4 Impact of CO2 on wellbore integrity

In spite of the many similarities of hydrocarbon production and temporal storage and geological storage of CO2, several significant contrasts between these activities have to be recognized. The distinct characteristics of CO2 storage operations include the repressurization of subsurface formations, the purpose of long-term containment (i.e. thousands of years) and the potential chemical interaction of aqueous CO2 with formations and well materials. This chapter describes the potential impact of CO2 injection on the wellbore and well cement and steel.

4.1 Potential CO2 leakage mechanisms

Based on a review of natural and industrial analogues for CO2 storage, Benson et al. (2002) concluded that the most important cause of failure of an injection well is the application of well construction materials that were incompatible with the injected fluids, leading to excessive casing corrosion. Alternatively, failure was reported to result from well mechanical flaws, damage to the well as a result of excessive injection pressures, inadequate monitoring of annulus pressure or lack of detection of fluid migration behind the casing (Benson et al., 2002). In general, several mechanisms can give rise to leakage of fluids along the well system as illustrated in Figure 4.1.

As wet CO2 or CO2 in solution are corrosive fluids, specific attention is required for chemical degradation of well materials in CO2 storage projects. The combination of CO2 and water will result in both chemical degradation of the oil well cement (e.g. Bruckdorfer 1986, Scherer et al. 2005, Barlet-Gouédard et al. 2006), thereby potentially enhancing porosity and permeability, and corrosion of the casing steel, creating pathways through the steel.

Mechanical deformation of cement, casing material or host rock due to operational activities (e.g. drilling, pressure and temperature cycles) or natural stresses can result in the development of cracks or shear strain, enabling highly permeable pathways through these media to develop (Ravi et al., 2002a; Shen & Pye, 1989). Furthermore, the loss of bonding between different materials (also called debonding) could cause annular pathways along the interfaces between casing, cement or host rock. This process could result from a poor cement placement job or cement shrinkage. Volumetric shrinkage of cement upon placement of 4% is not uncommon in oil and gas industry.

Cementing is critical to the mechanical performance and integrity of a wellbore both in terms of its method of placement and the type or class of cement used. In practice, a good quality of cement plugs and primary cement sheath has to be confirmed by e.g. (pressure/weight) testing, log data and drill-off performance. The primary cement sheath both prevents behind-casing flow of fluids as well as protects the casing from corrosion by e.g. aqueous CO2 or brines, not only at seal levels, but also at shallow depths (Watson & Bachu, 2007). It is sensitive to flaws resulting from e.g. bad mud removal, decentralized casing (especially in deviated wells), non-optimal placement (for details see Barclay et al., 2002), or cement failure under stress. Even a good cement bond log (CBL) is no guarantee for a channel-free cement sheath (Carey et al., 2007; Watson & Bachu, 2007).

Figure 4.1 Potential leakage pathways through an abandoned well: (a) along the casing-cement interface; (b) along the cement plug-casing interface; (c) through the cement pore space as a result of cement degradation or fracturing; (d) through casing as a result of steel deformation or corrosion; (e) through fractures in cement; and (f) along the cement-rock interface. After Gasda et al. (2004).

4.2 Chemical degradation of wellbore cement

4.2.1 Chemical degradation mechanism

Although CO2 itself is not corrosive, in combination with water (either wet supercritical CO2 or CO2 dissolved in water) CO2 dissociates to form carbonic acid:

CO2 + H2O ↔ H2CO3 ↔ H+ + HCO3 ↔ 2H+ + CO32− (1)

After dissolution of CO2, cement degradation commences with the progressive consumption of the solid cement constituents Portlandite (Ca(OH)2(s)) and Calcium Silicate Hydrates [C-S-H(s)] to produce carbonates (aragonite, vaterite and/or calcite), amorphous silica gel (SiOxOHx) and water. This stage is called ‘carbonation’:

Ca(OH)2(s) + 2H+ + CO32− → CaCO3(s) + 2H2O (2a)

Ca(OH)2(s) + H+ + HCO3 → CaCO3(s) + 2H2O (2b)

[C3,4-S2-H8](s) + 2H+ + CO32− → CaCO3(s) + SiOxOHx(s) (3a)

[C3,4-S2-H8](s) + H+ + HCO3 → CaCO3(s) + SiOxOHx(s) (3b)

Note that in the presence of magnesium ions in the formation water, also dolomite (CaMg(CO3)2) will be formed.

