3 Case studies

Several case studies are presented illustrating specific aspects of wellbore integrity. The De Lier case describes the evaluation of an onshore depleted gas reservoir in the Netherlands for proposed CO2 storage. This feasibility study demonstrated several concerns that may rise when proposing second life applications for depleted hydrocarbon fields. The following section reflects some issues reflected by case studies on Texas, USA. It discusses both the effect of regulatory developments on well leakage probability as well as insights on well material degradation from samples retrieved at wells that were exposed to CO2 for a minimum of 30 years. Finally, extensive well leakage evaluations in Alberta are described, focusing on either oil and gas recovery wells and CO2 and acid gas injection wells.

3.1 Proposed CO2 storage in the De Lier field (The Netherlands)

In 2006 the ‘Nederlandse Aardolie Maatschappij’ (Dutch Oil Company; NAM) considered the possibility of injecting and storing CO2 in the depleted gas reservoir of the De Lier field, located a few kilometres to the north of the port of Rotterdam in the Netherlands. The field is one of the older NAM assets in the onshore Rijswijk Concession, the production of which started in 1958 and ceased in 1992. The De Lier field consists of several stacked reservoirs (Figure 3.1, right). The De Lier gas reservoir has a thickness of 45 m and belongs to the Holland Greensand Member with the top structure at a depth of 1350 m below the earth’s surface. At deeper levels other hydrocarbon reservoirs are present, such as the oil-bearing De Lier sand-shale member with the top at approximately 1550 m depth. A reverse fault is bounding the SW flank of the structure, and a normal fault is bounding the structure from the NE side. The top seal of the Holland Greensand reservoir comprises a 30 m-thick Middle Holland Claystone layer, which is overlain by the 70-100 m-thick Upper Holland Marl. The initial pressure in the Holland Greensand gas reservoir was 150 bar, while the pressure of the depleted reservoir was 30 bar at abandonment. The reservoir temperature is 58°C.

Figure 3.1 Top Holland Greensand depth map of the De Lier field (left). Cross section through the De Lier field showing stacked oil and gas reservoirs (right). Source: NAM.

One of the concerns regarding health, safety and environment for second life application of the field (such as the proposed injection and storage of CO2 in the reservoir), was the integrity of previously abandoned wells. It should be noted that when considering geological storage of CO2 or other second life applications, the status of the field and the involved wells are integrally re-evaluated by the operator and the regulator with respect to the projected new purpose. Other abandonment requirements will apply, governed by the containment of corrosive fluids, rather than by abandonment of depleted hydrocarbon fields. Substantial effort may be needed to adapt the current state to requirements associated with the proposed completely new situation.

3.1.1 Regulatory framework

Regulatory instruments involved in the evaluation of abandonment programmes constitute the Dutch Mining Regulations, the Dutch Mining Decree and the Working Conditions legislation. While the Mining Regulations dictate prescriptive rules on well abandonment, the Mining Decree involves goal-setting legislation on the construction, intervention and abandonment of boreholes and wells to reach an acceptable risk level to these HSE interests. Compliance to these goal-setting elements of the Mining Decree is generally demonstrated using best practices and state-of-the-art techniques. In addition, the Mining Decree is coupled to the Safety Case regime under the Working Conditions Act, requiring justification of risks to workers on a mining location or installation, in each step of its lifetime, to be acceptable. The governing regulatory framework for (re-)evaluation of abandonment programmes is described in detail in Section 5.6.

Currently, the Mining Act, Decree and Legislation have not yet been revised for geological storage of CO2 or other second life applications. However, through the integral approach of both the goal-setting and prescriptive elements of the Mining legislation, supplemented with the Safety Case requirements of the Working Conditions legislation, regulatory instruments so far already seem reasonably comprehensive.

3.1.2 Review of well abandonment status

The De Lier gas reservoir is penetrated or transected by 52 wells, of which 51 are currently abandoned. In total, 49 wells were drilled in the De Lier field itself (Figure 3.1, left), while three wells from the adjacent Monster field also penetrate the Holland Greensand formation. All 49 wells in the De Lier field were plugged and abandoned, as well as two of the wells in the Monster field (one is still operating). Most of the abandoned wells transect the Holland Greensand Member to reach for reservoir stacks at deeper levels, such as the oil reservoir at the De Lier Sand-Shale level (Figure 3.1, right). Shell and NAM previously concluded that most abandonments are of recent date and meet high standards, except for three wells, which were abandoned under different standards that applied during the early sixties or late fifties (Van Luijk, 2003). Wells that are classified as being plugged and abandoned according to Dutch Mining Regulations contain cement plugs that were designed and tested according to the prescriptions. Furthermore, these wells did not show any gas bubbles or casing pressure build-up during the three months observation period after (re-)abandonment. NAM performed a preliminary review of the status of all 51 abandoned wells in the De Lier and Monster fields, resulting in a subdivision of the plugged and abandoned wells into three categories.

