Wellbore Leakage Incident Rates.

By the end of 2017 in BC, 25,119 wells had been drilled; 3,594 were drilled and abandoned before 1995 (Fig. 4). There are 2,322 cases of reported leakage from active wells and 7 reported instances from abandoned wells (Datasets S1 and S2). As previously stated, reporting of wellbore leakage in BC only goes as far back as 1995 (Fig. 4). This corresponds to the beginning of the database as well as the year that operators in BC were first required to test for SCVF leakage prior to abandonment. An increase in reported instances occurred in the late 2000s, potentially due to increased drilling activity and self-reporting from industry. This increase in reported leakage continued after 2010, the year of new regulations requiring leakage testing after drilling and during routine maintenance; 2010 is also the year that operators were first required to report OSCL of gas. However, testing for OSCL of gas in BC is only required when there are visible signs, such as dead vegetation or bubbling in water around the wellhead (12). For abandoned wells, there is currently no requirement for leakage testing.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Cumulative number of wells and number of reported leaky wells per year. The sharp increase in 2010 of the number of wells reporting leakage corresponds to stricter testing and reporting requirements.

Leakage testing is generally self-reported by industry, and reporting is only required for positive test results. If we consider all wells in BC except those abandoned before 1995, the year when leakage reporting began in the province, this represents a total test population of 21,525 wells. The 2,322 reported cases of active wells with leakage, therefore, represent a leakage incident rate of 10.8%. This figure is more than 2 times higher than the leakage occurrence rate of 4.6% determined by other researchers for wells in the neighboring province of Alberta (4). These researchers also point out that their number could be influenced by less stringent testing and reporting requirements in the past. According to these authors, it is plausible that many wells in Alberta abandoned before 1995 could present unreported or undiscovered leakage (4), similar to what is observed in BC. In Alberta as in BC, leakage is self-reported by industry (4). For Alberta, however, it is unclear if the self-reporting by industry necessarily includes the submission of negative test results or if it is limited only to the submission of data for wells testing positive for leakage, as is the case in BC. This is an important point given that a comparison study between the leakage database in BC and the results of a field campaign have shown that approximately half of wells with detected leakage do not appear in the database (10, 13). This suggests that the true percentage of leaky wellbores in BC could be much higher than the 10.8% calculated from the theoretical test population. We will see in a subsequent section of this article that the rate of reported leakage could also be a function of the timing and frequency of testing, which are linked to drilling and abandonment dates.

Tables 2 and 3 summarize the wellbore leakage statistics for all active and abandoned wells in BC according to leakage fluid types and exit pathways, respectively. The majority of these leakage incidents (90.7%) involve SCVF of gas, which does not pose a risk of aquifer contamination but which does contribute to GHG emissions. OSCL of gas is rarer and is usually accompanied by SCVF of gas; 5.4% of cases involve a combination of SCVF and OSCL of gas, whereas ∼0.5% involve only reported OSCL of gas.

Liquid leakage in active wells is rarer and is commonly accompanied by gas leakage. In total, 3.42% of leakage instances involve liquid leakage or a combination of gas and liquid, mostly in the form of SCVF (3.2%). Only 6 instances involving OCSL of liquid were discovered (including a combination of OSCL and SCVF as well as liquid and gas), representing 0.22% of active well leakage. The lesser number of instances of liquid leakage that are reported could be explained by the fact that liquid leakage is less likely to reach the surface where it can be detected. Indeed, liquid leakage requires a certain degree of hydraulic head in order to reach the surface; furthermore, liquid leaking along a wellbore will have a tendency to flow into a transmissive interval along the wellbore.

Reported leakage of abandoned wells is rarer than that for active wells. At the end of 2017, there were 7,268 abandoned wells in BC. Of these, only 7 or 0.1% reported leakage after their abandonment date in the form of CL or OSCL of gas (Tables 3 and 4). No liquid leakage from abandoned wells was reported. It is difficult to determine if this low rate of leakage occurrence for abandoned wells is due to the fact that they actually leak less or that leakage from abandoned wells is simply less frequently discovered and reported. Although current regulations stipulate that all incidences of leakage must be repaired prior to well abandonment, there is no program in place in Canada for monitoring wellbore leakage in wells after permanently buried and abandoned (1). It is unclear why and how the 7 abandoned wells under investigation in this study were identified to be leaking. Furthermore, it is important to note that the majority (4 wells) of these 7 wells were abandoned after 1995, when leakage testing prior to abandonment became mandatory (Table 4). Therefore, either the leakage in these 4 wells was not properly identified and repaired prior to abandonment, or leakage developed along the well later after abandonment.