The progressive dissolution of C-S-H creates a high-porosity ‘dissolution front’ in the cement (Figure 4.2). The produced calcium carbonates precipitate at the ‘carbonation front’ reducing the porosity. During the subsequent ‘leaching’ step, the calcium carbonate could be again dissolved by the CO2-brine (‘dissolution back-front’ in Figure 4.2), increasing the porosity and resulting in a strong degradation of the cement:

CaCO3(s) + H+ ↔ Ca2+ + HCO3 (4)

The entire cement degradation process as observed in the laboratory is formed by three subsequent steps that result in the distinct formation of progressive zones of alteration in samples under CO2 attack as shown in Figure 4.2 (Barlèt-Gouédard et al., 2006). During the cement carbonation stage the porosity of the cement is reduced by precipitation of calcium carbonates. If the permeability is reduced as well this process is likely to have a decelerating effect on the diffusion rate of CO2 (WBIN4, 2007). However, supply of additional CO2 may again lead to dissolution of CaCO3 during the leaching stage. The final result of this total degradation process would then be an amorphous silica gel. Rimmelé et al. (2008) found that also the apparent unaltered inner part of the sample rapidly reacted in the first days of exposure, leading to increased porosity and sporadic calcite precipitation throughout the sample.

Note that the combination of chemical reactions (1) to (3) involves no net increase or decrease of H+ ions in the solution. Upon consumption of hydrogen ions in reactions (2) and (3), supply of H+ ions is warranted by dissolution of CO2 according to reaction (1). Hence, no net effect on pH is expected from these reactions. However, if dissolution of calcium carbonate according to reaction (4) is incorporated, the pH would increase. Obviously, the occurrence of these reactions is governed by the chemical equilibrium. Due to the anticipated surplus of CO2 and its dissolution products from reaction (1), reactions (2) and (3) are expected to shift to the right-hand side. The occurrence of reaction (4) is less straightforward and the composition of the formation water will have an important effect on this ‘leaching’ step.

Figure 4.2 Back-scattered electron image of a Portland cement sample after degradation in a water-saturated supercritical CO2 fluid. After Barlèt-Gouédard et al. (2006).

At present the range of the permeability of degraded cement remains unclear. Although some authors indicate a relatively low permeability in the order of micro- to milli-Darcy based on laboratory experiments (e.g. Barlèt-Gouédard et al., 2006; Duguid et al., 2006; Lécolier et al., 2006) and on field observations over a maximum of 30 years of exposure to CO2 (Carey et al., 2007), it is uncertain how the permeability will develop over longer time scales. Furthermore, it has been established that the remaining amorphous silica gel has an extremely low strength. As a consequence it will be easily washed out by fluid flow or deformed when pressure gradients or buoyancy would be applied.

The w/c (water to cement) ratio of Portland cements has a major impact on the resistivity of cement to acid attack (e.g. Bruckdorfer 1986, Van Gerven et al. 2004). This ratio has to be kept as low as possible in order to keep the cement porosity and permeability low and to avoid capillary forces driving CO2 into the cement. If less viscous cements are needed in the field, the use of plasticizers is beneficial over adding water.

Experiments on the effects of cement additives returned mixed results. Some results indicated that addition of fly ash increases the degradation rate of cement significantly (WBIN4, 2007) or at least does not improve carbon dioxide resistance (Bruckdorfer, 1986). Bentonite, another commonly used additive, was found to dramatically decrease the cement’s resistance to acid attack (US DOE, 2006). Strazisar et al. (2008) studied the effect of Halliburton Pozmix A - a Class F fly ash - on material degradation under influence of CO2, qualitatively comparing results of similar experiments on 65:35 and 35:65 pozzolan-cement mixtures. While the 35:65 pozzolan-cement sample displayed phenomena probably associated with precipitation of calcium carbonate at the sample’s rim, presumably the 65:35 sample was entirely carbonated. However, neither sample showed significant alteration of its physical properties (Strazisar et al., 2008). On the other hand, Halliburton found that a 50:50 mixture of pozzolan and Portland cement is less susceptible to carbonic acid leaching and maintains a lower permeability over a longer period of time compared to formulas composed entirely of Portland cement (Onan, 1984).

4.2.2 Degradation rates

Numerous literature references deal with cement degradation under influence of CO2 (e.g. Barlet-Gouedard et al., 2006; Bruckdorfer, 1986; Duguid et al, 2004; 2006; Duguid, 2008; Lecolier et al, 2006; Shen and Pye, 1989; Van Gerven et al., 2004) allowing for estimation of cement degradation rates under in-situ reservoir conditions. An inexhaustive summary of experimental results reported in literature is presented in Table 4.1.