Categories I and II comprises wells that were successfully plugged and abandoned in compliance with prescriptions of Dutch Mining Regulations (DMR) for corrosive fluids: abandonment of 34 wells was successful at the first attempt (category I), while 6 wells required re-abandonment resulting in adequate plugging at the second attempt (category II). Its abandonment configurations comply with DMR requirements for underground storage of corrosive fluids. The prescriptions by Dutch Mining Regulations are comparable to the most rigid regulations around the world and significantly more stringent than regulations in e.g. the USA or Alberta, Canada ( see Chapter 5).

Category III consists of 11 wells that were successfully plugged and abandoned according to Mining Regulations, but do not comply with requirements for the presence/injection of corrosive fluids. For the current purpose of oil and gasfield abandonment this is satisfactory, as no corrosive fluids are present at this time. It should be noted that NAM’s abandonment requirements are amongst the most stringent worldwide and these wells still favourably compare with current abandonment requirements from e.g. the USA.

3.1.2.1 Wells of categories I and II

Upon the depletion of the gas and oil reservoirs, the wells of the De Lier field were abandoned according to Mining Regulations. In fact the abandonment measures performed on wells of categories I and II were rigid such that they even complied with prescriptions for the presence of corrosive fluids. On a first glance these wells therefore would seem adequate also for storage of CO2. However, even if these wells were adequately abandoned upon the end of hydrocarbon production, this does not necessarily implicate that this decommissioning is acceptable when taking into account future applications of the abandoned field or wells.

Regarding the proposed CO2 storage in the De Lier field, a complicating factor proved to be the stacked nature of the field. While storage was proposed in the depleted Holland Greensand reservoir, many wells transected this level targeting oil-bearing strata at greater depth. These wells were abandoned to successfully isolate the oil-bearing formations at the end of production, but since the wells were not perforated at the level of the Holland Greensand, they were not plugged at that level. This aspect can be illustrated by LIR-47 well. In this well, the top of the Holland Greensand Member is located at a depth of approximately 1440 m (Figure 3.2). At abandonment this formation did not act as a reservoir, and in compliance with Dutch Mining Regulations the well therefore was not plugged at this level, leaving it cased and cemented. However, in the light of proposed storage applications this abandonment configuration would need to be adjusted. Under the worst case assumption that, given sufficient time, the injected CO2 in combination with the connate reservoir water would be able to partially corrode the cement sheath and casing at the Holland Greensand level, CO2 would be able to enter the well and migrate upwards until the first cement plug at a depth of 205 m. As a result CO2 would accumulate at shallow depth (i.e. 205 m) with a single cement plug between the CO2 and the surface, or (once again due to corrosion) migrate out of the well to flow into shallow permeable formations.

Figure 3.2 Schematic diagram of well LIR-47 after abandonment, showing the absence of an abandonment plug at the level of the caprock of the proposed CO2 storage reservoir (Holland Greensand Member).

Similar to LIR-47, no abandonment plug is present at the top of the Holland Greensand Member (at 1390 m) in LIR-7. In case CO2 would be stored in the reservoir, ultimately corrosion of the cement sheath and casing at this level could lead to leakage of CO2 into the well. Subsequently, CO2 can migrate upwards through the well until the level of the mechanical bridge plug at 693 m, where it will accumulate and potentially corrode both the casing and the bridge plug. However, unlike well LIR-47, at LIR-7, the interval of the 7” casing between 911-621 m is not cemented (Figure 3.3). Therefore one less barrier is in place to contain CO2 in the well and after the casing at these levels would have been corroded, the CO2 reaches the uncemented well part and can migrate further upwards through the annulus. It remains unclear whether the potential pathway would extend to levels equivalent with the upper part of the cement plug at 693-599 m or to the connection to the LIR-46 sidetrack.

Figure 3.3 Schematic diagram of well LIR-7 after abandonment. Some uncertainty exists about the exact configuration near the side track to well LIR-46.

Taking into account the proposed storage of CO2 in the Holland Greensand reservoir, significant adaptations of the current abandonment configuration would result from the re-evaluation of these wells by the operator and the regulator to ensure safe and effective storage.

