s / sum(shape) for s in shape]
rows, cp = [], 0.0
for y, o in zip(years, oil):
    rec = gross * 0.65 * (1 - math.exp(-(y - 1) / 3.0))
    buy = gross - rec
    rows.append((y, o, buy, rec)); cp += buy
co = sum(oil); stored = cp * 0.995
print(" yr   oil       purchased  recycled   (CO2 in kt)")
for y, o, b, r in rows[::2]:
    ov = f"{o / 1e3:6.1f} km³" if SI else f"{o * 6.2898 / 1e3:6.0f} Mbbl"
    print(f" {y:2d}  {ov}   {b / 1e3:7.0f}   {r / 1e3:7.0f}")
gu = gross * 15 / co; nu = cp / co
if SI:
    print(f"Gross utilization {gu:.2f} t/m³, net utilization {nu:.2f} t/m³ of incremental oil")
else:
    print(f"Gross utilization {gu * MCF_PER_T / 6.2898:.1f} Mcf/bbl, net {nu * MCF_PER_T / 6.2898:.1f} Mcf/bbl of incremental oil")
print(f"CO2 stored {stored / 1e6:.2f} Mt; at {credit:.0f} dollars per tonne worth {stored * credit / 1e6:.0f} million dollars over 15 years")
print(f"Stored CO2 per unit of oil: {stored / co:.2f} t/m³" if SI else f"Stored CO2 per barrel: {stored / (co * 6.2898):.2f} t/bbl")

 yr   oil       purchased  recycled   (CO2 in kt)
  1    31.5 km³       180         0
  3    60.6 km³       123        57
  5    64.7 km³        94        86
  7    58.1 km³        79       101
  9    47.9 km³        71       109
 11    37.5 km³        67       113
 13    28.4 km³        65       115
 15    21.0 km³        64       116
Gross utilization 3.97 t/m³, net utilization 1.99 t/m³ of incremental oil
CO2 stored 1.35 Mt; at 40 dollars per tonne worth 54 million dollars over 15 years
Stored CO2 per unit of oil: 1.98 t/m³

The code reproduces the shape of real CO2 balances. Purchases fall by more than half as recycle grows, and the net utilization of about 6 Mcf per barrel, or 2.0 tonnes per cubic metre of oil (0.32 tonnes per barrel), is in the typical range. The purchased CO2 remains in the reservoir at the end of the flood, apart from small operational losses. That fact turns every CO2 flood into a CO2 storage project, whether or not that is its purpose.

Incentives for CO2 EOR

The economics of CO2 EOR depend on CO2 cost and on policy. Saskatchewan has long supported EOR through its royalty and tax structure for approved projects, and in 2021 extended its Oil Infrastructure Investment Program, which offers transferable royalty credits worth 20 percent of eligible costs, to pipelines carrying CO2 for EOR and storage. The federal investment tax credit for carbon capture, utilization and storage, by contrast, excludes capture projects whose CO2 is used for EOR, a policy choice that has been debated in the industry. Emissions credits under provincial pricing systems can apply where storage is quantified under an approved protocol. In the United States, the federal 45Q tax credit pays for each tonne of CO2 captured and either stored or used; legislation in 2025 raised the credit for CO2 used in EOR to parity with dedicated geological storage, about 85 dollars per tonne for qualifying projects.

Operations and storage

Facilities for CO2 floods must resist corrosion where CO2 and water meet: injection wells use corrosion-resistant alloys or lined tubing and CO2-resistant cement, and producers handle carbonic acid, scale and sometimes asphaltene precipitation, which CO2 can trigger in some oils. Old wellbores that penetrate the flooded zone are inspected and, where necessary, re-plugged, because they are the main potential leakage path for CO2. Safety systems deal with the asphyxiation hazard of CO2, which is heavier than air and pools in low areas, and with hydrogen sulphide where it is present.

Storage in EOR is called associated storage. Monitoring at Weyburn and Midale under the IEA Greenhouse Gas R&D Programme project from 2000 to 2011 used repeated seismic surveys, soil gas sampling, groundwater chemistry and geochemical tracing to follow the injected CO2, and found no evidence that it had leaked from the reservoir. Published estimates put the eventual storage in the two units at about 40 million tonnes. ISO 27916, published in 2019, sets out how to quantify and document CO2 stored through EOR. The deeper science of trapping mechanisms, long-term monitoring and dedicated saline aquifer storage is the subject of PET-460.

For Ethan's study, the lesson is that a CO2 flood in the Cardium A pool is technically plausible but depends on a CO2 source. He adds a line to his task list: identify the nearest sources of captured CO2, their volumes and the pipeline distance.

Monitoring a CO2 flood

Surveillance of a CO2 flood builds on the waterflood tools of Lesson 5 and adds gas-specific measurements. Operators track the CO2 content of produced gas at each well, because a sudden rise marks breakthrough along a high-permeability path; they adjust the WAG ratio, cycle length or injection rate of the supporting injectors in response. Injection profiles are run during both water and CO2 cycles, since the two fluids often enter different intervals. Pattern-level CO2 balances, comparing CO2 injected and produced, identify patterns that are losing CO2 out of the pattern or cycling it inefficiently through a thief zone. In mature Permian floods, managing the WAG schedule pattern by pattern in response to these data has been one of the main ways operators reduce net utilization and keep CO2 working in the unswept parts of the reservoir.