Diffusion of CO2 into the pores of the cement is considered to be the rate-controlling step in cement degradation (e.g. Garcia-Gonzalez et al., 2006; Duguid, 2008). However, not all authors presented diffusion coefficients from their results. For these studies, coefficients were estimated based on the reported results of penetration depth versus time. Reported and derived coefficients based on experiments at different conditions are summarized in Table 4.1 to aid comparison of the different results. Results from BarletGouédard et al. (2006) show that degradation under influence of water-saturated supercritical CO2 is somewhat faster than under influence of dissolved CO2. In general degradation rates increase when cement is exposed to CO2 at higher temperature or lower pH conditions. Kutchko et al. (2007) showed that also the curing conditions determine the cement’s susceptibility to degradation: curing at higher temperature and pressure results in reduced penetration, relative to samples hardened at lower temperature and/or pressure.

Table 4.1 Comparison of cement degradation rates from literature references

Reference C1 (mm·h½) Time (yr) for 25 mm of cement degradation CO2 penetration depth (m) after 104 yr Cement type P (bar) T (°C) pH w/c ratio Carbonation medium
Barlèt-Gouédard et al. (2006) 0.2622 1.0 2.45 Portland Class G 280 90 - 0.44 Static water-saturated supercritical CO2
0.2182 1.5 2.04 Portland Class G 280 90 - 0.44 Static CO2-saturated water fluid
Bruckdorfer (1986) 0.0623 18.4 0.58 Portland Class A 206.8 79.4 - 0.38 Static wet supercritical CO2
0.0513 27.1 0.48 Portland Class C 206.8 79.4 - 0.38 Static wet supercritical CO2
0.0770 12.0 0.72 Portland Class H 206.8 79.4 - 0.38 Static wet supercritical CO2
0.0843 10.0 0.79 Portland Class H plus 50% fly ash 206.8 79.4 - 0.38 Static wet supercritical CO2
Carey et al. (2008)2 0.0126 450 0.12 ‘Portland cement’ 140/2803 40 7.8-7.6 - Flowing CO2-brine (25,000 ppp NaCl, 4,000 ppm CaCl2, 1,000 ppm MgCl2, 200 ppm MnCl2)
Duguid et al. (2004) 0.0336 63.2 0.31 Portland Class H 1 50 5.03 0.38 Flowing carbonated brine (0.5M NaCl solution)
0.0250 114 0.23 Portland Class H 1 50 4.78 0.38 Flowing carbonated brine (0.5M NaCl solution)
Duguid et al. (2006) 0.0121 487 0.11 Portland Class H 1 23 3 0.38 Berea sandstone with static carbonated brine (0.5M NaCl)
0.00534 2,500 0.05 Portland Class H 1 23 - 0.38 Berea sandstone with static carbonated brine (0.5M NaCl)
Duguid (2008)4 0.00146 33,381 0.014 Portland Class H 1 20 3 0.38 Berea sandstone with static carbonated brine (0.5M NaCl)
0.00159 28,300 0.015 Portland Class H 1 50 3 0.38 Berea sandstone with static carbonated brine (0.5M NaCl)
0.00031 724,281 0.003 Portland Class H 1 20 5 0.38 Berea sandstone with static carbonated brine (0.5M NaCl)
Kutchko et al. (2007) 0.04014 44.2 0.38 Portland Class H 303 50 2.9 0.38 Static carbonated brine (1% NaCl), cured at 22°C, 1 bar
0.03062 76.1 0.29 Portland Class H 303 50 2.9 0.38 Static carbonated brine (1% NaCl) , cured at 22°C, 303 bar
0.03198 69.7 0.30 Portland Class H 303 50 2.9 0.38 Static carbonated brine (1% NaCl) , cured at 50°C, 1 bar
0.01497 318 0.14 Portland Class H 303 50 2.9 0.38 Static carbonated brine (1% NaCl) , cured at 50°C, 303 bar
Kutchko et al. (2008) 0.00329 6587 0.03 Portland Class H 303 50 2.9 0.38 Static water-saturated supercritical CO2
-6 5 0.0025 Portland Class H 303 50 2.9 0.38 Static CO2-saturated brine (1% NaCl)
Lécolier et al. (2006) 0.5568 0.23 5.21 ‘conventional Portland cement’ 150 120 4 0.44 Static H2S-saturated brine
Shen & Pye (1989) 1.3199 0.04 12.36 Portland Class G 69 204 - - Static wet CO2 gas
Van Gerven et al. (2004) 0.7778 0.12 7.28 Portland CEM I 42.5 R 400 80 - 0.34 Flowing supercritical CO2