3.1.2.2 Category III wells

The 11 wells that were successfully plugged and abandoned according to Mining Regulations, but do not comply with requirements for the presence/injection of corrosive fluids, obviously are candidates for repair requirements when CO2 would be injected in the Holland Greensand reservoir. Different reasons obstruct their suitability and should be remediated when taking into account long-term CO2 storage.

Two wells (LIR-10 and LIR-16) do not have a cement plug of at least 50 m on top of the mechanical bridge plug placed between the Holland Greensand and the underlying De Lier oil-bearing formation, which would be required in case of involvement of a corrosive fluid. The LIR-10 well lacks a cement plug between the Holland Greensand and the underlying De Lier oil-bearing formation. The LIR-16 well does not hold adequate isolation between the De Lier formation and the Vlieland formation, but includes a 150 m cement plug between the Holland Greensand and the De Lier formation. These issues should not pose a direct problem for the proposed CO2 storage purpose, as the involved lacking plugs are situated at deeper levels than the proposed CO2 storage reservoir.

After the first abandonment attempt, gas leakage from the Texel and/or Holland Greensand formations was observed from several wells. In order to reach an effective plugging in a second attempt, silicon squeezes were applied on six wells (LIR-4, LIR-18, LIR-19, LIR-41, LIR-48 and LIR-49). Although appropriate to seal off hydrocarbons, it remains uncertain whether the silicon is resistant to CO2 and CO2-rich waters.

Three wells (LIR-3, LIR-14 and LIR-15) were abandoned prior to the enforcement of abandonment regulations and therefore do not have the required cement column lengths according to the current Dutch Mining Regulations. However, except for the surface plugs of LIR-14 and LIR-15 wells, showing lengths of 6 and 10 m, respectively, all abandonment plugs significantly exceeds lengths recommended by e.g. API (1993), who advise a cement plug to extend at least 50 ft (15.24 m) above and below the casing shoe.

3.1.3 Implications

The De Lier case of well abandonment provides an excellent example of some typical issues that could arise when considering second life applications of fields. First of all, it is obvious that the historical development of abandonment regulations plays an important role. In the De Lier field three out of 51 abandoned wells were plugged before regulation was put into force in 1967. These wells show significantly shorter cement plugs relative to current standards. However, they comply with the philosophy of the Mining regulations and Decree, meeting standards to effectively isolate the current low pressure depleted gas reservoir.

However, in the feasibility phase of proposed redevelopment of the field, the status of all wells has been re-evaluated in the context of long-term CO2 storage in the Holland Greensand reservoir. The fact that the De Lier field consists of a stack of reservoirs leads to increased complexity. Several wells that transect the proposed CO2 storage reservoir at the Holland Greensand level are actually aimed at deeper strata. Furthermore, the performed abandonment measures at the time did not take into account the potential application of the reservoir for CO2 storage purposes. During the abandonment of these wells, no cement plugs were required at cap rock level of the Holland Greensand reservoir as long as no perforations were present at this level. The current abandonment status and configurations do not pose any threat to gas leakage to surface. However, if corrosive fluids would be injected and stored in the Holland Greensand reservoir, additional abandonment measures would be required to comply to the goal setting elements in the Mining Decree and Work Conditions legislation (see Chapter 5), by using best practices and state-of-the-art techniques in order to reach acceptable risk levels with respect to health, safety and environment interests.

Based on the outcome of the well evaluation, NAM decided that geological storage of CO2 in the De Lier field was not economically feasible at the time. Costly workover operations would be required on several wells to reduce CO2 leakage risks to acceptable levels. Some of these wells were not even accessible as a result of urban expansion. Moreover, on basis of the lessons learnt from this study NAM improved its company best practice on well abandonment. Although not required by Dutch Mining Regulations, as of 2006 NAM aims to abandon wells in a manner compatible to CO2 storage in case the site has been earmarked for potential future CO2 storage. NAM’s adoption of this abandonment philosophy enhances the potential of their assets for future applications, and consequently its value.

3.2 Well evaluation at Gulf Coast and SACROC (Texas, USA)

Hydrocarbon production in the USA kicked off with the first well drilled in 1859 in Pennsylvania. By 1992 approximately 3.2 million boreholes were drilled for oil and gas exploration and production throughout the USA (Calvert and Smith, 1994). Of these about 600,000 were still producing in 1992, leaving 2.6 million wells dry, inactive or abandoned. Since the 1950’s many of these wells have been converted into injection wells for disposal of oil field brine.

