spe-132228-pa-p cementing solutions for corrosive well environments_uu diem slag, chung minh no...

12
208 June 2011 SPE Drilling & Completion Cementing Solutions for Corrosive Well Environments A. Brandl, J. Cutler, A. Seholm, M. Sansil, and G. Braun, Baker Hughes Incorporated Copyright © 2011 Society of Petroleum Engineers This paper (SPE 132228) was accepted for presentation at the International Oil and Gas Conference and Exhibition in China, Beijing China, 8–10 June 2010, and revised for publication. Original manuscript received for review 23 March 2010. Revised manuscript received for review 8 October 2010. Paper peer approved 13 October 2010. Summary Main factors that are responsible for the corrosion of the cement sheath in wells are determined on the basis of laboratory tests and field analyses from a literature survey. A description of the chem- istry, mineralogy, and physical properties of American Petroleum Institute (API) well cements and their mechanisms of corrosion in the presence of aggressive formation and injection fluids [such as magnesium- or sulfate-containing brines and carbon dioxide (CO 2 )] is given. API cement hydration mainly produces calcium- silicate-hydrate (C-S-H) phases (xCaO·ySiO 2 ·zH 2 O), which are responsible for the strength, and portlandite [Ca(OH) 2 ], which is basically a weak point within the cement matrix. Increasing the permeability and the portlandite content reduces strength and chemical resistance of set cement toward corrosive media. The addition of pozzolanic materials eliminates portlandite and allows lowering the water content in the cement system. Both effects can reduce the permeability and improve the mechanical properties of set cement. Cement specimens were prepared and exposed to CO 2 -loaded water at 300°F and 3,000 psi for 6 months. Mechani- cal-properties tests, microscopy, and quantitative CaCO 3 analyses revealed significantly less corrosion and fewer negative impacts for a pozzolan/API cement blend compared with a conventional API cement design at the same density. These findings were related to the effect of the used pozzolans on the formed C-S-H. Practical concepts for improvements of the cement sheath with respect to cement-slurry design, cementing process, and the impact of factors such as temperature or cement admixtures are presented to mitigate cement corrosion. Introduction Well cementing is a fundamental and essential part of the process of drilling a well. The primary goals of the annular cement sheath in wells are zonal isolation, supporting the casings, and protecting the casings from corrosion. The cement sheath must not only with- stand downhole stresses induced by pressure and temperature fluc- tuations, but also corrosive attacks from aggressive formation and injection fluids (such as in CO 2 flooding for enhanced oil recovery or in CO 2 capture and storage). Cement-sheath failures leading to loss of zonal isolation are the biggest concerns because they affect the wellbore integrity and so the life of the well, with economic consequences: decline in production rates; loss of production time for remedial cementing; and in the worse case, even complete well failure/collapse, so that well abandonment necessary. These problems create an imperative to design cement systems that counteract all negative impacts on the sheath integrity during the life of a well to ensure maximum durability. Synthetic and epoxy-resin-based binders provide high chemical resistance and good mechanical properties in the well (Cole 1979). Nevertheless, their high costs limit their use to special applications. Calcium aluminate- or phosphate-based cements also proved to be more highly corrosion resistant than Portland cements (Sugama 2006). But these cements are sensitive to contamination with Portland cements and so must be handled separately, requiring advanced planning and expensive logistics. Portland-cement-based systems are still preferred because of their economical and practical advan- tages as well as their ready availability. The objective of this study was to find cementing solutions for corrosive well environments using Portland cement-based systems. The following steps were executed to approach this goal: 1. Understand the mineralogy, chemistry, and physical proper- ties of set API cements. 2. Study the mechanisms of different corrosion types for API cements and potential solutions based on a literature survey. 3. Test and evaluate a designed pozzolan API cement system in comparison with a conventional system for a corrosive well scenario (CO 2 , bottomhole static temperature of 300°F, 3,000-psi pressure) with occurring mechanical downhole stresses. 4. Compose cementing guidelines for best practices in corrosive well environments. Mineralogy and Properties of Set API Cements The API well cements (A–H) are Portland cements manufactured according to API Specification 10A. Today, the universal cements API Class G or H are normally used for well cementing. Dur- ing cement hydration, essentially C-S-H phases and portlandite are formed (Eqs. 1a and 1b) because 3CaO·SiO 2 and 2CaO·SiO 2 make together approximately 85 wt% in API Class G/H clinker. Hydration of 3CaO·Al 2 O 3 and 2CaO·(Al 2 O 3 ,Fe 2 O 3 ), in the presence of a setting control agent (e.g., gypsum CaSO 4 ·2H 2 O), results in [Ca 4 Al 2 (OH) 12 ](SO 4 )·6H 2 O monosulfate or [Ca 4 (Al,Fe) 2 (OH) 12 ] (SO 4 )·6H 2 O monosulfo-aluminate (-ferrite), respectively (Eqs. 1c and 1d). 3CaO·SiO 2 + xH 2 O yCaO·SiO 2 ·[y–(3–x)]H 2 O (= C-S-H) + (3–y) Ca(OH) 2 (= portlandite), . . . . . . . . (1a) 2CaO·SiO 2 + xH 2 O yCaO·SiO 2 ·[y–(2–x)]H 2 O (= C-S-H) + (2-y) Ca(OH) 2 (= portlandite), . . . . . . . . (1b) 3CaO·Al 2 O 3 + CaSO 4 ·2H 2 O + 10H 2 O [Ca 4 Al 2 (OH) 12 ](SO 4 )·6H 2 O, . . . . . . . . . . . . . . . . . . (1c) 3[2CaO·(Al 2 O 3 ,Fe 2 O 3 )] + 2CaSO 4 ·2H 2 O + 23H 2 O 2([Ca 4 (Al,Fe) 2 (OH) 12 ](SO 4 )·6H 2 O) + 2(Al,Fe)(OH) 3 . . . . . . . . . . . . . . . . . . . . . . . . (1d) C-S-H phases occur in fine needles—during cement hydration they grow in length and grab onto each other like a zipper, resulting in high strength. Portlandite, on the other hand, consists of large hexagonal crystals between the C-S-H phases. The amount of formed portlandite in a set, pure Portland cement depends on the degree of hydration, temperature, exact clinker-phase composition, water-to-cement ratio, and composition of formed C-S-H phases, and it can be up to 35 wt% (Milestone et al. 1986). Portlandite does not contribute to the strength and so is a weak point in the cement matrix. On a microscopic level, even a set Portland cement is still a porous, water-filled system, into which corrosive fluids can penetrate and cause deterioration. Increasing the amount of mix water increases the percentage of capillary pores in the hardened cement. Capillary pores are mainly responsible for the perme- ability. Increasing permeability reduces a cement’s strength and resistance to corrosive media.

Upload: johnsmith

Post on 22-Dec-2015

16 views

Category:

Documents


0 download

DESCRIPTION

PES P

TRANSCRIPT

Page 1: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

208 June 2011 SPE Drilling & Completion

Cementing Solutions for Corrosive Well Environments

A. Brandl, J. Cutler, A. Seholm, M. Sansil, and G. Braun, Baker Hughes Incorporated

Copyright © 2011 Society of Petroleum Engineers

This paper (SPE 132228) was accepted for presentation at the International Oil and Gas Conference and Exhibition in China, Beijing China, 8–10 June 2010, and revised for publication. Original manuscript received for review 23 March 2010. Revised manuscript received for review 8 October 2010. Paper peer approved 13 October 2010.

SummaryMain factors that are responsible for the corrosion of the cement sheath in wells are determined on the basis of laboratory tests and field analyses from a literature survey. A description of the chem-istry, mineralogy, and physical properties of American Petroleum Institute (API) well cements and their mechanisms of corrosion in the presence of aggressive formation and injection fluids [such as magnesium- or sulfate-containing brines and carbon dioxide (CO2)] is given. API cement hydration mainly produces calcium-silicate-hydrate (C-S-H) phases (xCaO·ySiO2·zH2O), which are responsible for the strength, and portlandite [Ca(OH)2], which is basically a weak point within the cement matrix. Increasing the permeability and the portlandite content reduces strength and chemical resistance of set cement toward corrosive media. The addition of pozzolanic materials eliminates portlandite and allows lowering the water content in the cement system. Both effects can reduce the permeability and improve the mechanical properties of set cement. Cement specimens were prepared and exposed to CO2-loaded water at 300°F and 3,000 psi for 6 months. Mechani-cal-properties tests, microscopy, and quantitative CaCO3 analyses revealed significantly less corrosion and fewer negative impacts for a pozzolan/API cement blend compared with a conventional API cement design at the same density. These findings were related to the effect of the used pozzolans on the formed C-S-H. Practical concepts for improvements of the cement sheath with respect to cement-slurry design, cementing process, and the impact of factors such as temperature or cement admixtures are presented to mitigate cement corrosion.