From the results by different authors, it appears that diffusion-based chemical degradation rates of cement are relatively low. Even under very high temperatures (i.e. 204 ºC at 69 bar) degradation rates would result in a maximum of 12.4 m of cement plug degradation after 10,000 years of exposure to CO2 (Shen and Pye, 1989), assuming that diffusion processes define the degradation mechanism. Depending on the governing abandonment regulations (See Chapter 5), cement plugs generally are required to extend beyond this length. This implies that the lower bound of prescribed plug lengths (i.e. 15 m) would only just exceed the maximum penetration depths under conditions described by Shen and Pye (1989), potentially compromising well integrity. On the other hand, many national regulations require plug lengths of 50 to 100 m, which is significantly longer than maximum observed CO2 penetration. For these plug lengths, diffusion-controlled cement degradation in vertical direction appears to form no significant hazard, assuming wells to be properly cemented and plugged at the level of the sealing cap rock with no fractures or annuli present or induced. However, in lateral directions the 2-5 cm thick primary cement sheath may be degraded in timeframes in the order of (tens of) years, subsequently enabling CO2 to attack the casing steel. Still it seems unlikely that significant amounts of CO2 will flow through an adequate caprock to radially reach and affect the primary cement sheath over long depth intervals.

Lécolier et al. (2007) showed that different cement curing conditions have significant effects on its mechanical strength. Curing in a brine resulted in decreasing mechanical strengths of 20% and 50%, for static and flowing conditions, respectively. Aging in crude oil did not show any mechanical degradation. This could have serious implications for the evaluation of different experimental studies. Although tests performed with brine and pure water show similar alteration patterns, te type of fluid controlling CO2 solubility plays a significant role in the cement degradation rate (Barlet-Gouédard et al., 2009).

Furthermore, Duguid (2008) established that after 2 to 3 months initial degradation rates are reduced to significantly lower values, possibly as a result of calcium carbonate precipitation or reflecting the switch from portlandite degradation to leaching of calcium carbonate. The author defined functions to describe the degradation rates during both of these phases according to Fick’s Law (d = C . t ½ , where d is the penetration depth and t represents time). The diffusion coefficients (C) resulting from these experiments given in Table 4.1 only represent the reduced degradation rates, as these are expected to govern diffusion-based cement degradation on longer time scales (Duguid, 2008). A similar effect was observed by Kutchko et al. (2008), who also reported significant reduction of initially high penetration rates after 2 to 3 months for their cement samples subjected to a CO2-saturated brine. They concluded that the penetration rate is retarded by a diffusive barrier of calcium carbonate, which precipitated during the CO2 attack. However, instead of fitting two separate functions to the data, Kutchko et al. (2008) used an Elovich curve to successfully fit their results in a single relation.

4.2.3 Relevance to in-situ degradation

In order to translate experimental results to field cases, several limiting factors apply with respect to degradation rates. Whereas cement samples in the laboratory in certain cases were immersed in a bath of supercritical CO2, well material in reality will be partially surrounded by reservoir rock, limiting the available reaction surface, the supply of CO2 and the transportation of reaction products. Although some experiments were performed with cement samples embedded in rock material (e.g. Duguid et al., 2004; 2006; Duguid, 2008), this is not taken into account in all reported experiments.

Furthermore, in specific field cases, especially in depleted gas fields, the availability of water necessary for degradation is far more limited compared to the experiments. Moreover, injected CO2 will push back the brine present in the storage formation. As dissolution will take place slowly, many wells will not come across the CO2-water contact at or near critical levels, such as the cap rock. The presence of only connate water would significantly limit the chemical reactivity of CO2.