Significant numbers of the relatively shallow wells, drilled and abandoned before 1930, were not plugged with cement and are not governed by operators. These are the so-called orphan wells. However, also in recent times, such as after the 1986 oil crisis many deeper wells were left unplugged after operating companies became insolvent. Although legal responsibility is lacking, frequently these wells are monitored by state authorities. In contrast, abandoned wells are generally not monitored for leakage by state authorities, as successfully plugged wells are considered to maintain their sealing integrity (Ide et al, 2006). However Hovorka et al. (2004; 2005; in: Nicot, 2008) concluded from the evaluation of 19 penetrations in the area of review of a CO2 injection well at the Frio experiment, that only 3 abandoned wells were plugged with cement below the lowermost drinking water zone.

Texas has a long tradition of oil production. The state still holds the largest crude oil reserves of the USA, amounting 5.1 billion barrels in 2007 (EIA, 2007). Except for the Gulf of Mexico (1.3 million barrels crude oil per day in 2006), the highest oil production rates in the USA are found in Texas. With a production of 1.088 million barrels per day Texas accounted for 21% of the total domestic production in 2006 (EIA, 2006). The first oil wells in Texas were drilled in 1866 (Ide et al., 2006). As the earliest state rules on abandonment were not enforced before 1920-1930, many wells in the state of Texas that were abandoned before these regulations were in place will have inadequate plugging configurations (Calvert and Smith, 1994).

Similar to other states, Texas holds many orphan wells, i.e. wells that were never plugged and are no longer under active company control. The costs to plug orphan wells, averaged at approximately $ 4,500 per well, are partly funded by tax revenues on oil production. However, with approximately 135,000 orphan wells situated in Texas alone, the available funds in the order of $ 10 million are only a few percent of the required amount (Calvert and Smith, 1994; Ide et al, 2006). Between 1965 and 2005 over 20,000 wells were plugged in Texas using the Railroad Commission’s ‘Well Plugging Funds’ (Nicot, 2008). In general the orphan wells comprise wells abandoned prior to 1930 that were drilled to relatively shallow depths.

3.2.1 Gulf Coast – well density and data availability

The Texas Gulf Coast comprises oil and gas districts 2, 3 and 4 of the Texas Railroad Commission (Figure 3.4, right-hand side). The region is put forward as a promising target for geological storage of CO2 by Nicot et al. (2006a). It is characterized by intensive oil and gas exploration and production, reflected in the large amount of wells drilled (about 127,000) and extremely high average well density of 2.4 wells/km2 (Figure 3.5; Nicot, 2008). Especially in the northern part of the Gulf Coast, the Houston area, the well distribution is not spatially uniform as it is governed by the occurrence of piercing salt diapirs, resulting in numerous and complex traps where well densities can amount hundreds to of wells per square kilometer. In the Corpus Christi area towards the south hydrocarbon traps are controlled by more common structures along growth faults (Nicot et al., 2006a). It should be noted that the number of penetrations per square kilometer decreases with depth. The Gulf Coast is one of the areas in Texas with a large number of deeper wells, although the majority of wells is situated at depths between 1,500-3,000 m. Only 10% of the fields in the region is deeper than 3,000 m (Nicot, 2008). Well defects leading to leakage are in general less probable for deep reservoirs, as the quality of abandonment practices improved through time along with drilling depth. Figure 3.4 represents wellbores of three onshore districts of the Texas Railroad commission only, but similar time-depth relations likely hold for other areas.

Figure 3.4 Well total depth distribution versus the year of abandonment (left) in the Texas Railroad Commission Oil and Gas Districts 2, 3 and 4 (right). After: Nicot et al. (2006b).

In Texas there is no comprehensive database incorporating all oil and gas wells ever drilled. Of all known wells in the Gulf Coast between Corpus Christi and Houston, about 30% are abandoned wells with electronic plugging records at the Texas Railroad Commission. The remainder consist of wells that are either still in operation, only have microfilm or paper plugging documentation, or no records at all (Nicot et al., 2006a). The Railroad Commission holds data on some 1.1 million well locations across Texas. However, information on some 400,000 older wells in this database is restricted to locations transferred from old maps. Furthermore, the database is not exhaustive. For instance, it comprises only 78 wells drilled before 1934, where hundreds of thousands of wells have been drilled and abandoned before the 1930s. Some of these may have been recompleted or deepened and plugged and entering the records in a later stage, while the majority likely remained unregistered. Also more recent data is lacking. Only after the implementation of the Safe Drinking Water Act in 1974 was plugging and abandonment generally reported.

Figure 3.5 Surface well location along the Texas Gulf Coast, showing over 100,000 wells (source: RRC digital map data). Lower insert shows the average well density per county. After: Nicot, 2008.