IntroductionWell cementing is a fundamental and essential part of the process of drilling a well. The primary goals of the annular cement sheath in wells are zonal isolation, supporting the casings, and protecting the casings from corrosion. The cement sheath must not only with-stand downhole stresses induced by pressure and temperature fluc-tuations, but also corrosive attacks from aggressive formation and injection fluids (such as in CO2 flooding for enhanced oil recovery or in CO2 capture and storage). Cement-sheath failures leading to loss of zonal isolation are the biggest concerns because they affect the wellbore integrity and so the life of the well, with economic consequences: decline in production rates; loss of production time for remedial cementing; and in the worse case, even complete well failure/collapse, so that well abandonment necessary.

These problems create an imperative to design cement systems that counteract all negative impacts on the sheath integrity during the life of a well to ensure maximum durability. Synthetic and epoxy-resin-based binders provide high chemical resistance and good mechanical properties in the well (Cole 1979). Nevertheless, their high costs limit their use to special applications. Calcium aluminate- or phosphate-based cements also proved to be more highly corrosion resistant than Portland cements (Sugama 2006). But these cements are sensitive to contamination with Portland cements and so must be handled separately, requiring advanced

planning and expensive logistics. Portland-cement-based systems are still preferred because of their economical and practical advan-tages as well as their ready availability. The objective of this study was to find cementing solutions for corrosive well environments using Portland cement-based systems. The following steps were executed to approach this goal:

1. Understand the mineralogy, chemistry, and physical proper-ties of set API cements.

2. Study the mechanisms of different corrosion types for API cements and potential solutions based on a literature survey.

3. Test and evaluate a designed pozzolan API cement system in comparison with a conventional system for a corrosive well scenario (CO2, bottomhole static temperature of 300°F, 3,000-psi pressure) with occurring mechanical downhole stresses.

4. Compose cementing guidelines for best practices in corrosive well environments.

Mineralogy and Properties of Set API CementsThe API well cements (A–H) are Portland cements manufactured according to API Specification 10A. Today, the universal cements API Class G or H are normally used for well cementing. Dur-ing cement hydration, essentially C-S-H phases and portlandite are formed (Eqs. 1a and 1b) because 3CaO·SiO2 and 2CaO·SiO2 make together approximately 85 wt% in API Class G/H clinker. Hydration of 3CaO·Al2O3 and 2CaO·(Al2O3,Fe2O3), in the presence of a setting control agent (e.g., gypsum CaSO4·2H2O), results in [Ca4Al2(OH)12](SO4)·6H2O monosulfate or [Ca4(Al,Fe)2(OH)12](SO4)·6H2O monosulfo-aluminate (-ferrite), respectively (Eqs. 1c and 1d).

3CaO·SiO2 + xH2O → yCaO·SiO2·[y–(3–x)]H2O(= C-S-H) + (3–y) Ca(OH)2 (= portlandite), . . . . . . . . (1a)

2CaO·SiO2 + xH2O → yCaO·SiO2·[y–(2–x)]H2O(= C-S-H) + (2-y) Ca(OH)2 (= portlandite), . . . . . . . . (1b)

3CaO·Al2O3 + CaSO4·2H2O + 10H2O → [Ca4Al2(OH)12](SO4)·6H2O, . . . . . . . . . . . . . . . . . . (1c)

3[2CaO·(Al2O3,Fe2O3)] + 2CaSO4·2H2O + 23H2O → 2([Ca4(Al,Fe)2(OH)12](SO4)·6H2O) + 2(Al,Fe)(OH)3.

. . . . . . . . . . . . . . . . . . . . . . . (1d)

C-S-H phases occur in fine needles—during cement hydration they grow in length and grab onto each other like a zipper, resultingin high strength. Portlandite, on the other hand, consists of large hexagonal crystals between the C-S-H phases. The amount of formed portlandite in a set, pure Portland cement depends on the degree of hydration, temperature, exact clinker-phase composition, water-to-cement ratio, and composition of formed C-S-H phases, and it can be up to 35 wt% (Milestone et al. 1986). Portlandite does not contribute to the strength and so is a weak point in the cement matrix. On a microscopic level, even a set Portland cement is still a porous, water-filled system, into which corrosive fluids can penetrate and cause deterioration. Increasing the amount of mix water increases the percentage of capillary pores in the hardened cement. Capillary pores are mainly responsible for the perme-ability. Increasing permeability reduces a cement’s strength and resistance to corrosive media.

HNgoc6
Highlight
HNgoc6
Highlight
Page 2: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

June 2011 SPE Drilling & Completion 209

Effect of Pozzolanic Materials. An effi cient way to reduce the amount of portlandite, strengthen the matrix, and decrease the perme-ability of set cement is to add fi ne pozzolanic materials (e.g., diato-maceous earth, volcanic ash, silica fume, metakaolin, calcined clay, and/or fl y ash) to the cement. The SiO2 (and Al2O3) constituents of the pozzolanic additions react with portlandite and convert it to further C-S-H phases (or C-A-S-H phases: xCaO·wAl2O3·ySiO2·zH2O, with A standing for Al) according to the pozzolanic reaction:

xCa(OH)2 + ySiO2 (and wAl2O3) → xCaO·ySiO2·zH2O(and xCaO·wAl2O3·ySiO2·zH2O). . . . . . . . . . . . . . . . . . (2)

Pozzolanic materials have typically been added to API cements as extenders to design filler slurries. Pozzolanic materials generally have a lower specific gravity (s.g. of 2.0–3.0 g/cm3) than cement (s.g. of 3.2 g/cm3). Consequently, at a constant water/solid ratio, the pozzolanic additions will reduce the density of the cement slurry, or, at a constant slurry density, the pozzolan additions will reduce the water/solid ratio in the cement slurry. This will decrease the permeability of the set cement significantly (Nelson and Guillout 2007). Finally, the partial substitution for cement by relatively chemically inert pozzolanic materials essentially contributes to further improvement of corrosion resistance.

Cement Corrosion Because of High Temperatures At elevated temperatures, a number of thermodynamically stable C-S-H phases occur, depending on the formal molar ratio of CaO to SiO2 in the Portland cement (Taylor 1964). The formal molar CaO/SiO2 ratio for a neat cement is approximately three, resulting in the formation of conventional C-S-H phases. Above a tempera-ture of 230°F, these C-S-H phases will transform into a thermo-dynamically more stable �-Ca2[SiO3(OH)](OH) (�-C2SH) phase with higher crystalline density, reducing compressive strength and increasing permeability of set cement (strength retrogression). This conversion reaction can be prevented by adding 35–40% by weight of cement (BWOC) of SiO2 to the cement (silica-stabilized cement). This reduces the formal molar CaO/SiO2 ratio to approxi-mately unity, and the more-stable tobermorite phase is formed at temperatures of approximately 230°F, preserving high com-pressive strength and low permeability in the set cement. Above 302°F, tobermorite converts into xonotlite and/or gyrolite. Both of these higher-temperature phases result in improved well-cementing qualities (low permeability and high compressive strength).

Cement Corrosion Caused by Poor Cement Quality or Incom-patible Additions. Insuffi cient conversion of CaO or MgO during the cement-manufacturing process results in high free-lime (CaO) or periclase (free-MgO) content of the clinker. Hydration of free lime or periclase after cement setting can cause destructive expan-sion within the set cement (Odler 1998):

CaO, MgO + H2O → Ca(OH)2, Mg(OH)2 => expansive . . . . . . . . . . . . . . . . . . . . . . . (2)

Therefore, the free-lime and periclase contents must be limited to 2 wt% or 6 wt%, respectively, to mitigate expansion issues. Increasing the alkali content increases the hydroxide concentration in the pores of set cement. The combination of reactive (amor-phous) silica aggregates (such as chert, quartzite, opal, strained quartz crystals, and glass) and high concentrations of hydroxides results in a gel (alkali-silica reaction), increasing in volume by taking up water and so exerting an expansive pressure:

NaOH, KOH + SiO2 (amorphous) + water

→ alkali-silica-gel => expansive. . . . . . . . . . . . . . . . . . (4)

The results are cracks throughout the cement matrix. Therefore, the total alkali content must be limited to 0.75 wt% of Na2O equivalents, as required for API cements. Another way to reduce

the likelihood of alkali-silica reaction is to decrease water content or use pozzolans as a partial cement replacement in the system; the latter also reduces the alkalinity of the pore fluid.