Several field investigations conducted in other study zones outside of BC indicate much higher incident rates of abandoned well leakage than those estimated in this study (6, 7, 14, 15). In general, field investigations tend to find that plugged abandoned wells, such as those in BC, leak far less than unplugged abandoned wells (7, 14). Still, the percentage of abandoned and plugged wells with detectable positive methane flow rates ranges from 0.8 (14) to 69% (7), which correspond to 8 and 700 times, respectively, the incident rate of 0.1% calculated in this study. The incident rate of leakage from abandoned wells could be even higher when considering that field investigations measuring methane fluxes at the surface may be unable to detect leakage due to oxidation and dispersion of methane in the subsurface (15, 16). In general, it is difficult to quantitatively compare the results of our inventory study with direct field measurements, as the database does not provide any information on testing methodology and detection limits. However, it is likely that the true percentage of abandoned and leaking wells in BC is higher than our inventory estimates considering the results of these field investigations and considering the absence of a monitoring program in BC.

Wellbore Leakage Pathways.

Two subanalyses were conducted based on a sample of 29 active well leakage incidents and 7 abandoned well leakage incidents extracted from the databases; the 29 incidents occurring in active wells were chosen based on availability of completion workover reports containing sufficient detail. The goal was to obtain a preliminary idea of the proportion distribution of the various pathways for well leakage (i.e., how frequently one leakage pathway occurs vs. another). The percentage distributions of leakage incidents by type of pathway are shown in Table 5 for 29 active wells and in Table 6 for 7 abandoned wells.

Leakage Pathways in Active Wells.

The completion workover reports of 29 active wells with leakage from the E-library were investigated. The most commonly reported entry pathway occurs through deficiencies in the production casing caused by corrosion or rupture of casing strings (44.8%). This typically resulted in SCVF; however, in 1 incident, leakage from deficiencies in the production casing exited in the form of OSCL. Chemical, electrochemical, and mechanical corrosion of steel casings in contact with highly saline and often acidic subsurface liquids is a common phenomenon (1). This corrosion can also occur in cemented sections of the wellbore, typically along, but not restricted to, zones of poor cement quality (4). These types of leaks were often detected by pressure testing of the production casing.

Leakage from uncemented intervals along the production casing occurred in 34.5% of the cases. Lack of cementing allows for intermediate subsurface fluids to enter unimpeded into the annular space. These always resulted in SCVF as an exit point. In these cases, leakage was repaired by squeeze cementing the unprotected intervals of the casing. Leakage due to the failure of wellhead seals accounted for 10.3% of reported incidents. In these cases, the leakage was remediated by replacing or repairing the wellhead.

Two cases of leakage originating above the surface casing shoe were reported (6.9% of reported incidents). In 1 of these 2 incidences, leakage was detected during drilling and completion operations before the production casing was installed (Well 2552). In the other incident, the surface casing was not completely cemented to the surface as required, which allowed freshwater from an aquifer located at a depth of 30 to 40 m to enter directly into the well’s outer annulus (Well 22912).

Lastly, the type of pathway least frequently reported for leakage was through a production casing failure where the annulus was cemented. Only a single completion workover report among the 29 was found where SCVF was repaired by squeeze cementing an already cemented interval. Prior to remediation, this leakage exited the well in the form of SCVF.

Leakage Pathways in Abandoned Wells.

Table 6 summarizes the observed leakage pathways for abandoned wells. As previously mentioned, all cases of abandoned well leakage involved gases rather than liquids. The majority of cases of abandoned well leakage originate from uncemented intervals below the surface casing (57.2%). The next most common entry points were cemented intervals below the surface casing (28.6%) and plug failure (14.3%). There were no reported cases of abandoned well leakage originating above the surface casing shoe. In all cases, leakage exited the well in the form of CL. In about half of these cases, leakage also exited the well in the form of OSCL in addition to CL.