The absence of records on many abandoned wells leads to difficulty in locating these wells. Various approaches are used to find abandoned wells if documentation is not available (Calvert and Smith, 1994), such as visual search of the area, interviews with local residents or land owners, aerial photography, or application of metal detectors. Geophysical techniques involve ground penetrating radar, electrical resistivity or infrared. Airborne geophysics was effectively applied to identify brine leakage to the surface in West Texas (Paine et al., 1999). The survey showed that shallow groundwater and surface water was significantly salinated by leakage through old wells. For 39 of the 718 wells penetrating the shallow artesian Coleman Junction formation (at 800 ft depth), mostly dating back to 1920-1960, anomalous geophysical profiles were observed matching that of a leaking well. Leakage rates through these wells were unknown (Nicot et al., 2006a).

Although the region holds abundant saline aquifers that might be suitable for carbon storage, the large number of existing wells causing multiple perforations obviously is an important issue with respect to long-term storage.

3.2.2 SACROC – field samples

3.2.2.1 Field description

The SACROC Unit is located in Scurry County, West Texas, and lies on the north eastern fringe of the Permian Basin (Figure 3.6). The Unit was discovered in 1948 as the Kelly Snyder Field. The SACROC Unit is the 7th largest onshore oil field in northern America with 2.8 billion barrels of original oil in place (OOIP). The reservoir comprises the Pennsylvanian Cisco and Canyon formations. The reservoir is overlain by the Wolfcamp shale (Brnak et al., 2006; Gonzalez et al., 2007). The top of the reservoir is situated at about 2100 m depth and the reservoir has an average thickness of 240 m and temperature and pressure of 54°C and 18 MPa (Carey et al., 2007). The porosity is around 10% (7.6% according to Brnak et al., 2006) and permeability ranges between 10-100 mD. The main producing reservoir, the Canyon Reef limestone formation, is highly heterogeneous with non-producing zones with porosity and permeability at <2% and <1 mD (Carey et al., 2007).

Figure 3.6 Location of the SACROC Unit, situated in West Texas. Modified after: Vest (1970) in Larkin and Creel (2008).

By 1952, 1,200 wells had been drilled in the SACROC Unit by 400 different operators. As pressure declined rapidly, a water flooding operation was started in 1954 to counteract the pressure drop and to improve oil recovery (Larkin and Creel, 2008). In 1972 CO2 flooding commenced together with the injection of H2S. Shortly thereafter the unit’s oil production peaked with rates exceeding 200,000 barrels per day (Brnak et al., 2006). Since 1972, 68 million metric tonnes of CO2 have been effectively sequestered (Carey et al., 2007). The SACROC unit thus holds wells that have been in contact with CO2 for over 35 years. This provides an excellent opportunity to investigate the behaviour of well materials under influence of CO2 under reservoir conditions. To this purpose side-core samples, consisting of casing, primary cement and caprock, were retrieved just above the reservoir-caprock contact by Carey et al. (2007).

3.2.2.2 Field samples

Carey et al. (2007) investigated CO2-cement interactions and its impact on cement performance on the base of wellbore samples Side-core samples were taken from well 49-6, located in the northern region of the reservoir. This well was drilled in 1950 to a depth of 2131 m, with the shale-limestone reservoir contact at 2000 m. The analyzed cement samples were retrieved from a depth of 1994-2000.1 m. At this depth THE casing was cemented during the completion stage, apparently using a neat Portland cement with density 1857 kg/m3, probably of Class 1 cement, equivalent to API Class A (Carey et al., 2007). Subsequently an acid stimulation using 477,000 l of HCl was performed. The well was exposed to injected CO2 for the first time in 1975, functioning as a production well for 10 years and as an injector for 7 years. Alternatively, small amounts of CO2 may have been introduced to the well much earlier in its life cycle by generation from the initial acid stimulation (Carey et al., 2007).

The cross section, which was reconstructed from the samples by Carey et al. (2007), comprises casing, cement and shale caprock (Figure 3.7). Upon retrieval the casing was found to be in excellent condition showing little evidence of corrosion. This is in agreement with evaluation by Crow et al. (2008) of core samples retrieved from a natural CO2 production well as well as corrosion log data indicating minimal wall thickness losses. All 20 K-55 carbon steel casing samples were in excellent condition with only limited corrosion. In contrast to the intact casing the cement of the SACROC field sample, however, is characterized by two distinct alteration zones of the cement.