Cement Corrosion Caused by Expansive Attack. Generally, there are two categories of cement corrosion induced by chemical attacks: that by expansive attack and that by dissolving attack. During the expansive (or expanding) attack, the corrosive fl uid penetrates the cement pores and forms voluminous water-insoluble products. As their crystal size increases, these products create high pressure inside the set cement, resulting in cracks, fractures, and fragments. The most well-known expansive attack is the sulfate attack. Sulfate-containing formation fl uids penetrate into the pores of the cement sheath and react with the 3CaO·Al2O3 phase of the cement and its hydration products to form secondary or delayed ettringite crystals, which fi ll the cement pores (Morales at al. 2003):

3CaO·Al2O3 + 3(CaSO4·2H2O) + 26H2O → [Ca6Al2(OH)12](SO4)3·26H2O ettringite, . . . . . . . . . . (5)

3CaO·Al2O3·6H2O + 3(CaSO4·2H2O) + 20H2O → [Ca6Al2(OH)12](SO4)3·26H2O ettringite, . . . . . . . . . . (6)

3CaO·Al2O3·CaSO4·12H2O + 2(CaSO4·2H2O) + 16H2O → [Ca6Al2(OH)12](SO4)3·26H2O ettringite. . . . . . . . . . . (7)

These formed ettringite crystals can reach sizes up to 50 µm, caus-ing high internal pressures in the cement pores until fractures of the cement sheath occur. Besides ettringite, sulfate can also react with portlandite of the set cement to form secondary (or delayed) gypsum crystals:

Ca(OH)2 + Na2SO4 + 2H2O → CaSO4·2H2O secondary gypsum + 2NaOH. . . . . . . (8)

Bellmann and Stark (2006) found that cement deterioration caused by the formation of secondary ettringite and gypsum at ambient conditions occurs only above a sulfate concentration of 1500 mg/L or 3000 mg/L, respectively. Because the water solubility of sodium and magnesium sulfates is low at temperatures above 140°F, sul-fate attack is not a big issue at this temperature or higher (Suman and Ellis 1977). For temperatures below 140°F, it is essential to use high sulfate-resistance cements (3CaO·Al2O3 phase content below 3 wt% for API cements) and to eliminate portlandite in the set cement. Morales et al. (2003) discovered that decreasing the permeability is an even more efficient way to avoid sulfate attack. Using pore-blocking additives (such as latex, sodium silicate, or microsilica) or a lower water/cement ratio will improve the sulfate resistance of an API Class A cement.

Magnesium attack is another example of expansive attack. Magnesium-containing formation fluids precipitate under high pH conditions in the pores of set cement or react with the portlandite to form brucite [Mg(OH)2] by ion base exchange:

Ca(OH)2 + Mg2+ → Ca2+ + Mg(OH)2 => expansive. . . . . . . . . . . . . . . . . . . . . . . . . (9)

The formation of expansive Mg(OH)2 induces mechanical stresses in the set cement, resulting in destruction. In addition, calcium ions in the C-S-H phase lattice are exchanged with magnesium ions, resulting in strength retrogression and ultimately complete deterio-ration, as reported by Steinmann (1972). Wollherr (1999) tested the chemical resistance of a conventional API Class G cement com-pared to a blast-furnace-slag/pozzolan/Portland-cement system (high-magnesium-resistance cement) in a high-MgCl2 (200-g/L) brine at 257°F. Only the high-magnesium-resistance cement still possessed integrity after 6 months; the conventional API Class G cement system showed severe expansive deterioration and loss of integrity.

HNgoc6
Highlight
Page 3: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

210 June 2011 SPE Drilling & Completion

Cement Corrosion Caused by Dissolving Attack. When cement exposure to corrosive fl uids creates water-soluble products, the set cement is leached from the surface (dissolving attack). During reservoir stimulation, the cement sheath (in particular around the perforation zone) may be exposed to strong acids [hydrochloric acid (HCl) and hydrofl uoric acid (HF)]. Lécolier et al. (2010) estimated that 40% of all gas reservoirs contain at least 100 ppm of sour gas (H2S), a relatively weak acid. Conventional set API cements do not withstand acidic conditions and will dissolve over time with decreasing pH. The portlandite becomes unstable at a pH below 12.6 (Grabau 1994) and so will leach out fi rst:

2H+ + Ca(OH)2 → Ca2+ + 2H2O. . . . . . . . . . . . . . . . . . . . (10)

Below a pH of 8, the strength-giving C-S-H phases (e.g., 3CaO·2SiO2·3H2O) are also destabilized, followed by leaching of Ca2+ and formation of amorphous silica (SiO2):

6H+ + 3CaO·2SiO2·3H2O → 3Ca2+ + 2SiO2 + 6H2O. . . . . . . . . . . . . . . . . . . . . . . . (11)

The remaining amorphous silica forms a protective coating that slows the acid reaction rate. The rate of deterioration is accelerated if the amorphous silica is washed away (e.g., dynamic conditions). As reported by Brady et al. (1989), conventional Class G cement showed a significant weight loss (96%) within a short time when exposed to 12–3% HF/HCl (HCl-/HF-containing mud) at 190°F under dynamic conditions. Mud-acid systems are typically used to stimulate sandstone reservoirs. According to Blount et al. (1991), the resistance of Portland cement toward acid attacks can be improved by 700% when substituting for the cement with greater amounts of American Society of Testing Materials (ASTM) Class F fly ash in conjunction with the use of liquid-latex dispersion. The fly ash eliminates acid-susceptible portlandite, reduces the perme-ability of the set cement, and partly replaces Portland cement with relatively chemically inert siliceous materials. Adding latex will form an artificial barrier between the cementitious phases and the attacking fluid. Adding ASTM Class F fly ash or the liquid latex alone was found to improve the mud-acid resistance by 48% or >85%, respectively.

Al-Taq et al. (2009) recently explained that pretreating a neat cement with a mixture of 10% acetic acid + 1.5% HF leads to a protective layer of CaF2, which significantly improves resistance to mud-acid attack. The weight loss for pretreated neat cement in mud acid at 150°F after 2 hours was only 2 wt% compared to 32 wt% for untreated cement. Therefore, it was recommended to start stimulation of sandstone reservoirs with a preflush of 10% acetic acid + 1.5% HF. This is a simple and cost-effective procedure to protect conventional cement between the casing and the formation from severe acid corrosion.

The corrosive H2S attack on Portland cement under well conditions was studied by Lécolier et al. (2010). A low-perme-ability Portland cement containing silica fume was exposed to H2S-saturated fluid (pH = 4) at 248°F and 2,176 psi for 21 days. Within a corrosion depth of 11 mm, all portlandite was completely removed and calcium from the C-S-H phases was partly leached out. It was concluded that alternative cements must be used to guarantee long-term integrity for H2S-affected injection wells. Noïk and Rivereau (1999) designed a “reactive powder cement” composed of API Class G cement with sand and silica fume. The system mixed at 15.8 lbm/gal (ppg) and had a low water/cement ratio of 0.27, resulting in low water permeability of less than 10–6 md after curing at 356°F for 3 months. Reactive-powder-cement samples were exposed to seawater in the presence of 10% H2S gas at 248°F. After 1 year of aging, the compressive strength actually increased by 30%, which proved durability of the system in H2S under the given conditions.

Cement Corrosion in the Presence of CO2. Lécolier et al. (2010) also reported that 40% of the world’s remaining gas reserves naturally contain more than 2 vol% of CO2. Some gas reserves contain more than 50 vol% of CO2. Moderate concentrations of CO2 (5–30 vol%)

are most common in gas wells. Furthermore, supercritical CO2 is used for enhanced oil recovery during CO2 fl ooding. In recent years, CO2-sequestration/-injector wells have come to have great impor-tance in reducing CO2 emissions and so the effect of greenhouse gas. Onan (1984) was the fi rst to test CO2 attack (under well condi-tions at elevated temperatures and pressure) and concluded that the presence of CO2 in wells could be a serious problem.

As part of the present study, CO2 attack was simulated by exposing a 2-in.-long, 1-in.-diameter cylinder of set cement (Class G + 40% BWOC silica flour mixed at 15.8 ppg) to CO2-loaded water at 250°F and 650-psi CO2 pressure for 30 days. A thin sec-tion was prepared from a cylinder half and was analyzed under the optical microscope (Fig. 1). Evaluation of the photomicrograph confirms the findings published earlier by Barlet-Gouédard et al. (2006) and Kutchko et al. (2007, 2008).

Cement corrosion in the presence of CO2 is a special case of dissolving attack because it consists of three sequential steps: First, (Step 1 in Fig. 1c) is the formation of carbonic acid in the presence of moisture (Eq. 12). The solubility of CO2 gas in water is (with 1.45 g/L under ambient conditions) relatively low, and only 0.2% of the dissolved CO2 reacts with water to form carbonic acid and its dissociated form. This explains that CO2 or carbonic acid, respec-tively, is a very weak acid compared to HCl or mud acid. Even at high partial pressure, the pH value will not drop below 3.5.

The second step (Step 2 in Fig. 1c) is the carbonation of the cement components: The carbonic acid penetrates the cement matrix and preferentially reacts with the portlandite, converting it into calcium carbonate and water (Eq. 13). The conversion of portlandite to calcium carbonate reduces the pH from 13 to less than 10. Carbonic acid also reacts with the C-S-H phases to form CaCO3 and amorphous (porous) silica (Eq. 14). Carbonation of portlandite is believed to be faster than that for the C-S-H phases (Santra et al. 2009). Because calcium carbonate has a higher molar volume than portlandite (+11%) (Stark and Wicht 2001), the total pore volume is reduced and so the permeability of the cement matrix is initially decreased by carbonation. Carbonation of the cement also increases its mass and its compressive and tensile strength. However, the amorphous silica gel does not contribute to strength like the CaCO3 (calcite).

