4.3 Leakage via Wellbores

In addition to leakage of CO2 from reservoirs via the ‘natural’ pathways discussed above, potential release via wellbores must also be considered. There are two principal ways in which wellbore leakage can occur. The first is catastrophic well failure, or blowout, resulting in the rapid release of large volumes of CO2 to the atmosphere. The second is leakage or seepage of CO2 along high permeability channels associated with the wellbore, delivering CO2 to higher stratigraphic levels, and potentially delivering CO2 to the seafloor or the ocean surface. Indeed, while the probabilities of catastrophic release are slight, the migration of CO2 along preferential pathways associated with the wellbore is widely accepted to be the most likely mode of leakage or seepage of CO2 from a geological storage reservoir.

4.3.1 Well Blowout

A well blowout is the uncontrolled flow of reservoir fluids into the wellbore, and sometimes catastrophically to the surface. If reservoir fluids flow into another formation and do not flow to the surface, the result is called an underground blowout. In this case, if the wellbore has significant openhole intervals, it is possible that the well will bridge over downhole (seal itself with rock fragments from collapsing formations). However, in the event of catastrophic failure, large volumes of CO2 would be released directly to the atmosphere. Due to the possibility of severe accidents, the main concern in the event of blowout must be the hazard to workers in the vicinity of the release (both at the time of blowout and during remediation).

The likelihood of well blowout can be assessed from offshore Oil & Gas experience through the use of an analogue database from which the relative frequency, duration, and magnitude of well releases for a given well category can be ascertained. Combining this with the number and type of wells in the storage reservoir then allows a site specific risk of blowout to be estimated (e.g. FutureGen, 2007). Commercial databases (restricted access) are available to the industry sector, such as the SINTEF Offshore Blowout Database which includes information on 515 offshore blowouts and well releases that have occurred world-wide since 1955 (see www.sintef.no for more details). Table 4.1 shows the mean well blowout frequency from North Sea and US Offshore Oil & Gas experience (after Holland, 1997). The frequency of production well blowout is low, at 1 event every 16,000-20,000 well-years, with slightly higher frequencies of blowout in the US during well workover (repair and maintenance). Blowout frequency during drilling suggests 1 event for every 150-600 wells drilled. While these frequencies indicate that the likelihood of catastrophic well blow-outs is exceedingly slight, the eventuality that stored CO2 escapes rapidly in great amounts at once cannot be completely neglected.

Table 4.1: offshore well blowout frequency - oil & Gas experience (after Holland, 1997)

  North Sea US
Production Well(per well-year) 6.00×10-5 5.00×10-5
Workover Well(per well-year) 6.00×10-5 1.70×10-4
Exploration Well Drilling(per well drilled) 6.66×10-3 5.93×10-3
Development Well Drilling(per well drilled) 1.65×10-3 3.99×10-3

In the event of blowout, the maximum rate at which CO2 will vent to the atmosphere is constrained by choked flow through the well bore. For an ideal gas, the mass flow rate under choked flow is given by7:

[Eq. 4.1]


[Eq. 4.2]

Where QCO2 is the mass flow rate of CO2 (kg/s), C is the discharge coefficient for the orifice (typical values are near 1), A is the cross sectional area of the orifice (m2), γCO2 is the specific heat ratio (the specific heat capacity of CO2 at constant pressure divided by the specific heat capacity of CO2 at constant volume, Cp(CO2)/Cv(CO2), 1.308 at STP), ρCO2 is the density of CO2 (kg/m3), and PCO2 is the pressure of CO2 in the wellbore (Pa).

CO2 will continue to flow at this maximum rate while pressure in the wellbore meets the pressure criterion:

[Eq. 4.3]

For example, for γCO2 of 1.308, choked flow will continue while PCO2 is more than 1.84 times higher than atmospheric pressure (Patm). Thus in the case of storage in an overpressured reservoir, choked flow could be expected to continue until the wellbore is controlled.