Finally, higher salinity of formation water will likely decrease the solubility of CO2 and reaction products, thus reducing cement degradation rates (WBIN4, 2007). Especially relatively high calcium concentrations in the fluid, which is often observed in field situations, will significantly reduce dissolution of calcium carbonates. This would have major consequences on the extent of well degradation. First, reduced dissolution of calcite could leave the cement in a state of low porosity and sufficient mechanical strength. No ‘dissolution back-front’ may be developed and progressive degradation may be hampered by the low porosity ‘carbonation front’. The actual occurrence of these phenomena would be strongly depending on the exact location of calcite precipitation (reactions (2) and (3)). If calcite would be transported away from the reaction front to be deposited elsewhere, the effects described above would be limited. Secondly, the pH would not be increased by these reactions, i.e. no pH buffering by cement would occur. These propositions are contrasting with observations in several experimental studies, where calcium carbonates were witnessed to be dissolved (e.g. Barlèt-Gouédard et al., 2006; Duguid et al., 2004; 2006). In this perspective it should be noted that these studies are performed in sodium chloride solutions, enabling dissolution of calcium carbonate during the final ‘leaching’ step.

Investigations of cement field samples that were exposed to CO2 over 30 years during CO2-EOR operations in the SACROC Unit in West Texas (Carey et al., 2007) and from a natural CO2 producer (Crow et al., 2008) support the concept of limited leaching of calcium carbonate. Results from these studies show that CO2 carbonation of cement is capable of healing pathways to some extent (see Section 3.2.2). In contrast to the laboratory results (by e.g. Barlèt-Gouédard et al., 2006), calcium carbonate precipitated from a saturated solution, healing fractures. These observations indicate that specific formation water chemistry may decrease dissolution of calcite and as a result reduce progressive degradation and weakening of the cement.

4.3 Corrosion of casing steel

4.3.1 Corrosion mechanism

The presence of natural or injected CO2 in subsurface reservoirs enhances corrosion of well casing steel. Corrosion of steel is an electrochemical process that requires (1) a cathodic and (2) an anodic reaction, (3) electrical conduction by electrons and (4) transport of reaction products and reactants from and to the reaction surface by an electrolyte. At typical reservoir levels anaerobic conditions are prevalent and CO2 may be in a supercritical state. Under anaerobic conditions the corrosion process of steel by dissolved CO2 is dominated by the following cathodic and anodic reactions:

Cathodic: H2CO3 ↔ H+ + HCO3 ↔ 2H+ + CO32− (5a)

2H+ + 2e ↔ H2(g) (5b)

Anodic: Fe(s) ↔ Fe2+ + 2e (6)

The cumulative reaction involves the formation of Fe2+ ions and hydrogen gas from iron metal and dissolved H+ ions. This mechanism involves the corrosion of iron metal under influence of the presence of dissolved H+ ions in combination with the formation of hydrogen gas. The gas escaping the solution drives the reactions. The electrochemical cell describing this reaction is shown in Figure 4.3.

Figure 4.3 Schematic representation of an electrochemical cell for CO2 corrosion on a steel surface, showing carbonic acid formation, an anodic reaction site with iron dissolution, metallic conduction of the electrons, the cathodic reaction with the consumption of the electrons and transport of ions and species in the electrolyte. Note that the shown cathodic reaction is composed from equilibrium reactions (1), (2) and (3), and all components of these reactions will be present in the system.

The corrosion rate of steel under aqueous CO2 conditions is a function of temperature and partial CO2 pressure. Under typical reservoir conditions corrosion rates could be high, i.e. in the order of ten’s of mm per year (Cailly, 2005). However, wet CO2

corrosion rates on carbon steel at high pressure are observed to be far smaller (Institute for Energy Technology in Norway supported by Statoil, as referred to by Cailly, 2005).

4.3.2 Corrosion rates

The corrosion rate of steel under aqueous CO2 conditions is a function of temperature and partial CO2 pressure and theoretically can be high, i.e. in the order of ten’s of mm per year (Cailly, 2005). Cui et al. (2006) performed weight loss test on P110, N80 and J55 graded steel samples in simulated production water for a range of pH values of 4-6. The resulting corrosion rates decrease with the exposure time as well as with temperature. After 144 hours of exposure to CO2, corrosion rates are in the order of 2 mm/yr (Figure 4.4). Similar observations were made by Wu et al. (2004) on exposure of J55 steel to CO2 leading to a rate of 1.62 mm/yr after 144 hours at 90 ºC and 83 bar. With increasing temperature corrosion rates drop from initial values between 2.2 and 3.5 mm/yr at 60 ºC to values between 0.7 and 1.1 mm/yr at 150 ºC (Figure 4.4). Values in the same order are reported by Lin et al. (2006) from experiments with formation water. It should be noted that in general higher grade steel is more susceptible to corrosion than lower grade samples.

Figure 4.4 Corrosion rate dependency on temperature (left) and exposure time (right) for different steel grades. After Cui et al. (2006).