The final step (Step 3 in Fig. 1c) is the leaching (Eq. 15) and deposition process (Eq. 16), which is actually the detrimental reac-tion: Further carbonic acid reacts with the formed calcium carbon-ate and converts it into calcium bicarbonate, which is highly water soluble and can be leached out of the cement matrix easily. All reactions are thermodynamically driven, but their kinetics depend on the given conditions (e.g., temperature, pressure, concentrations of educts and products within the reaction equilibria).

In summary, then, cement corrosion caused by CO2 attack can be described with the following chemical equations.

Formation of carbonic acid:

CO2 + H2O ⇌ H2CO3 ⇌ H+ + HCO3–. . . . . . . . . . . . . . . . (12)

Carbonation of the cementitious phases:

Ca(OH)2 + H+ + HCO3– ⇌ CaCO3 + 2H2O, . . . . . . . . . . . (13)

xCaO·ySiO2·zH2O (= C-S-H ) + x H+ + x HCO3–

⇌ xCaCO3 + y SiO2 (amorphous and porous silica) + (z+x) H2O. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (14)

Leaching out and deposition, respectively, of calcium carbonates:

CO2 + H2O + CaCO3 ⇌ Ca(HCO3)2, . . . . . . . . . . . . . . . . (15)

Ca(HCO3)2 + Ca(OH)2 ⇌ CaCO3 + H2O. . . . . . . . . . . . . (16)

In the Scurry Area Canyon Reef Operators Committee field study by Carey et al. (2007), a cement section with casing and

HNgoc6
Highlight
Page 4: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

June 2011 SPE Drilling & Completion 211

formation right above the perforation zone from a 55-year-old well was cored and analyzed. The original pumped system was a neat Portland cement with a density of approximately 15.5 ppg. After 25 years of production, CO2 was injected for enhanced oil recovery for 30 years. The recovered cement samples showed a corrosion depth of 2 to 10 mm. This suggests that cement corrosion in the presence of CO2 is a very slow process. It was concluded that “the cement retained its capacity to prevent significant transport of fluid through the cement matrix.” Recovered field samples from this well indicated that degradation occurs mainly along existing or induced migration pathways. Consequently, the cement’s abil-ity to resist CO2 attack seems to be secondary: More important is the ability to obtain a good initial bond with the casing and formation.

Crow et al. (2009) published a similar detailed field study (CCP—CO2 Capture Project Phase 2) concerning a 30-year-old well from a natural-CO2-production reservoir (96% CO2). Infor-mation about cement and casing samples recovered from different side walls and reservoir-fluid and -pressure measurements were collected and analyzed. The well was cemented 30 years ago with a system of 50:50 fly ash/API Class H cement including 3% benton-ite mixed at 14.2 ppg (extended pozzolan API cement system). The temperature was up to 140°F, and the pressure was approximately 1,500 psi. Mineralogy of the cement cores revealed maximum CaCO3 content near the CO2 reservoir and original cement phases more abundant further from the reservoir. This pozzolan extended API cement system inhibited CO2 migration, although carbonation occurred and permeability and porosity increased. No casing pressure was observed during the life of the well, confirming no leakage. Furthermore, evaluation-log information was presented in detail, indicating generally good cement-bond quality. Con-sequently, Portland-cement-based systems can be used in a CO2 environment if good cementing practices are employed.

Milestone et al. (1986) investigated the effect of silica addi-tions to Portland-cement systems after 14 days of exposure to CO2 at 302°F. They found that increasing additions of silica will reduce the portlandite content and increase the carbonation depth in the set cement system. Any portlandite present will react with CO2 to form a protective layer of calcite, which decreases perme-ability and inhibits further attack. Therefore, the authors suggested

that the amount of added silica be reduced from 35–40% to 15% BWOC to retain a certain amount of portlandite in the set cement to improve CO2 resistance. The evaluation of the recommended design regarding water permeability and long-term integrity was not presented. Consequently, it was not proved that the reduced silica content is sufficient to prevent strength retrogression or that the actual detrimental leaching step caused by further CO2 attack was mitigated.

For the current study, a pozzolan-containing API cement sys-tem, with sufficient silica additions to prevent strength retrogres-sion, was designed at a density of 15.0 ppg without bentonite but with reduced water content. The goal of the study was to evaluate the effect of CO2 on the integrity of the pozzolan cement system at 300°F and 3,000 psi over a period of 6 months.

Testing and Evaluation Cement Designs and Chemical Analyses. For the CO2 tests a pozzolan/API-Class-G/silica-fl our system mixed with seawater at a density of 15.0 ppg was used (pozzolan) (Table 1). This system was compared directly to a “conventional” cement system (API Class G with 35% BWOC silica fl our) at the same density. Less mixing water was required for the pozzolan system. The same chemical admixture package was added to both systems to stabilize the slurry (no free water, no channeling, no settling) and to achieve static API fl uid-loss control below 100 cm3/30 min at 300°F and thickening times of 7 hours at bottomhole circulating temperature of 260°F and 3,000 psi. Slurries were mixed accord-ing to API RP 10B procedures, and cement specimens [2-in.-long, 1-in.-diameter cylinders and briquettes (“dog bones”) according to ASTM-307] were cured at 3,000 psi and 300°F for 96 hours. After recovering from the curing chamber, all grease was properly removed from the specimens by intensive washing with water and soap. Specimens showing cracks or irregularities on their surfaces were discarded. Analyses of both cured systems revealed almost identical composition (Table 1). For both systems, the formal CaO content was 47 wt% (±1). Suffi cient silica was present in both cement systems to reduce the formal molar CaO/SiO2 ratio to approximately 1.1 to prevent strength retrogression. According to X-ray powder diffraction analyses, practically no portlandite was found in either system.

Fig. 1—(a) Set-cement cylinder (Class G+40% BWOC silica flour mixed at 15.8 ppg) after 30-day exposure to CO2-loaded water at 250°F/650-psi CO2 pressure. (b) Cylinder was cut in half, and a thin section was prepared. (c) Thin-section petrography: Unaltered set cement (I ) still exists having C-S-H phases and portlandite. On the far right side is the corrosive fluid (V ) with the CO2 that has reacted with water to form carbonic acid (Step 1). The carbonated cement and its propagation front (Step 2) can be clearly seen in the middle part (III ). The leaching of the carbonated cement with further attack of carbonic acid (Step 3) has already taken place, resulting in a soft and porous rim (IV) dark black zone at the rim because of stronger penetration of the blue dye is indicating higher porosity).

Page 5: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

212 June 2011 SPE Drilling & Completion

CO2 Tests. To simulate the CO2 or carbonic acid attack within the cement specimens under static conditions, a modifi ed high-temperature/high-pressure (HT/HP) curing chamber (Model 4000, Nordman Instrument Incorporated, Houston, Texas) was used and was attached to a CO2 bottle (Fig. 2a). The rack with the cement specimens (Fig. 2c) was released into the 5.1-L interior of the curing chamber (Fig. 2b). The chamber was then completely fi lled with tap water, and CO2 gas was injected constantly from the bottom of the chamber with the help of a high-pressure booster for 5 minutes to generate CO2-loaded water (measured pH = 5). The curing chamber was pressurized with CO2 gas to 3,000 psi and heated to 300°F within 4 hours. Any pressure reduction was automatically compen-sated with further CO2 injection through pump cycles, guaranteeing a continuous supply of CO2.. After 1, 3, and 6 months of exposure, the pressure and temperature in the HT/HP curing chamber were gradually decreased over 12 hours and a set of cement specimens was recovered for mechanical and mineralogical analyses. Because of the increase of the pH to 6, the CO2-loaded water was refreshed each time samples were taken. The initial volume ratio between the corrosive fl uid and all cement specimens was calculated to be 4:1, but then was increased to 5:1 and fi nally 6:1 because some samples were recovered after 1 and 3 months.

Testing and Evaluation Methodology. Thin-section petrography. Each cement cylinder was cut in half, and one of the halves was impregnated with blue epoxy resin under vacuum to enhance the display of porosity. The sample was then attached to a glass slide and ground to a fi nal 30-µm thickness.

Scanning Electron Microscopy (SEM). All cement specimens were prepared such that a freshly broken surface was presented to the electron beam. For low magnifications of 1,000X or less, samples were not coated and were viewed using the “wet mode” (gaseous secondary electron detector—1.0-torr H2O pressure) of the SEM. For higher magnification (5,000X), cement specimens were dried at low temperature and subsequently sputter coated with gold under vacuum. A secondary electron detector was used for the high-magnification imaging.

Chemical Analysis. An energy-dispersive-spectrometry (EDS) system from IXRF System Incorporated was used for all analyses. The excitation voltage on the SEM was 20 kV.