Although the rate of CO2 release is very fast in the event of blowout, such events are readily detected (primarily due to the rapid rate of release) and remedial action is possible, where Offshore Oil and Gas experience indicates that the typical duration of a high release event (time from blowout to control) ranges from 0.5 to 5 days (Holland, 1997; average duration for all North Sea and US well blowouts). The total amount of CO2 likely to be released in the event of a catastrophic blowout can be estimated from the mass flow rate and the duration of the leakage event. For a simple illustration, Figure 4.2 shows the mass flow rate of CO2 assuming choked flow at typical injection conditions (T=35°C, P=15-20 MPa) as a function of wellbore diameter. If it is assumed that pressure in the wellbore does not decrease significantly over time, then for a wellbore diameter of ~10cm approximately 50 ktCO2 would be released per day.

Figure 4.2: CO2 release rate under choked flow conditions.

4.3.2 Leakage or Seepage along a Wellbore

As noted above, existing well penetrations represent the most likely pathway for CO2 migration out of the storage reservoir. When CO2 dissolves in water it creates an acidic environment that is detrimental to the long term integrity of both cement and casings (e.g. Davis & McDonald, 2005; Scherer et al., 2005). Over time, cement degradation and casing corrosion could create preferential channels for CO2 migration. In addition, there may be formation damage due to drilling of the well, and wells may be poorly completed, with poor bonding between the cement and formation or cement and casing, contaminated cement, or the absence of cement altogether (e.g. Watson et al., 2002). Potential leakage pathways along an individual wellbore include (see Figure 4.3):

  • through damage zones in the formation immediately adjacent to the wellbore
  • between the cement sheath and the formation
  • between the cement sheath and the casing
  • through the cement (sheath and plug)
  • through corroded casing
  • through fractures in the cement or regions without cement

Figure 4.3: Potential CO2 leakage pathways via wellbores (after Celia at., 2004).

Any wellbore within the spatial footprint of the storage reservoir may act as a conduit for leakage. This includes wells that penetrate either deeper or shallower strata than the target storage reservoir, abandoned wells, exploration wells, producing wells, and the CO2 injection wells themselves. In the case of leakage up an improperly abandoned wellbore, the CO2 could leak through a corroded casing wall and travel upward through the casing to the surface. This represents a potential escape route that not only breaches the seal, but also the overburden, thereby bypassing any potential secondary trapping mechanisms that may act in the overburden (see Chapter 5). Alternatively, migration outside the casing could deliver CO2 into shallower formations, which may eventually release CO2 to the sediment-seawater interface. Similarly, in the case of leakage up a poorly sealed or failed injection well casing, the CO2 could leak out the well through a corroded casing wall travel upward outside the casing or migrate into adjacent formations, again potentially reaching the seafloor.

The principal control on the actual migration path is the distribution of high and low permeability segments both along a single well and between different wells (Celia et al., 2006). For example, CO2 may migrate out of the storage formation along one wellbore, then migrate laterally through a permeable formation and find another flow path along a second wellbore to the sediment-seawater interface. While intact cements have low permeability (~10-5 md), degraded cements will have significantly increased permeability, and an annular opening (a ring shaped gap) around 1mm wide between the rock and the cement sheath (due to poor bonding) has an effective permeability on the order of 100 d (105 md) (Celia et al., 2004). Furthermore, at the 10s of cm scale there could be large pockets with no cement between the formation and the casing.

4.3.3 Relative Risk of Leakage along Different Wellbore Types

Decommissioned wellbores are the most likely to have high permeability segments along the wellbore, and therefore the probability of leakage is highest along these wells. Specifically, the highest probabilities of leakage are associated with:

  • Undocumented Wells
  • Poorly Constructed &/or Improperly Abandoned Deep Wells (penetrates primary seal)
  • Poorly Constructed &/or Improperly Abandoned Shallow Wells (above primary seal)

In comparison, the likelihood of leakage along currently producing wells is expected to be lower, and the lowest probability of leakage is associated with the CO2 injection wells themselves. Undocumented Wells & House Keeping Records

Undocumented wells represent the highest risk of leakage as no remedial action can be taken to ensure seal and plug integrity prior to CO2 injection. Furthermore, should leakage along an undocumented wellbore occur, it is highly unlikely to be detected. Thus to determine the overall likelihood of leakage from a storage reservoir an assessment of the number of undocumented wells within the plume footprint will need to be made.

For the North Sea, the number of undocumented wells is expected to be low, if not zero, as house keeping records are understood to be excellent. In areas where records are not as good, a judgement will have to be made using information based on historical exploration in the area (c.f. FutureGen, 2007).