Figure 3.7 Photograph of samples retrieved from well 49-6, showing (from left to right) casing, cement with alteration zones at both interfaces, a zone of fragmented shale and shale caprock (Carey et al., 2007).

The first zone forms a 1-3 mm thick dark calcite-aragonite rind between the casing and cement. Furthermore, a carbonate vein is deposited against the casing. The CO2 that caused these deposits could have migrated along the casing-cement interface or may have penetrated from the inside of the casing (e.g. via casing threads or corrosion pits).

The intact cement has a measured porosity of 33.5%, which is typical for Portland cements, and air-dried permeability of 0.1 mD. It shows a subparallel set of fractures, filled with calcium carbonate, that could be associated with induced stresses and differential pressures during the subsequent production and injection phases. No pervasive flux of CO2 could have affected the grey cement as apart from calcite and aragonite, also the rapidly reacting portlandite is still present in the sample (Carey et al., 2007).

Between the unaltered cement and the shale caprock a 1-10 mm thick intensely carbonated orange alteration zone can be observed. All of the portlandite in this zone is consumed. It now consist of an amorphous component (about 50%) and three calcium carbonate polymorphs (calcite, aragonite and vaterite), indicating a strong supersaturation of the carbonating fluids. Permeability of this zone is similar to that of the intact cement (i.e. air-dried permeability of 0.2 mD, corresponding to Klinkenberg-corrected value of 0.01 mD under confining pressures up to 27.6 MPa). The texture indicates that CO2 was supplied along the cement-shale interface. The interface between this zone and the cement is formed by a narrow (<1 mm) dark grey deposit of amorphous silica, silica-carbonate and carbonates that may reflect successive carbonation fronts. The dense zone appears to form a continuous barrier between the altered and gray cement zones (Carey et al., 2007).

Finally, between the altered cement and the shale a complex zone of shale fragments embedded in a fine-grained matrix composed of drilling residue mixed with cement (comprising calcite, aragonite, dolomite, silica and significant amounts of halite). The zone is characterized by fragmented shale occurs that contains zones of high or open porosity. Therefore it seems plausible that CO2 migrated from the reservoir through the fragmented shale zone rather than along the shale interface.

The bonding between cement and casing was not preserved anywhere, probably as a result of the retrieving method, but could also be detached before sampling (Carey et al., 2007). In the latter case, however, there are no indications that any fluids migrated along these cracks. This observation is supported by the interpretation of a cement bond log (CBL) that shows adequate bonding of cement to both casing and shale from the top of the reservoir (at 2000 m) to 1950 m.

The analysis of the annular cement (Portland Class H, with 50% fly ash, 3% Bentonite gel; 1710 kg/m3) of a natural CO2 producer after 30 year well life by Crow et al. (2008) shows that cement cores close to the CO2 reservoir (at 1430 m measured depth and 58°C, 10 MPa) were almost completely converted to calcium carbonate. Average porosity and permeability were measured at 41% and 21 μD. In contrast, samples taken from the top of the caprock (at 1372 m MD) display minimal alteration and porosity and permeability of 25% and 1 μD. Cement interfaces with both casing and caprock show apparently tight contacts without alteration zones or debris (Figure 3.8). However, a Vertical Interference Test (VIT) between two perforated intervals shows evidence for some hydraulic communication, likely along the interfaces (Crow et al., 2008).

Figure 3.8 Core samples from a natural CO2 producer taken at 1427 m MD (left) and 1417 m MD (right), showing tight fits at the cement-formation and casing-cement interfaces, respectively. After: Crow et al. (2008).

Carey et al. (2007) conclude that the cement at the SACROC 49-6 well survived and retained its structural integrity after 30 years under influence of CO2. Similar conclusions are drawn by Crow et al. (2008) evaluating cement from a natural CO2 producer. In spite of the somewhat enhanced permeability with respect to virgin Portland cement, the cement in both investigated wells is expected to provide adequate resistance against significant CO2 migration through the cement matrix. On the other hand migration along the interfaces between cement and both casing and caprock seems to have occurred, leading to the observed alteration zones. Therefore, the integrity of these interfaces appears to be the most critical aspect in well integrity for geological CO2 storage.