Young’s Modulus, Poisson’s Ratio, and Compressive Strength. These were determined from uniaxial and triaxial stress/strain tests performed on cement cylinders under a confining pressure of 1,000 psi. Testing procedures and apparatus closely follow ASTM D 7012-07 “Standard Test Method for Compressive Strength and

TABLE 1—CEMENT-SYSTEM DESIGN, PROPERTIES, AND OXIDIC COMPOSITION

Cement System Pozzolan Conventional

Base blend Pozzolan + API Class G + silica flour

API Class G + 35% BWOC of silica flour

Slurry density (lbm/gal)

15.0 15.0

27.0 55.0 )tw/tw( dilos :retaWPortlandite (wt%) Not detected Not detected

Oxidic Composition of Set Sement Systems (Calculated From EDS Analyses)

5.74 6.64 )%tw( OaCSiO2 8.93 8.04 )%tw(

3.2 4.2 )%tw( CNa2 6.0 6.0 )%tw( O

4.0 4.0 )%tw( OgMAl2O3 2.2 4.2 )%tw( SO3 8.2 1.3 )%tw(

6.0 7.0 )%tw( lCK2 2.0 1.0 )%tw( OFe2O3 4.3 9.2 )%tw(

0.001 0.001 )%tw( latoT

Fig. 2—(a) Modified HT/HP curing chamber connected with a CO2 bottle; (b) interior of the chamber; (c) cement-cylinder specimens on the rack after 3-month exposure to CO2-loaded water.

Page 6: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

June 2011 SPE Drilling & Completion 213

Elastic Moduli of Intact Rock Core Specimens under Varying States of Stress and Temperatures” recommended practices.

Tensile Strength. These data were determined from a direct uniaxial tensile-strength method according to ASTM Standard C-307 tensile strength using briquette specimens (dog bones) and a United press model STM-20k.

Microindentation. Brinell hardness (BH) was determined by forcing a hard steel sphere of a specified diameter under a specified load into the surface of a material and measuring the diameter of the indentation left after the test. BH is obtained by dividing the load used, in kilograms, by the actual surface area of the indenta-tion, in square millimeters. The result is a pressure measurement, but the units are rarely stated.

Water Permeability. Cement cylinders were loaded into a preheated (200°F) Hassler-style core holder, and a confining pres-sure of 4,500 psi (1,500-psi net confining pressure) was applied. Using an Isco syringe pump, deionized water was injected at con-stant pressure into one end. Injection pressure was introduced in 500-psi increments to 3,000 psi. Water flow through the sample was monitored by volume change in a pipette.

Quantitative Determination of CaCO3. Cement specimens were dried under vacuum and ambient temperature for 5 days. The amount of lost water during the drying process was taken into account for the quantitative determination of the total CaCO3 content (sum of aragonite and calcite; vaterite was not found). Intensities were acquired using a Phillips XRG 3100 X-ray diffrac-tometer system equipped with a diffracted-beam monochromator and MDI JADE 8 X-ray diffraction pattern processing and iden-tification software. Corundum was used as a standard. Quantities were determined by X-ray powder diffraction using the reference intensity ratio method.

Results and Discussions. Comparison of the Testing Conditions in the Laboratory With the Field Reality. During the laboratory tests, the cement samples were completely soaked and exposed from all sides (except the bottom) to the corrosive fl uid; in the wellbore, the cement sheath is sheltered by the formation and casing (good cementing practices provided) so that only a small area from one side is actually attacked. Also, the ratio of corrosive fl uid to cement and the availability of moisture (necessary to form carbonic acid) were much higher during the laboratory tests than they would be in a well, resulting in a stronger corrosive attack and so simulating worst-case conditions. On the other hand, the laboratory tests were performed under static conditions, whereas in the fi eld, certain scenerios (e.g., CO2 injection or CO2 gas fl ow) are

dynamic and so abrasion could become an important factor. These differences have to be considered when extrapolating conclusions from the laboratory results to the fi eld.

Evaluation. Cement specimens were evaluated after 1-, 3-, and 6-month exposure to CO2-loaded water at 300°F and 3,000 psi. The cement cylinder sets that were exposed for 1 month showed no visual cracks or fractures (Fig. 3a). Cylinders were cut into two pieces along the width. One part of each cut cylinder was used for measuring water permeability and testing microindenta-tion (BH). The other part of the cylinder was sliced into halves along the length to determine the progress of CO2 attack with the help of quantitative CaCO3 measurements. Phenolphthalein tests (pH<10) indicated that no alkalinity was left along the specimens’ profiles for both systems (Fig. 3c), confirming that carbonic acid completely penetrated the cement samples. Only the conventional system (Fig. 3c right) had a distinct brownish rim, indicating stronger alteration. BH could not be measured accurately at this brownish rim because of shattering (Fig. 3b, Number 1), confirm-ing a weaker or brittle zone. To give reliable numbers and to avoid the impact of shattering, BH was tested 1 mm from the specimens’ edges defined as the outer zone (Fig. 3b, Numbers 2 and 3). Every BH test was repeated at least three times on different sites and the results were averaged (shattered data points were not taken into account). The outer zone had a significantly lower hardness (BH = 9) and lower CaCO3 content (38 wt%) than the interior of the conventional cement specimen (BH = 20, CaCO3 = 60 wt%), con-firming a softer zone because of the depletion of CaCO3 according to the leaching mechanism described in Eq. 15. On the other hand, the CaCO3 content (35–37 wt%) and the BH (15–17) between the outer zone and the interior of the pozzolan cement specimen were identical, indicating it was less affected by corrosion after 1 month. (The darker color in the interior of the pozzolan cement specimen is an artifact caused by sawing the cement cylinders.)

Both cement specimens (of the pozzolan and conventional sys-tems) tested for 1, 3, and 6 months retained sufficient low water permeabilities (< 0.01 md) and high compressive strengths >5,000 psi (under a confining pressure of 1,000 psi) to provide zonal isola-tion (Fig. 4, left). Because both systems contained sufficient silica (formal molar CaO/SiO2 ≈1.1; see Table 1), strength retrogression (caused by the high temperature of 300°F) is not of concern. The changes in sizes (length and diameter) for all specimens were marginal, but both systems gained some weight (Fig. 4, right) with hydration and carbonation. The increase in weight (+5% after 1 month to +7% after 6 months) of the pozzolan cement specimens directly correlates with the steady increase in total CaCO3 content

Fig. 3—(a) Cylinders of the pozzolan and conventional cement systems after exposure to CO2-loaded water at 300°F and 3,000 psi for 1 month and sample preparation for evaluation. (b) Testing BH. (c) Profile of the pozzolan (left) and conventional (right) cement systems with their determined BH and CaCO3 content.

Page 7: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

214 June 2011 SPE Drilling & Completion

(35 wt% after 1 month to 40 wt% after 6 months) and confirms that carbonation has probably not been completed even after 6 months. For the conventional cement system, the maximum increase in weight (+8.3 wt%) and maximum carbonation (CaCO3 content = 60 wt%) were reached after 1 month. Afterward, the weight con-tinuously dropped, indicating leaching of cementitious material. After 6 months, the total CaCO3 content was significantly less for the pozzolan (35–40 wt%) than for the conventional cement system (56–60 wt%), even though both systems contained the same formal CaO content from the beginning (Table 1). Obviously, there exist certain CaO or C-S-H phases within the pozzolan system that are less susceptible to carbonation than those formed in the conven-tional system. A possible explanation for this behavior is given in the Effect of the Pozzolan on the Formed C-S-H subsection.

After the cement cylinders from both cement systems were trimmed for testing permeabilities, severe spallings were observed only for the conventional cement system after 6-month exposure to CO2 (Fig. 5a). The deterioration of the conventional system increased when the cement cylinders were put back into tap water because the brownish rim completely spalled all around the speci-men (Fig. 5c for the conventional compared with Fig. 5b for the pozzolan system). As a result, the cement cylinder’s diameter was reduced by 0.6 mm (–2.4%); the spallings had an approximate thickness of 0.3 mm. A gap of 0.3 mm along interfaces of the cement sheath with the casing and formation is large enough to be

of serious concern as a potential channel and migration pathway for corrosive fluids (e.g., CO2).

To verify the findings, a second specimen of each cement sys-tem was recovered from the HT/HP curing chamber after 6-month exposure to CO2 and was split along the length for analyses (Fig. 6, top). The brownish rim was again found only for the conventional system. SEM images confirmed a channel at the rim of the conven-tional system caused by spallings, whereas integrity exists for the pozzolan system (see Fig. 6, bottom). The observed spallings may not occur naturally under confined pressure in the wellbore, but the findings confirm a potential risk of losing effective zonal isolation and barrier-system integrity over time. These results demonstrated that for the conventional cement system, the corrosion process in the presence of CO2 progressed faster than for the pozzolan cement system. The latter did not show any indication of detrimental leach-ing or loss of integrity after 6-month exposure to CO2.