In all cases, care should be taken both to preserve and catalogue existing documentation of drilling activities, and to ensure excellent records are kept and maintained in the future. Indeed, a continuously updated record of storage sites, including wellbore characteristics, monitoring and repair activities, can be considered vital to the long term success and stewardship of storage projects. Given the importance of such records, and the need to preserve them over long timescales (1000+ years) it seems sensible to suggest they should be lodged with a regulatory body (be it national or international). Decommissioned Wells & Abandonment Procedures

At present there are no specific abandonment procedures for ‘CO2’ wells (defined here as any well that falls within the footprint of a target CO2 storage reservoir), and a review and adaptation of standard plugging-and-abandonment (P&A) procedures is required to ensure that both sheaths and plugs are fit for purpose (e.g. Barlet-Gouédard et al., 2006).

At abandonment, the well is typically sealed by a cement plug. For the vast majority of abandoned wells, CO2 injection would not have been contemplated at the time of decommissioning. Thus inappropriate decommissioning of wells (e.g. insufficient plug) seems likely to be a widespread problem (Barlet-Gouédard et al., 2006). For all ‘CO2’ wells, an assessment should be made of the location, thickness and material (e.g. type of cement) of the plug and casing sheath, and the likely continuity of the sheath. If required, remedial action can be taken prior to CO2 injection, where documented wellbores can be re-entered and re-plugged. Barlet-Gouédard et al., (2006) suggest that if deemed necessary by the criticality of the plug, the casing and cement can be milled all the way to the storage formation before placing a cement plug directly in contact with the formation. Ideally, all ‘CO2’ wells should be resealed using chemically resistant cements (see Section below). However, given the expense of this procedure it seems more likely that existing wells with intact Portland cement seals will be monitored for leakage during storage operations, and if leakage is detected, then the wellbore will be re-entered and resealed (see for example the Gorgon project risk assessment, ChevronTexaco Australia Pty Ltd., 2005). Current Wells & CO2 Injection Wells

For current producing or exploration wells that penetrate a potential future CO2 storage reservoir, there exists an opportunity while the wells remain accessible to ensure that both the plug and casing sheath are of sufficient quality for CO2 storage purposes. For example, for a well that will continue to produce during CO2 storage operations (e.g. via EOR, or if it penetrates a hydrocarbon reservoir that either over or underlies the storage reservoir) an assessment should be made of the quality of the sheath prior to CO2 injection, and remedial action taken if necessary. Likewise, for a well that is to be decommissioned, the sheath should be checked (or the casing removed altogether) prior to plugging.

As CO2 injection wells will be specifically engineered and constructed to withstand the corrosive conditions in the storage reservoir (e.g. using chemically resistant cements), these wells can reasonably be considered to represent the lowest risk of leakage. Other Drill Holes

In addition to Offshore Oil & Gas operations, there may be other drill holes located in a target storage reservoir. For example, the Integrated Ocean Drilling Program (previously the Deep Sea Drilling Program, then the Ocean Drilling Program) is an international research program that explores the history and structure of the earth as recorded in seafloor sediments and rocks. These scientific drilling programs have now cored seafloor sediments and rocks at over 1300 sites. While the majority of drill holes at these sites only extend a few hundred metres into the overburden, 28% have drill holes that extend beyond 500m, and 5% have drill holes that extend beyond 1000m. While the likelihood of a deep DSDP/ODP/IODP drill hole occurring within the footprint of a target storage reservoir is low, it should nevertheless be checked. Cement Integrity

Traditionally wells have been sealed (both sheaths and plugs) with standard Portland cement, a material that is inherently unstable with respect to reactions with CO2. New materials are now becoming available that offer either flexibility and expansion (lightweight or foam cements, of benefit to ensure good cement distribution in horizontal or deviated wells) or long term durability to CO2 attack (chemically resistant cements). Data is needed on the long-term integrity of these materials when exposed to a CO2-rich environment. At the decade scale, this can be obtained either by studying the condition of cement seals that are known to have been exposed to CO2 over a given period of time (e.g. Carey et al., 2006), or by carrying out laboratory testing, and using data obtained over a comparatively short period to extrapolate to long term seal performance (e.g. Barlet-Gouédard et al., 2006).