Fang et al. (2006a; 2006b) determined the influence of high salinity brines (3 wt% - 25 wt% NaCl) on the corrosion rate of C1018 carbon steel under CO2 saturated conditions at pH 4.0 and temperatures ranging from 1 to 25°C. The high salt content retarded the anodic (dissolution of iron) and the cathodic process on the steel surface. The corrosion rates of the steel in a 10 wt% NaCl brine were 2-3 times lower compared to the corrosion rates in a 3 wt% NaCl brine.

4.3.3 Siderite precipitation

The corrosion of carbon steels under CO2 corrosion conditions can be obstructed by the formation of a (partial) siderite (FeCO3) layer under specific conditions (e.g. Johnson and Tomson,1991; Van Hunnik et al., 1996). Combination of reaction products Fe2+ and CO3 form iron carbonate. Having a relatively low solubility, FeCO3 readily precipitates under favourable conditions.

Fe2+ + CO32− ↔ FeCO3(s) (7)

The morphology of a steel surface under CO2 corrosion conditions is primarily determined by the balance of corrosion reactions (resulting in loss of metal) and precipitation reactions (potentially protecting the underlying steel surface). If the precipitation rate exceeds the corrosion rate, a protective film forms at the metal surface which may significantly reduce corrosion rates (Xiao and Nešiü, 2005). The precipitation of solid FeCO3 occurs when the product of the concentrations of Fe2+ and CO32− exceeds the solubility limit. The process takes place in two stages: nucleation and growth. Both processes are related to the level of supersaturation, i.e. the level at which the amount of dissolved material exceeds the saturation limit, which could occur for instance as a result of cooling. At a high supersaturation, iron carbonate crystals will rapidly nucleate on a large number of locations and grow fast, forming a tight and very protective surface FeCO3 film with small crystal size (Xiao and Nešiü, 2005). At low supersaturation, nucleation occurs at a significantly smaller number of locations, crystal growth proceeds slowly, resulting in very large crystal sizes. These large crystals form a much thicker and looser surface layer that is less protective and is easier damaged or washed away by fluid flow (Xiao and Nešiü, 2005). Lin et al. (2006) concluded that when CO2 pressure exceeds a critical value, the grain size and the thickness of the scale reach a minimum value.

The formation of a FeCO3 layer is mainly governed by partial CO2 pressure, temperature and pH. High partial CO2 pressures or high temperatures will result in both increasing corrosion rates as well as enhanced precipitation of iron carbonate. Protective iron carbonate films have been observed in systems with high Fe2+ concentrations, high temperature and partial CO2 pressure, and pH>5 (Nešiü et al., 2003). A general temperature dependency of the film formation was given by Burke (1984) and Palacios & Shadley (1991). While FeCO3 precipitation at temperatures below 60 ºC leads to the formation of a non-protective, porous layer on the steel surface, between 60-100 ºC the film starts to show some protective properties. Above 100 ºC the FeCO3 film becomes tight and adhesive, which is even enhanced by the co-precipitation of magnetite at temperatures over 150 ºC.

It is well established that to form iron carbonate films in CO2 corrosion, the pH has to exceed a critical value, which primarily depends on temperature, Fe2+ concentration, and ionic strength. Corrosion data obtained from experimental investigation by Nešiü et al. (2003) show that in laboratory experiments at 20 °C siderite films form very slowly, and a reasonably high pH value (>6) and very long exposure times are required to form protective films.

4.4 Mechanical deformation of wellbore cement

The majority of experimental data on cement degradation as well as retrieved cement-casing cores from wells in CO2 operations (see Section 3.2.2) indicate that cement is expected to provide adequate resistance against significant CO2 migration through its matrix. Hence, it seems that the mechanical integrity of the cement plug and sheath as well as the quality of its placement could be of more significance than the chemical degradation processes (Scherer et al., 2005). Migration pathways are most likely to occur at the interfaces between cement and both casing and caprock. Therefore, the integrity of these interfaces appears to be the most critical aspect in well integrity for geological CO2 storage (Shen and Pye, 1989; Carey et al., 2008; Bachu and Bennion, 2008), supported by evidence of CO2 migration along these interfaces in field cases (Carey et al., 2007; Crow et al., 2008). This occurrence of micro-fractures or micro-annuli could result from poor cementing jobs (e.g. Barclay et al., 2002) or cement shrinkage (Ravi et al., 2002). This can give rise to debonding and enhanced permeability pathways resulting from e.g. cement shrinkage, poor mud removal or decentralized casing especially in deviated wells (Watson and Bachu, 2007).