Effect of Pozzolan on the Formed C-S-H. An examination of the effect of the pozzolan on the formed C-S-H phases is necessary to understand why the set pozzolan system shows less carbonation than the conventional system and especially why it preserves integrity after 6-month exposure to CO2, whereas the conventional system does not.

SEM images were prepared from the pozzolan system after 4-day (Figs. 7a and 7b) and 3-month (Fig. 7c) curing at 300°F and 3,000 psi in tap water. During the hydration of the pozzolan

0

1500

3000

4500

6000

0 1 3 6

Exposure time to CO2 loaded water (months)

Com

pres

sive

Str

engt

h (p

si)

0.0001

0.001

0.01

0.1

1

Wat

er P

erm

eabi

lity

(mD

)

pozzolanconventional

0%

5%

10%

15%

20%

25%

30%

35%

0 1 3 6

Exposure time to CO2 loaded water (months)

Wei

ght I

ncre

ase

(%)

0%

10%

20%

30%

40%

50%

60%

70%

Tota

l CaC

O3 C

onte

nt (%

)

pozzolanconventional

Fig. 4—Left: (Confined) compressive strengths (dotted line) and water permeabilities (solid line) of the pozzolan and conventional cement systems as a function of exposure time to CO2. Right: Weight increase (dotted line) and total CaCO3 content (solid line) of the pozzolan and conventional cement systems as a function of exposure time to CO2.

Fig. 5—(a) Pozzolan (left cylinder) and conventional cement systems (right cylinder) after recovering from exposure to CO2-loaded water at 300°F and 3,000 psi for 6 months; specimens were trimmed at the bottom and top for testing water permeability—only the conventional cylinder showed detrimental spalling. (b) Pozzolan cement cylinder stored in fresh water without any changes; (c) the rim from the conventional cement cylinder immediately spalled when exposed to tap water resulting in a decrease in diameter of 0.6 mm (–2.4%).

Page 8: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

June 2011 SPE Drilling & Completion 215

cement system, a dublex film (Huettl 2000; Diamond et al. 1980) out of C-S-H phases (blue arrow in Fig. 7b) forms on the surface of the pozzolan according to the pozzolanic reaction (Eq. 2). This dublex film serves as a crystal nucleus for epitaxial growth of fur-ther C-S-H phases, resulting in a sheath of densified C-S-H phases around the pozzolan spheres (green arrows in Fig. 7). C-S-H phases within the cement matrix are generally nanocrystalline or roentgen amorphous and so cannot be detected by X-ray diffraction analy-ses. However, the chemical analyses of the sheath by EDS confirm its nature to be out of C-S-H phases. After 6-month exposure to CO2-loaded water at 300°F and 3,000 psi, sheaths of densified C-S-H phases can be still found (Fig. 7d) throughout the speci-men, seemingly unaltered by the CO2 exposure. This may explain the lower CaCO3 content or carbonation, respectively, found for the pozzolan compared with the conventional system. The densi-fied C-S-H phases in the spherical sheaths around the pozzolans are compact (lower surface area), resulting in better integrity and higher durability under CO2 exposure.

Testing Mechanical Properties. The effects of CO2 attack on the mechanical properties of the pozzolan and the conventional cement systems were also studied. Young’s modulus, Poisson’s ratio, compressive strength, and tensile strength were determined before and after 1-month exposure to CO2-loaded water (Table 2). The differential in mechanical properties of set cement usu-ally decreases over time because, after setting, cement hydration is diffusion controlled. Thus, changes in mechanical properties significantly decrease after a curing time of 96 hours at 300°F and 3,000 psi. This might not be exactly correct for the pozzolan system because cement/pozzolan blends typically develop strength slowly but continuously over time. The gain in weight caused by carbonation and further hydration resulted in an increase in density from 15.0 to 16.1 ppg (+4.8 wt%) for the pozzolan and to 16.5 ppg (+8.3 wt%) for the conventional system. Water perme-

abilities of both systems increased from 0.00021 to 0.00442 md (pozzolan) or 0.00032 md to 0.00375 md (conventional), but are still sufficiently low to provide zonal isolation. At the same time, the Young’s modulus (approximately –40% for both systems) and Poisson’s ratio (approximately –20% for the pozzolan and –40% for the conventional) decreased significantly. As mentioned earlier, cement carbonation results in higher compressive and tensile strengths. Compressive strengths were initially relatively high for both systems and remained greater than 5,800 psi after exposure to CO2. Tensile strengths increased from the original 354 psi (pozzolan) and 258 psi (conventional) to 468 psi (+30%) or 438 psi (+70%), respectively. The mechanical properties of the conventional cement system are more affected after 1 month of CO2 exposure because of stronger carbonation than the properties of the pozzolan cement system.

Evaluation of the Wellbore Stress Modeling. Portland cement is a brittle material by nature, basically because of a high Young’s modulus and a relatively low tensile strength (compared to com-pressive strength). Therefore, the cement sheath in the wellbore is more likely to fail in tension than in compression. The pozzolan cement system has generally higher tensile strengths and lower Young’s moduli (before and after CO2 exposure) than the conven-tional cement system. On the basis of its more-favorable mechani-cal properties, the pozzolan cement system tends to withstand wellbore stresses better than the conventional cement system.

A numerical wellbore-stress model was run to illustrate the different performances of each cement system for an identical well scenario. The model predicts likely cement sheath failure as a pass/fail feature on the basis of maximum calculated stress compared to the known strength of the cement as described by Mueller et al. (2004) and Myers et al. (2005). The model assumes that casing, cement, and formation are linear elastic, isotropic, and homogeneous as well as mechanically coupled. Furthermore,

Fig. 6—Top: Pozzolan (left cylinder half) and conventional cement systems (right cylinder half) after recovering from exposure to CO2-loaded water at 300°F and 3,000 psi for 6 months. Bottom: SEM images taken from the rim (arrow) of each recovered cement system.

Page 9: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

216 June 2011 SPE Drilling & Completion

it assumes that the casing is centered in the annulus, the cement sheath is defect free, and all horizontal stresses are equal. A change in temperature (ΔT = –50 °F) and pressure (Δp = +600 psi) caused by an injection fluid (e.g., CO2 flooding) was simulated to exer-cise mechanical stresses on each cement sheath (surrounding a 13.375-in. casing in a 17.5-in. hole) at a reference depth of 6,000 ft. Induced stresses, applied to the cement sheath, have both radial and tangential components. Radial stress occurs from the center of the wellbore outward. Tangential stress is perpendicular to the radial stress component. A cement-sheath failure is more likely to

Fig. 7—SEMs of the set pozzolan cement system. (a and b) After curing for 4 days in tap water at 300°F and 3,000 psi. (c) After curing for 3 months in tap water at 300°F and 3,000 psi (pozzolan remained on the opposite fragment). (d) After 6-month exposure to CO2-loaded water at 300°F and 3,000 psi (pozzolan remained on the opposite fragment). EDS analyses reveal the chemical composition of the sheath.

TABLE 2—MECHANICAL PROPERTIES OF THE POZZOLAN AND CONVENTIONAL CEMENT SYSTEMS BEFORE AND AFTER CO2 EXPOSURE

Cement System Density (ppg)

Permeability (md)

Young’s Modulus (Mpsi)

Confining Stress: 1,000 psi

Poisson’s Ratio Confining Stress:

1,000 psi

Compressive Strength (psi)

Confining Stress: 1,000 psi

Tensile Strength (psi) Unconfined

After 96 hours curing at 3,000 psi/300°F and before CO2 exposure

Pozzolan 15.0 0.00021 1.52 0.32 >5,800 354 Conventional 15.0 0.00032 2.07 0.33 >5,860 258

After 30 days exposure to CO2 loaded water at 3,000 psi/300°F

Pozzolan 16.1 0.00442 0.85 0.26 >5,850 468 Conventional 16.5 0.00375 1.17 0.23 >5,850 438

occur if the calculated stresses (in tension or compression) exceed the found tensile or compressive strength of the cement system.