Carey et al. (2006) investigated the impact of CO2-cement interactions on wellbore samples from the world’s second oldest continuous CO2-flooding operation, the SACROC unit, located in the Permian Basin of West Texas. The SCAROC unit is a limestone oil reservoir with a shale caprock. Samples were analysed from a region extending from the limestone-shale contact, to 6m above the reservoir, where the well from which samples were taken was drilled and cemented in 1950 (using an additive-free Portland type 1 cement) and first exposed to CO2 in 1975. The analytical results (structural integrity, permeability, and petrography) indicate that the cement retained its capacity to prevent significant flux through the cement matrix itself. However, significant carbonate precipitation was observed at both the interface between the shale caprock and the cement, and between the casing and cement, indicating migration of CO2 along the shale-cement and casing-cement interfaces. The origin of the CO2 at the casing-cement interface may have been derived by migration along this interface from the reservoir or from the interior of the well at casing joints or regions of casing corrosion (Carey et al., 2006). Carey et al. (2006) therefore concluded that the integrity of these interfaces appears to be the most critical issue in wellbore performance for CO2 sequestration (Carey et al., 2006).

In a laboratory study, Barlet-Gouédard et al. (2006) tested the integrity of Portland cement and a new resistant cement with respect to both CO2 saturated aqueous fluid and wet supercritical CO2. They conducted their experiments over a range of temperatures and pressures that encompass expected in-situ conditions (T = 30-300°C, P = 1-50 MPa) with experiment durations ranging from 2 days to 3 months. The evolution of the cement chemical composition and porosity with time was fully characterized by scanning electron microscopy, chemical analyses, backscattered electron image analysis and Hg-porosimetry measurements (Barlet-Gouédard et al., 2006). The alteration observed was then used to create a predictive model of cement integrity over longer time periods. They found that Portland cement was not mechanically resistant to either wet supercritical CO2 or to CO2-saturated water, where an initial sealing by carbonation is followed by a dissolution stage, which starts earlier in CO2-saturated water than in wet supercritical CO2. The model predicts an alteration front of 100 mm after 20 years of CO2-attack possibly, potentially destroying zonal isolation and triggering casing corrosion. In contrast, the CO2 resistant cement was found to be comparably inert, where the good stability of this material was confirmed by weight, density, compressive strength, microstructural characterizations and Hg porosity measurements.

The differences between these two studies, with the laboratory tests indicating Portland cement failure after a period of 20 years, and the SACROC data indicating the cement itself had maintained its integrity over a period of 30 years, will reflect a number of different parameters, the most obvious of which will be CO2-water-cement ratio (e.g. a cement plug completely immersed in either wet supercritical CO2 or to CO2-saturated water versus a cement plug where supply of CO2 is limited by migration). Nevertheless, the laboratory study indicates that carbonation does not continuously plug Portland cement, suggesting that the SACROC cement can be expected to deteriorate further over time. Given the requirement for secure storage over 1000+ years, all new wells and any remedial measures taken to reseal existing wells should utilise chemically resistant cements. Site Screening

The existence of multiple known wellbores and/or a high possibility of multiple undocumented wells may render a site unsuitable for storage purposes (i.e. if the cumulative probability of leakage along wellbores is deemed too high, or if costs of any required remedial action outweighs any economic benefits associated with that site).

Based on analysis of documented wellbore leakage in Alberta Canada, a mature on-shore sedimentary basin with a statistically significant number of wells (over 300,000), Bachu et al. (2006) recommended that risk assessments should be based on the following criteria:

  • Well age
  • Well status (active, inactive, abandoned)
  • Well casing, or lack thereof
  • Well direction (vertical, deviated, horizontal)
  • Cementing intervals
  • Level of drilling activity
  • Global and local events that may have affected drilling practices
  • Regulations and their timeline of being introduced.

While offshore basins do not generally face the same level of difficulties as onshore basins in North America (e.g. generally lower drilling densities), these criteria seem equally valid for preliminary screening of offshore storage sites. In the absence of hard data, the categorisation of wells according to this scheme, combined with the use of an analogue database, would allow a preliminary assessment of the likely distribution of permeability along the wellbores that fall within the footprint of the storage reservoir.