Deformation on a reservoir scale can affect the mechanical integrity of the wellbore, i.e. well deformations due to decompaction of the reservoir and casing shear due to shear deformations at the reservoir-cap rock interface or at reservoir bounding faults. Reservoir pressure changes due to depletion or fluid injection can lead to wellbore casing deformation and to casing, cement sheath or plug failure, and therefore affect the integrity of existing wells.

Another important process resulting in damaged cement sheaths is the development of fractures as a result of cement failure under stress, for instance due to high injection pressures and/or temperature changes or cycles (Ravi et al., 2002a, Shen and Pye, 1989). These processes could enhance the widening of existing or developing pathways through the cement or rock (Dusseault et al., 2000).

4.4.1 Reservoir decompaction

As the pore fluid pressure decreases during production and depletion of a reservoir, the effective stress in the reservoir increases, causing elastic and inelastic compaction and surface subsidence (Dusseault et al., 2000; Hettema et al., 2002; Settari, 2002). However, due to CO2 injection the reservoir pressure will increase, leading to decompaction of the reservoir. Abandoned wells in prospective CO2 storage reservoirs (especially depleted oil and gas fields) may have seen effects of preceding hydrocarbon production, depending on the timing of abandonment versus depletion. Certainly these wells have been completed and abandoned before CO2 injection and will be exposed to expansion in the injection phase. This could affect the mechanical stability of the wellbore system.

The reservoir rock and cement possess similar mechanical properties. However, because the elasticity modulus of steel is about 15 times higher that that of cement and reservoir rock, axial deformation of the casing will be negligible relative to the cement and rock deformation. As a result of high strain incompatibility at the cement-steel interface, debonding is likely to take place under extensional loading, leading to the formation of micro-annuli at the interface.

Tensile fractures may occur in the cement when an axial extensional stress would be applied that exceeds the tensile strength of cement. As a result it is most likely that both horizontal tensile cracks and vertical cracks along the cement-rock interface occur in the cement sheath along the well in the reservoir section. The exact type of cracking depends heavily on the local geometric conditions of the cement-rock interface: a highly irregular interface implies a higher value for cement bond strength resulting in a higher probability for tensile cracking.

4.4.2 Shear deformation

(Re-)injection of fluids will in turn cause a decrease of the effective stress in the reservoir. In depleted reservoirs part of the strain will be recovered. These changes in stress and strain could induce shear stresses at the interface between wellbore and reservoir rock, potentially leading to failure of wellbore cement (Dusseault et al., 2000).

As a consequence of reservoir decompaction, shear strains, and possibly slip planes, tend to develop either along interfaces between rock types of different stiffness, or along existing discontinuities or planes of weakness. Particular attention therefore should be paid to the interfaces between the sealing formations and the reservoir rock. Philippacopoulos and Berndt (2000) established that 80-120 mm of lateral displacement as a result of shear stress could result in yielding of the casing.

4.4.3 Micro-fractures and micro-annuli

The primary cement sheath both prevents behind-casing flow of fluids as well as protects the casing from corrosion by e.g. aqueous CO2 or brines, not only at seal levels, but also at shallow depths (Watson and Bachu, 2007). It is sensitive to flaws resulting from e.g. poor mud removal, decentralized casing (especially in deviated wells), non-optimal placement (described in Barclay et al., 2002), or cement shrinkage. Another important process resulting in damaged cement sheaths is the development of fractures as a result of cement failure under stress, for instance due to high injection pressures and/or temperature changes or cycles (Ravi et al. 2002a, Shen and Pye, 1989).

Cement failure via micro-annulus debonding and micro-fracturing of cement is commonly observed in oil and gas industry. Inadequate isolation is predominantly associated with conductive pathways within the cement through micro fractures, or along the interfaces between cement and both casing and formation (Nelson and Guillot, 2006). Even a good cement bond log (CBL) is no guarantee for a channel-free cement sheath (Carey et al., 2007; Watson and Bachu, 2007). The presence or development of fractures or annular pathways in or along the cement dominates the permeability of the cement (Shen and Pye, 1989; Carey et al., 2007; 2008; Crow et al., 2008; Bachu and Bennion, 2008). If present, such pathways can play an important role in leakage mechanisms, significantly enhancing cement degradation (Bennaceur et al., 2004). Bachu and Bennion (2008) showed that annuli or cracks with apertures of 0.01-0.3 mm result in an increase of effective permeability of several orders of magnitude, i.e. from 1 nD for pristine cement to 0.1-1 mD of the sample with annulus or radial cracks.