In the graphs of Fig. 8, stresses in tension and compression are shown as positive and negative values, respectively. For the simulated decrease in temperature and increase in pressure, all mathematical models show that the highest radial and tangential stress values are in tension rather than compression. (Figs. 8a through 8d). The cement sheath made from the pozzolan cement system before CO2 exposure (Fig. 8a) has adequate tensile strength (354 psi; Table 2) to withstand the maximum calculated stress in

Page 10: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

June 2011 SPE Drilling & Completion 217

240

230

220

210

200

194

Rad

ial S

tres

s (p

si)

8 7 6.6Dist from Borehole Axis (in)

Radial Stress

20

30

40

50

6064

Tang

entia

l Str

ess

(psi

)

7 8 8.8Dist from Borehole Axis (in)

Tangential StressRadial/Tangential Stress Field

240

220

200195

Rad

ial S

tres

s (p

si)

8 7 6.6Dist from Borehole Axis (in)

Radial Stress

-20

0

20

40

Tang

entia

l Str

ess

(psi

)

7 8 8.8Dist from Borehole Axis (in)

Tangential StressRadial/Tangential Stress Field

190

180

170

160156

Rad

ial S

tres

s (p

si)

8 7 6.6Dist from Borehole Axis (in)

Radial Stress

10

20

30

4044

Tang

entia

l Str

ess

(psi

)

7 8 8.8Dist from Borehole Axis (in)

Tangential StressRadial/Tangential Stress Field

200

180

160

Rad

ial S

tres

s (p

si)

8 7 6.6Dist from Borehole Axis (in)

Radial Stress

-40

-20

0

15

Tang

entia

l Str

ess

(psi

)

7 8 8.8Dist from Borehole Axis (in)

Tangential StressRadial/Tangential Stress Field

pozzolan cement system before CO2 exposure

conventional cement system before CO2 exposure

pozzolan cement system after CO2 exposure

conventional cement system after CO2 exposure

A

B

C

D

Fig. 8—Numerical wellbore-stress model for (a) pozzolan cement system before exposure to CO2, (b) conventional cement system before exposure to CO2, (c) pozzolan cement system after exposure to CO2, and (d) conventional cement system after exposure to CO2. Radial (left) and tangential (right) stresses simulate a change in temperature (ΔT = –50°F) and pressure (Δp = +600 psi)because of the injection of a fluid. Stresses in tension appear as positive and in compression as negative values. Different shadings in color reflect the stress as a portion of the cement strength: blue = <25%, green = 25–50%, yellow = 50–75%, orange =75–100%, and red = >100%.

tension (241 psi; Fig. 8a, left). The conventional cement system sheath before CO2 exposure has lower tensile strength (258 psi; Table 2), nearly the same as the calculated stress in tension (254 psi; Fig. 8b). Therefore, the cement sheath made out of the pozzo-lan system is more likely than the conventional system to withstand the simulated stresses without failures.

After 1 month of exposure to CO2, the mechanical properties for both cement systems changed (Table 2) as did the calculated stresses exercised on each cement sheath during the simulation

(Figs. 8c and 8d). The measured tensile strengths (468 and 438 psi, respectively, for the pozzolan and conventional systems) increased significantly because of carbonation during exposure to CO2. At the same time, the maximum calculated (radial) stresses in tension after exposure to CO2 dropped to 198 and 214 psi, respectively, for the pozzolan and conventional systems. These effects combined to reduce the negative impacts of mechanical stresses on each cement sheath during the simulated temperature and pressure fluctuations. The cement sheath made out of the pozzolan system is expected

Page 11: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

218 June 2011 SPE Drilling & Completion

to have better durability because of the lower experienced (radial) stresses in tension at highest tensile strength.

Cementing Guidelines for Corrosive Well EnvironmentsThe published field studies (Crow et al. 2009; Carey et al. 2007) revealed that corrosion occurred mainly along the cement/forma-tion interface and to lesser extent through the cement matrix. Consequently, defects in the cement sheath, existing annuli, and leakage pathways for migration allowed an increase of contact time and area between the corrosive fluid and the cement, enhancing the corrosion process. On the basis of the findings from the literature survey about cement corrosion, the following cementing guidelines were composed:

• Follow good cementing practices for a successful primary cement job. This requires well preparation and properly designed cement slurry for the well conditions. Engineered spacers, central-izing the pipe, rotating, and reciprocating, in combination with the use of simulation software for the cement-pumping process, help to maximize mud displacement efficiency.

• Improve cement bonding and reduce permeability of the formation. Newhall (2006) reported that a preflush with a sodium silicate system will improve cement bond. An aqueous film of silicates on the surface of the formation and the pipe will form calcium silicate precipitation when in contact with the cement slurry. Furthermore, the permeability of formations containing calcium and magnesium will be reduced because of the precipita-tion reaction. The sealing of formations containing corrosive fluids is beneficial to minimize chemical attacks on the cement sheath. However, this method is not applicable for every section in the well (e.g., perforation zone).

• Design the cement system with suitable mechanical proper-ties. Mathematical wellbore-stress simulations help to predict if the cement sheath will withstand induced tensile and compressive forces created by changing wellbore conditions without failures over the life of the well.

• Use fit-for-purpopse cement systems designed to counteract all corrosion attacks expected. The literature survey revealed that different corrosion types can occur and must be simultaneously and selectively mitigated. In particular, expansion caused by magnesium- and sulfate-containing fluids results in a stronger damage of the cement sheath than a dissolving attack from acids. In addition, chemical reactions are normally accelerated with increasing temperatures. As a rule of thumb, an increase of +10°C in temperature doubles chemical reactions and so doubles the cor-rosion rate. Increasing pressure also enhances the penetration rate of corrosive fluids into the cement pores, and so enhances their corrosive effect.

Cement designs should be developed on the basis of chemi-cal composition of downhole fluids and any potential injection operations with corrosive fluids. The use of API cements is recom-mended to meet minimum quality standards while avoiding initial issues during cement hydration. A reduced C3A content improves the sulfate resistance. The addition of sufficient silica to reduce the CaO/SiO2 ratio in the cement system to approximately unity is necessary to prevent strength retrogression above 230°F.

Portlandite is a general weak point in the set cement and must be minimized because it is sensitive toward most chemical attacks (see Eqs. 8, 9, and 10). It might be true that CO2 converts portlandite into calcium carbonate (see Eq. 13), which temporarily seals cement pores and so slows down further CO2 attack. But at the same time, portlandite will also be easily leached out of the cement matrix in the presence of acids or will react with magne-sium or sulfates causing destructive expansion. The latter results in more-severe corrosion than that caused by CO2 attack alone. Field studies have proved that cement corrosion caused only by CO2 attack seems not to be of great concern but is accelerated in the presence of defects caused, for example, by other corrosion types. Consequently, the elimination of portlandite and the addition of sufficient silica help to mitigate the worst corrosion attacks first.

All chemical attacks within the cement are diffusion-controlled processes, and so the corrosion rate can be effectively reduced with decreasing permeability of the set cement. A cement system with initial minimal permeability contributes to improve durability. The partial substitution for cement with pozzolanic materials at same slurry density can be an elegant “all in one” solution because this reduces the total C3A content, decreases the CaO/SiO2 ratio to pre-vent strength retrogression at temperatures above 230°F, eliminates the portlandite, and decreases the permeability of the set cement.

ConclusionAPI-cement-based systems can be a practical solution for corrosive well environments, as demonstrated in the literature about field studies and laboratory tests. Following good cementing practices is of utmost importance to minimize corrosion of the cement sheath. The literature survey furthermore revealed that properly designed API cement systems containing selected pozzolanic materials gen-erally improve the durability in contact with corrosive fluids.

In the present comparison study, the evaluation of a pozzolan and a conventional API cement system exposed to CO2-loaded water at 300°F and 3,000 psi results in the following: Under the given HT/HP test conditions, CO2-loaded water completely pene-trates all specimens within 1 month. Therefore, this test method can serve to predict long-term effects within a reasonable time frame. Carbonation of both cement systems increases their water perme-abilities and compressive and tensile strengths, whereas Young’s moduli and Poisson’s ratios are reduced. Because of these changes of mechanical properties, mathematical wellbore models predict lower stresses on each cement sheath than before CO2 exposure. Therefore, the durability of each cement sheath improves in regard to stresses induced by temperature and pressure changes—as long as no chemical leaching or spalling occurs.

After 6 months of exposure to CO2, the tested pozzolan API cement system maintains better integrity than the conventional system: Significantly less carbonation is found for the pozzolan system, and only the specimens of the conventional system show severe spalling effects that may risk the loss of zonal isolation in the wellbore. The better performance of the pozzolan cement system appears to be linked to the observed formation of sheaths of densified C-S-H phases around the selected pozzolans.

AcknowledgmentsThe authors wish to thank the management of Baker Hughes Incorporated for permission to publish this paper.

ReferencesAl-Taq, A.A., Nasr-El-Din, H.A., and Al-Shafei, T.A. 2009. A New Tech-

nique to Enhance Cement Resistance to Mud Acids. Paper SPE 119994 presented at the 8th European Formation Damage Conference, Scheve-ningen, The Netherlands, 27–29 May. doi: 10.2118/119994-MS.

Barlet-Gouédard, V., Rimmelé, G., Goffé, B., and Porcherie, O. 2006. Miti-gation Strategies for the Risk of CO2 Migration Through Wellbores. Paper SPE 98924 presented at the IADC/SPE Drilling Conference, Miami, Florida, USA, 21–23 February. doi: 10.2118/98924-MS.

Bellmann, F. and Stark, J. 2006. Neue Erkenntnisse bei der Prüfung des Sulfatwiderstands von Mörteln (New findings when testing the sulphate resistance of mortars). ZKG International 59 (6): 68–76.