If the permeability distribution along all wellbores is known, then potential leakage pathways and rates can readily be determined using computationally fast semi-analytical models (e.g. Celia et al., 2006). The use of such a model at the site selection stage would allow identification of any critical wellbores that require remedial action prior to CO2 injection.

As an illustrative example, Figure 4.4 shows a simulated permeability distribution for 1000 segments along wellbore casings. This example assumes a log-normal distribution for two classes of cement segments (1) intact cement segments with a mean permeability κ1 of 10-5 md and a variance (σ2) of var[log10κ1]=12 and (2) degraded or poorly bonded cement segments with a mean permeability κ2 of 0.1 md and a variance of var[log10 κ2]=22 (after Celia et al., 2006). In the example shown in Figure 4.4(a) a 1:1 ratio of intact to degraded segments is assumed. Figure 4.4(b) shows the cumulative permeability distribution for varying proportions of intact to degraded segments. If we compare these permeability distributions to caprock permeabilities, the following becomes apparent: All well segments (even when 100% intact segments) are more permeable than the best evaporite seals (halite, 10-8 to 10-9 md). A significant proportion of well segments will be more permeable than the poorest quality shale seals (~10-1 md), where as the proportion of degraded segments increases from 25% to 50% to 75%, the number of segments with relatively high permeability (≥0.1 md) increases from 12.5% to 25% to 37.5%.

Figure 4.4:(a) Permeabiity distribution for 1:1 ratio of intact to degraded cement segments (b) Cumulative permeability distribution for varying proportions of intact to degraded cement segments(after Celia et al., 2006) Well Monitoring & Repair

Evaluating and monitoring the integrity over time of steel casings and cement sheaths and plugs is of the utmost importance for the long term security of CO2 storage.

In CO2 injection wells, and other active wells in the footprint of the storage reservoir, well integrity should be regularly checked across the injection interval, the cap rock and shallower zones (Barlet-Gouédard et al., 2006). Over the past 20-30 years there has been considerable improvement in the tools available for this task, and in active wells it is possible to assess the integrity of both the steel casing and the sealing material in the annulus behind the casing (see review by Vu-Hoang et al., 2006). For example, modern tools such as the Isolation Scanner and Sonic Scanner enable both identification of channelling and examination of the full cement sheath between the casing and formation, with high vertical and azimuthal resolution, thus providing the ability to create a detailed Cement Bond Log (Vu-Hoang et al., 2006). In addition, casing corrosion can be evaluated by combining electromagnetic and ultrasonic measurements of the metal thickness (Vu-Hoang et al., 2006).

However, as these techniques require passing the measurement instrument down the wellbore (wireline logging), they cannot be readily employed on decommissioned wells. Monitoring the integrity of abandoned wells thus requires more attention, and new research on appropriate monitoring techniques will most likely be required. Methods for monitoring these aspects are currently under investigation within the Carbon Dioxide Capture project (www.co2captureproject.org/), the results of which are not yet publicly available.

Adequate remedial operations must also be concurrently developed to re-establish zonal isolation when a potential leakage path has been detected. For instance, squeeze jobs will allow repair of any faulty primary cement job (Barlet-Gouédard et al., 2006).

4.3.3 Scale & Duration of Leakage along Wellbores

The potential scale of leakage along wellbores ranges from relatively slow migration through a high permeability pathway, to faster flow along an annulus or via surface casing vent flow, to rapid release due to well blowout. Based on expert opinion, a number of risk assessments for on-shore geological storage sites have estimated probable slow leakage rates on the order of ~200 tCO2/yr (e.g. Hooper et al., 2005; FutureGen, 2007). Release of CO2 at such low rates (~6 gCO2/s) is likely to go undetected, and could therefore remain active for long periods of time (in excess of 1000 years). Conversely, rapid release (at rates in excess of 10,000 tCO2/yr, or over 300 gCO2/s) due to well blowout, can be assumed to be detected and mitigated, and therefore be active for short periods only (e.g. 0.5 to 5 days).

7 Note: As CO2 is a non-ideal gas this equation provides an approximation for the mass flow rate under choked flow. For an accurate description of CO2 mass flow rates real gas effects such as compressibility and variable specific heats need to be taken into account, where these effects can make a significant difference to the mass flow rate. Nevertheless, this equation allows an order of magnitude calculation to be made for illustrative purposes.