4.5 Interaction of processes

Until recently, most laboratory experiments focused on the investigation of chemical degradation mechanisms and rates of individual well materials, i.e. cement and steel. Nevertheless, especially at annuli or interfaces between cement and casing or formation, the combined response of associated materials to CO2 plays an important role. Furthermore, the analysis of chemical, mechanical and physical processes has not been iteratively coupled. The evaluation of well integrity in field cases, however, requires profound understanding of the interaction between different processes.

4.5.1 Interaction of casing corrosion and cement degradation along micro annuli

Recently, dedicated experimental work was performed on potential CO2-brine flow along micro-annuli. Carey et al. (2008) executed experiments on a composite sample of cement and steel exposed to a CO2-brine mixture at 40°C, 14 MPa pore pressure and 28 MPa confining pressure. Rather than significant mass loss of the steel, the results showed scaling on the steel surface formed due to the precipitation of siderite (FeCO3). This required a high supersaturation of FeCO3. The precipitation rate is controlled by temperature and pH. It could not be established if the scales accounted for any protection of the casing steel. However, the cement showed no loss of mass and the degradation was restricted to alteration of the cement until depths of 50-250 μm after 394 hours of exposure. Penetration is consistent with 1-D diffusion of CO2 from the fluid into the cement, showing a fairly low diffusion coefficient based on the maximum observed penetration depth (Table 4.1; Carey et al., 2008). Based on the apparent sensitivity of casing to corrosion in a composite sample, Carey et al. (2008) suggest applying relatively simple corrosion logging techniques as an indicator of flow of corrosive CO2 along the casing-cement interface.

4.5.2 Interaction of chemical, mechanical and physical processes

The presence of water or wet supercritical CO2 has a high impact on the mechanical strength of cement, significantly reducing its strength. However the observed difference between effects of water and wet supercritical CO2 is limited (Liteanu et al., 2008). As mentioned above, under certain conditions healing of fractures or annuli could occur by precipitation of for instance calcium or iron carbonates. This was observed to an extent e.g. in field samples from SACROC (Carey et al., 2007). Huerta et al. (2008) performed an experimental study to evaluate the effects of cement degradation along conduits and confining pressures on leakage rates. They show that acid attack did not substantially change the relationship between fracture aperture and confining stress for one of the samples, due to limited reactivity. However, another sample showed significant alteration of the sample’s mechanical properties during acid treatment. Mechanical weakening results in a considerable more rapid closure of the fracture with increasing confining stress, showing that leakage pathways could heal under influence of CO2-rich fluids at decreasing fluid pressures (Huerta et al., 2008). Experiments by Lécolier et al. (2008) confirm self-healing behaviour at the cement-casing interface based on one month of exposure to a CO2-saturated brine at 80°C and 7 MPa. During the experiment the permeability of the composite sample decreased by three orders of magnitude showing continuously decreasing flow rates. This effect is attributed to the formation and precipitation of carbonation products (Lécolier et al., 2008).

From the above it is evident that the evaluation of coupled processes and interacting materials is crucial for understanding of the long-term behaviour of the well system. While studies are increasing their focus on these fields, research directed at the integral behaviour of the system under influence of CO2 will remain imperative.

1 C = constant factor in Fick’s Second Law describing diffusion: d = C · t ½, where d in [mm]; t in [h]. Note that these values are estimated from results on CO2 penetration depth versus time: these C-values are not given in the references (except for Barlèt-Gouédard et al., 2006).

2 Experiment performed on composite sample of steel and cement with CO2-brine flow along the annulus. Diffusion coefficient is based on the assumption of 1-D diffusion perpendicular to the annular flowpath.

3 140 bar pore pressure, 280 bar confining pressure

4 Presented diffusion coefficients (C) from Duguid (2008) reflect reduced degradation rates observed after 2 to 3 months.

5 C = constant factor in Fick’s Second Law describing diffusion: d = C . t ½ , where d in [mm]; t in [h]. Note that these values are estimated from results on CO2 penetration depth versus time: these C-values are not given in the references (except for Barlèt-Gouédard et al., 2006).

6 Fick’s Law does not fit penetration results obtained by Kutchko et al. (2008) for cement degradation by CO2-saturated brine. Kutchko et al. (2008) successfully matched an Elovich equation of the form d = 0.0901. ln(t/24) + 0.1708 where d in [mm]; t in [h].