Blount, C.G., Brady, J.L., Fife, D.M., Gantt, L.L., Heusser, J.M., and High-tower, M.C. 1991. HCl/HF Acid-Resistant Cement Blend: Model Study and Field Application. J Pet Technol 43 (2): 226–232. SPE 19541-PA. doi: 10.2118/19541-PA.

Brady, J.L., Gantt, L.L., Fife, D.M., Rich, D.A., Almond, S.W., and Ross, D.A. 1989. Cement Solubility in Acids. Paper SPE 18986 presented at the Low Permeability Reservoirs Symposium, Denver, 6–8 March. doi: 10.2118/18986-MS.

Carey, J.W., Wigand, M., Chipera, S.J., WoldeGabriel, G., Pawar, R., Lichtner, P.C., Wehner, S.C., Raines, M.A., and Guthrie, G.D. Jr. 2007. Analysis and performance of oil well cement with 30 years of CO2 exposure from the SACROC Unit, West Texas, USA. International Journal of Greenhouse Gas Control 1 (1): 75–85. doi: 10.1016/S1750-5836(06)00004-1.

Page 12: SPE-132228-PA-P Cementing Solutions for Corrosive Well Environments_uu Diem Slag, Chung Minh No Hdong

June 2011 SPE Drilling & Completion 219

Cole, R.C. 1979. Epoxy Sealant for Combating Well Corrosion. Paper SPE 7874 presented at the SPE Oilfield and Geothermal Chemistry Sympo-sium, Houston, 22–24 January. doi: 10.2118/7874-MS.

Crow, W., Williams, D.B., Carey, J.W., Celia, M., and Gasda, S. 2009. Wellbore integrity analysis of a natural CO2 producer. Energy Procedia 1 (1): 3561–3569. doi: 10.1016/j.egypro.2009.02.150.

Diamond, S., Ravina D., and Lovell, J. 1980. The occurrence of duplex films on fly ash surfaces. Cement and Concrete Research 10 (2): 297–300. doi: 10.1016/0008-8846(80)90086-1.

Grabau, J. 1994. Untersuchungen zur Korrosion zementgebundener Mate-rialien durch saure Wässer unter besonderer Berücksichtigung des Schwefelsäureangriffs. PhD dissertation, TU Hamburg-Harburg, Ham-burg-Harburg, Germany.

Huettl, R. 2000. Der Wirkungsmechanismus von Steinkohlenflugasche als Betonzusatzstoff. PhD dissertation, TU Berlin, Berlin, Germany (September 2000).

Kutchko, B., Strazisar, B., Dzombak, D., Lowry, G., and Thaulow, N. 2008. Degradation Rate of Well Cement and Effect of Additives. Presented at the 4th Wellbore Integrity Network Meeting, Paris, 18–19 March.

Kutchko, B.G., Strazisar, B.R., Dzombak, D.A., Lowry, G.V., and Thau-low, N. 2007. Degradation of Well Cement by CO2 under Geologic Sequestration Conditions. Environ Sci. Technol. 41 (13): 4787–4792. doi: 10.1021/es062828c.

Lécolier, E., Rivereau, A., Ferrer, N., Audibert, A., and Longaygue, X. 2010. Durability of Oilwell Cement Formulations Aged in H2S-Con-taining Fluids. SPE Drill & Compl 25 (1): 90-95. SPE-99105-PA. doi: 10.2118/99105-PA.

Milestone, N.B., Sugama, T., Kukacka, L.E., and Carciello, N. 1986. Carbonation of geothermal grouts—Part 1: CO2 attack at 150°C. Cement and Concrete Research 16 (6): 941–950. doi: 10.1016/0008-8846(86)90018-9.

Morales, M., Morris, W., Criado, M.A., Robles, J., and Bianchi, G. 2003. Improving the Sulfate Resistance Performance of API Cement Class A upon Appropriate Slurry Design. Paper SPE 81000 presented at the SPE Latin American and Caribbean Petroleum Engineering Conference, Port-of-Spain, Trinidad and Tobago, 27–30 April. doi: 10.2118/81000-MS.

Mueller, D.T., GoBoncan, V., Dillenbeck, R.L., and Heinhold, T. 2004. Characterizing Casing-Cement-Formation Interactions Under Stress Conditions: Impact on Long-Term Zonal Isolation. Paper SPE 90450 presented at the SPE Annual Technical Conference and Exhibition, Houston, 26–29 September. doi: 10.2118/90450-MS.

Myers, S., El Shaari, N., and Dillenbeck, L. 2005. A New Method to Evalu-ate Cement Systems Design Requirements for Cyclic Steam. Paper SPE 93909 presented at the SPE Western Regional Meeting, Irvine, California, USA, 30 March–1 April. doi: 10.2118/93909-MS.

Nelson, E.B. and Guillout, D. 2007. Well Cementing, second edition, 3–5. Sugar Land, Texas: Schlumberger.

Newhall, C. 2006. Improving Cement Bond in the Appalachian Basin With Adjustments to Preflush and Spacer Design. Paper SPE 104576 presented at the SPE Eastern Regional Meeting, Canton, Ohio, USA, 11–13 October. doi: 10.2118/104576-MS.

Noïk, C. and Rivereau, A. 1999. Oilwell Cement Durability. Paper SPE 56538 presented at the SPE Annual Technical Conference and Exhibi-tion, Houston, 3–6 October. doi: 10.2118/56538-MS.

Odler, I. 1998. Hydration, Setting and Hardening of Portland Cement. In Lea’s Chemistry of Cement and Concrete, fourth edition, ed. P.C. Hewlett, Chap. 6, 241–289. Oxford, UK: Elsevier.

Onan, D.D. 1984. Effects of Supercritical Carbon Dioxide on Well Cements. Paper SPE 12593 presented at the Permian Basin Oil and Gas Recovery Conference, Midland, Texas, USA, 8–9 March. doi: 10.2118/12593-MS.

Santra, A., Reddy, B.R., Liang, F., and Fitzgerald, R. 2009. Reaction of CO2 With Portland Cement at Downhole Conditions and the Role of Pozzolanic Supplements. Paper SPE 121103 presented at the SPE Inter-national Symposium on Oilfield Chemistry, The Woodlands, Texas, USA, 20–22 April. doi: 10.2118/121103-MS.

Stark, J. and Wicht, B. 2001. Dauerhaftigkeit von Beton—Der Baustoff als Werkstoff. Basel, Germany: Birkhäuser.

Steinmann, K. 1972. Untersuchungen an Tiefbohrzement auf Elektrolytbe-ständigkeit unter besonderer Berücksichtigung der Bohrlochverhält-nisse, 88, 472. Hamburg, Germany: Erdöl Erdgas Kohle.

Sugama, T. 2006. Advanced Cements for Geothermal Wells. Final Report No. BNL-77901-2007-IR, Brookhaven National Laboratory (July 2006), http://www.bnl.gov/isd/documents/35393.pdf.

Suman, G.O. and Ellis, R.C. 1977. World oil’s cementing oil and gas wells: Including casing handling procedures. Houston, Texas: World Oil Handbooks, Gulf Publishing Company.

Taylor, H.F.W. ed. 1964. The Chemistry of Cements, 106–122. London, UK: Academic Press.

Wollherr, H. 1999. Auswahl und Standardisierung geeigneter Zement-Rezepturen für Tiefbohrungen in Norrdeutschland. Erdöl Erdgas Kohle 115 (6) (1999): 298.

Andreas Brandl joined the Cementing Technology group of Baker Hughes Incorporated “BHI” (formerly BJ Services Company) in Tomball, Texas in 2009. He started his career as a scientist with BJ Services in Hambühren (Germany) in 2007 and worked as a cementing specialist in Asia Pacific during 2008. Brandl holds a diploma in chemistry from the Technische Universität München, from which he also received a Doctor of Science for his studies about additives for well cementing. Jennifer Cutler is a senior research scientist with BHI with over 30 years of experience in the petroleum industry. She has worked for and with industry leaders at major oil companies and service companies in domestic and international assignments. Since joining BJ Services in 2004, Cutler’s responsibilities include regu-larly consulting with domestic and international personnel to address their field requirements as well as with research person-nel to provide information toward their projects. Amanda Seholm is a geologist with BHI in Tomball, Texas. She holds a BA in geology from Trinity University in San Antonio, Texas, and an MS in geology from the University of Aarhus in Aarhus, Denmark. Seholm spe-cializes in formation evaluation and materials characterization studies to identify potential completion problems and to cre-ate reservoir stimulation and remedial treatment recommenda-tions. She joined the Geological Services Group in 2006. Michelle Sansil joined BHI in 2007 and works as an associate chemist in the geomechanics group at Baker Hughes Technology Center in Tomball, Texas. Gerald Braun is a senior research scientist with BHI in Tomball, Texas with 30 years’ experience in quantitative phase analysis and electron microscopy. Braun holds BS and MS degrees in geology from the University of Arizona and a PhD in mineralogy-petrology from the University of Minnesota.