3.3 The Alberta Basin (Alberta, Canada)

The Alberta Basin in western Canada is a mature sedimentary basin, hosting large oil and gas fields. Hydrocarbon exploration started in the late 19th century and the oldest recorde abandoned well dates back to 1893. Drilling and production regulations were not in place before 1938. Upon a major oil discovery in 1947, rapid growth followed that continues until today following the oil and gas market economical cycle (Bachu and Watson, 2006). Over 320,000 wells have been drilled, with drilling rates of 12,000 wells per year over the last decade (Gasda et al., 2004; Figure 3.9). The Alberta Basin shows a very high quality and complete database on oil and gas wells. The database holds data on all deep wells and is managed by the Alberta Energy and Utilities Board (Bachu and Watson, 2006). In western Canada, similar to elsewhere around the world, current practice in oil and gas industry involves well monitoring during its active lifetime, while no monitoring is required after adequate abandonment is reported (Gasda et al., 2004).

3.3.1 Spatial and temporal characterization of wells

Gasda et al. (2004) performed a quantitative characterization of wells penetrating the Viking Formation, which contains approximately 5% and 8%, respectively, of oil and gas reserves in the Alberta Basin. The Viking Formation forms a wedge-shaped interval dipping to the southwest. Along its western boundary the formation is deepest exceeding 3000 m depth. Towards the outcrop in the northeast, the burial depth gradually decreases. Over 200,000 wells were drilled to penetrate or transect the Viking Formation, resulting in an average well density of 0.48 wells/km2. Of the existing wells 31% is active (either production, injection or disposal wells), while 68% is inactive (standing, suspended or abandoned) (Gasda et al., 2004). Figures for the entire province of Alberta show that 51% of the wells is active, whereas the remainder is abandoned or inactive/suspended, i.e. 34% and 15%, respectively (Bachu and Watson, 2006). In spite of the favourable potential CO2 storage capacity and infrastructure in the Alberta Basin, the abundance of existing wells could pose a risk for long-term storage.

The spatial distribution of wells obviously is determined by the location of oil and gas reservoirs. As a result, clusters of high well density exist around known hydrocarbon fields. The surface expression of the wells is complicated by the presence of stacks of reservoirs: adjacent wells may be aimed at different vertical levels that contain hydrocarbons. Gasda et al. (2004) show that approximately one third of the wells in the Viking Formation is situated in high to very high density clusters. It should be noted that zones showing very high densities (involving 5.2% of the wells) are typically associated with heavy oil recovery, as production techniques require closer spacing of well relative to less viscous oil or gas. In the Alberta Basin heavy oil is generally found in horizons that are too shallow for potential long-term CO2 storage (Gasda et al., 2004). As a result the clusters showing the highest densities will not impede carbon storage. The lower well density regions show relatively many dry holes, i.e. unsuccessful exploration wells that were abandoned immediately after drilling.

Figure 3.9 Location of the Viking Formation in the Alberta Basin, Canada, showing the location of all wells penetrating the formation (left) and a cross-section of the Alberta Basin (right). After: Gasda et al., 2004.

These wells generally do not include a steel casing and plugging is performed in the open hole. Approximately 50% of wells in the lower density areas is classified as abandoned (Gasda et al., 2004). Throughout Alberta, roughly 50% of the abandoned wells were drilled and abandoned without production casing in place (Bachu and Watson, 2006).

The temporal evolution of wells penetrating the Viking Formation reflects the general production developments in the Alberta Basin, starting in the late 19th century. Drilling activity rapidly increased after a major oil discovery in 1947. Subsequently, the number of wells drilled steadily increases with approximately 60,000 well per decade, peaking with 12,000 wells per year over the last years. The continuous development of exploration and production results in higher well densities, especially for wells now classified in lower density areas.

The observations above illustrate the significant differences in well density and therefore in the number of wells that could potentially be impacted by CO2 storage, depending on the location of injection. Based on this, CO2 injection would be preferred in low well density areas. On the other hand, storage operations could economically benefit from the better infrastructure that already exists in areas with larger well clusters.

3.3.2 Failure of oil and gas wells

Historically wellbore construction mainly focused on the production of economic, conventional hydrocarbon reservoirs. As a result little attention was paid to unconventional resources, such as coal bed methane and shale gas, which in Alberta typically are situated at relatively shallow levels (Bachu and Watson, 2006). In many cases, well cementing was not required at these levels, leading to the absence or inadequacy of the primary cement sheath. Since the implementation of the Oil and Gas Act in 1949, well cementing of deeper formations is required. As a result, leakage of shallow gas resources through or along the wellbore is observed by sustained casing pressures or soil gas migration outside the surface casing. In 70% of the reported cases of well leakage, the gas originated from depths less than 500 m, while approximately 90% was derived from less than 700 m depth (Bachu and Watson, 2006). Although sustained casing pressure and gas migration originate from shallower formations, these kind of well failures could provide leakage pathways to the surface once zonal isolation or casing fails (Bachu and Watson, 2008).

