ReFINED well integrity

This week's big news was the release of the latest paper from Durham University's ReFINE Group (Research on Fracking IN Europe - academics love a dodgy acronym!). In it, they compile statistics on well integrity from a range of sources, as well as looking at well abandonment and orphaned wells (where the company owning a well goes bust, leaving a well with no-one to look after it).

This follows ReFINE papers on induced seismicity, and on hydraulic fracture height growth. Their approach in each of these cases is to use as much data as they can possibly get their hands on, with little thought for quality control, or on whether they are comparing apples with apples. In their own words:

This paper draws on a variety of datasets, mostly published, but in some instances sourced from online repositories or national databases, and follows the approach of Davies et al. (2013). In that study, the risk of induced seismicity due to hydraulic fracturing was reviewed, and intentionally included all datasets in the public domain that were considered to be reliable, rather than de-selecting any data (Davies et al., 2013). This inclusive approach has a drawback because well barrier and well integrity failure frequencies are probably specific to the geology, age of wells, and era of well construction (King and King, 2013). A wide range of failure statistics is therefore reported, and although they are presented on a single graph to show the spread of results, this is not intended to imply that direct comparisons between very different datasets (i.e. size, age of wells, geology) can be made.

This means that data from recent drilling in the Marcellus (which is probably relevant) is presented alongside less relevant data from offshore wells (drilling offshore is always a more challenging prospect, with hundreds of meters of water between your platform and the well-head), from China, or even from the 1920s in California.



I sympathise with the ReFINE group in their desire to include as much data as possible, regardless of quality, given the general paucity of it in the public domain. When I spoke to Prof Davies recently at the Unconventional Gas Aberdeen conference this week, he said that the main thing to take away from this paper is simply that there isn't enough quality data to draw any conclusions.

However, I suspect that the less-well-informed audience will skip straight past the above caveats and draw conclusions where the data, when considered properly, do not support them.

The second issue, again correctly identified in the ReFINE paper, is that there is a big difference between a well with integrity issues and a well spewing hydrocarbons into the environment. Neither are desirable, but it is wrong to conflate statistics of the one to infer risks associated with the other. It'd be like claiming the number of cars that have failed an MOT is equal to the number of people killed in car crashes (MOTs are a subject dear to my heart right now, as my old banger didn't fare too well in its recent test, necessitating over £500 of repairs!).

Again, the ReFINE paper expresses this well in the introduction:

The terms ‘well barrier failure’ and ‘well integrity failure’ were differentiated by King and King (2013). They used ‘well integrity failure’ for cases where all well barriers fail, establishing a pathway that enables leakage into the surrounding environment (e.g. groundwater, surface water, underground rock layers, soil, atmosphere). ‘Well barrier failure’ was used to refer to the failure of individual or multiple well barriers (e.g. production tubing, casing, cement) that has not resulted in a detectable leak into the sur- rounding environment. The same terminology is used in this paper: ‘well integrity failure’ includes cases when gas or fluids are reported to have leaked into soils, rock strata or the atmosphere, and ‘well barrier failure’ includes cases where a barrier failure has occurred but there is no information that indicates that fluids have leaked out of the well.

Wells are deliberately designed with multiple barriers separating the environment from the contents of the well. This is done for a reason. While individual barriers may develop issues, it is very unlikely that all will fail to allow leakage. While the ReFINE paper cites the King and King (2013) SPE paper, it sadly misses one all-important conclusion

For US wells, while individual barrier failures (containment maintained and no pollution indicated) in a specific well group may range from very low to several percent (depending on geographical area, operator, era, well type and maintenance quality), actual well integrity failures are very rare. Well integrity failure is where all barriers fail and a leak is possible. True well integrity failure rates are two to three orders of magnitude lower than single barrier failure rates.

The ReFINE paper does draw the distinction between barrier failure and complete loss of integrity in their introduction. However, in the results section there is no attempt to differentiate them, meaning both issues are lumped together. This makes it very easy to mis-interpret the results, meaning that people may try to claim that the study shows that several percent of all wells are leaking hydrocarbons to the environment.

The most relevant dataset used in the study is that presented by Considine et al (2013). While the Considine paper itself is behind a paywall, the same data is available in this report. This examines drilling violations for companies operating in the Marcellus shale. 3,500 wells were drilled during this period. They classify major and minor impacts:

Major environmental events are defined in this study to include major site restoration failures, serious contamination of local water supplies, major land spills, blowouts and venting, and gas migration.

Non-major environmental events concern site restoration, water contamination, land spills, and cement and casing events that do not involve what is classified as having major environmental impact. Many of the NOVs in this category, while resulting in measurable pollution, were rather minor, involving, for example, a gallon of diesel fuel or antifreeze spilled on the ground.

In total they identify 25 cases of the former, and 820 of the latter. I think it is fair to focus our attention on the former, major incidences, while noting that even a gallon of diesel spilled on the ground results in a violation (I'd better not bring my old leaky banger to Pennsylvania then!). Of these 25 cases:

nine involve major spills of materials on land, another eight entail spills that contaminated local water supplies, four incidents concern well blowouts and venting, two events incur major site restoration impacts, and two events concern gas migration. There were no reported cases of hydraulic fracturing fluid migrating into potable water supplies.

As I've discussed before, the major issue is how we handle, store and transport fluids at the surface, rather than anything going on underground. In only two cases out of 3,500 was subsurface gas migration found to be an issue - a rate of 0.05%.

Moreover, of the 25 cases judged as major environmental impact, all bar 6 have been remediated completely:

the environmental damages resulting from these events were mitigated with the exception of six cases, two of which are too early to determine if remediation has been completed and for the other four cases, remediation efforts have been undertaken but not verified as completely effective. Hence, even when there are serious environmental impacts, regulators and drilling companies act to completely remediate the environmental damages. This implies that the PA DEP is acting effectively to minimize and in many cases prevent environmental harm from occurring. Hence, the Pennsylvania data shows that of the polluting environmental events that resulted in environmental damage, the regulatory agencies and drilling companies acted to completely remediate those damages.

What is equally significant is that the rate of violation decreased substantially over the study period. During this period, PA DEP introduced new regulations for drilling, and it is clear that these had an effect (showing again the importance of regulations and good practice when drilling). One important conclusion made by Considine et al:

The fourth and final conclusion is that the majority of the events were due to operator error, negligence, or a failure to follow proper procedures when drilling. This suggests that the industry has room for improvement, and the frequency of environmental events can be reduced.

This why the Royal Society, Royal Academy of Engineers, CIWEM, PHE, have all said that well-regulated drilling, following best practice, does not pose a risk to health or the environment.

While I feel that the Considine paper is far more informative than the ReFINE paper, there is one major conclusion made by the ReFINE group with which I agree, and that is the need for better data. DECC should consider putting structures in place to ensure that this happens:

A much tighter constraint on the risks and impacts would be obtainable if systematic, long-term monitoring data for both active and abandoned well sites were in the public domain

and

in terms of monitoring, abandoned wells could be checked 2-3 months after cement plugging for sustained casing pressure and gas migration. If the well has no evidence for barrier or integrity failure, it could be cut and buried as per regulations. Soils above well sites could be monitored every 5 years for emissions that are above a pre-determined statutory level. As there are 2152 wells in UK at present, only 430 would need to be checked each year. Monitoring could be intensified or scaled down based upon the results of the first complete survey. Monitoring a proportion of future abandoned shale gas and oil wells should also be feasible. A mechanism may need to be established in the UK and/or Europe to fund repairs on orphaned wells, and an ownership or liability survey of existing wells would be timely.