Reported incidences of sustained casing pressure or gas migration by operators since 1995 indicate that these types of leakage can be caused by the absence of cement, degradation by formation fluids, deformation resulting from swelling clays or (thermal) stimulation and production operations, or poor cementing jobs especially in deviated wells (poor mud removal, cement shrinkage, cement contamination or poor curing). Of the approximately 316,500 wells recorded by the Alberta Energy and Utilities Board, Bachu and Watson (2006) reported that 3.9% sustained casing pressure, 0.6% showed gas migration and 0.1% showed both phenomena. Well leakage is higher in cased wellbores than in open-hole wells (i.e. 6.1% and 0.5%, respectively), possibly due to historically more stringent regulations (requiring multiple abandonment plugs) on open-hole well abandonment (Figure 3.10).

Figure 3.10 Typical well abandonment configurations in Alberta, Canada, showing drilled and abandoned open-hole wells (left) and cased, completed and abandoned wells (right). After: Watson and Bachu, 2007.

Barclay et al. (2001) described remedial cementing to seal small gas vents in wells in Alberta, using an ultralow-rate squeeze-cementing technique. In single wells in different fields two subsequent conventional squeeze jobs failed to eliminate gas migration. Application of ultralow-rate squeeze operations with advanced cement, using optimized particle-sized distributions, lead to the successful permanent abandonment of both wells.

Since the 1950’s observed gas migration and sustained casing pressure are mainly governed by intensive drilling of newly discovered resources, oil price developments and advances in regulatory demands (especially in the late 1960’s) or technology. Immediate repair of well leakage is required for serious leakage, i.e. high leakage rates exceeding 300 m3/day, high pressure build-up compared to hydrostatic gradient, or leakage of liquids or sour gas. Non-serious leakage repairs may be postponed to abandonment of the well. It should be noted that only 2% of recorded leakage cases showed rates over 300 m3/day and 0.8% involved flow of liquids (Bachu and Watson, 2006).

Watson and Bachu (2008) investigated abandoned wells in Alberta, Canada, and found that only 50% of the wells with a cement plug placed using the dump bailer method were intact. This high rate of failure is believed to be caused by mechanical failure of the bridge plug. Therefore, this abandonment method seems less secure, leaving plugs with shorter expected life relative to other plugging methods (Watson and Bachu, 2008).

3.3.3 Failure of CO2 and acid gas wells

The first CO2-EOR operation in Alberta was performed in 1981. In 2008 at seven sites CO2 stimulation is used to enhance oil recovery. The first acid disposal operations started in 1991. However, only in 1994 a provincial directive was implemented, governing Class III injection and disposal wells by the requirement of hydraulic isolation of the reservoir zone and cementing over intervals bearing groundwater resources. Approximately half of the CO2 injection wells and most acid gas disposal wells were drilled after 1994, showing distinctively lower failure rates compared to wells drilled prior to the improvement of regulatory requirements.

Bachu and Watson (2008) reviewed failure of wells associated to CO2-EOR (31 wells) and acid gas disposal operations (48 wells) in Alberta (Figure 3.11). Most CO2 and acid gas injection wells are converted from existing wells that were originally drilled for different purposes. In the majority of evaluated wells conventional materials (e.g. Class G neat or bentonitic cement, J55 or K55 steel grade) were used in well construction. In 19 wells sulphur-resistant casing material (i.e. L80 or N80) was applied.

Different types of well failure were recognized, distinguishing various sections or equipment of the wellbore, in order to allow adequate comparison of the results against previous studies of the general well population in Alberta. Reviewed categories of well failure comprise tubing and/or casing failure, packer failure, zonal isolation failure, and sustained casing pressure or gas migration outside the surface casing (Bachu and Watson, 2008).

Figure 3.11 Locations of acid gas and CO2 injection wells in Alberta, Canada, in 2008. After: Bachu and Watson, 2008.

In contrast to casing failure and sustained casing pressure that were predominantly observed in the general population of deep wells, the majority of well failures associated with acid gas and CO2 injection wells is caused by tubing and packer failures. Furthermore, converted wells are recorded to be more prone to wellbore failure than in wells drilled for purpose, presumably as a result of better cementing at the top section of the well. Comparing acid gas disposal wells and CO2 injection wells, the former show lower failure probabilities, most likely due to more stringent design standards and maintenance standards. For both types of operations wells built for purpose show significantly lower failure rates. It should be noticed that this is mainly due to the occurrence of sustained casing pressures and gas migration and therefore not directly related to the injection operations (Bachu and Watson, 2008).