Among the deep drivers of my career, two stand out: an attraction to cement, and a profound respect, tinged with envy, for geologists. Whereas the first is physical, gravitational almost (who can seriously love grey dust?), the second one is emotional: poets that can read the language of nature and discern hidden structures and relationships. If only geologists could speak mathematics the world would be theirs...
Well Integrity Consultant, Germany
But I digress. The two drivers have come together over the past few years as I have delved deeper into the role creeping formations play in well integrity. The fulguration came during a study of wells in the Paris Basin and the risk of contaminating aquifers. The basin, a picture-perfect example of those sedimentary animals, takes its name from the overlying metropolis and has a numbers of challenging characteristics:
- It has a series of fresh water aquifers culminating in the Albian-Neocomian, which lies at a depth of around 800 m (a sort of scaled-down version of the Great Artesian Basin).
- Around 1,000 m deeper we find the Dogger, another aquifer but saline. Its hydraulic head is 85 m higher than the Albian's, meaning that any hydraulic connection - say a leaking well - will cause pollution of the upper, strategic water reserve.
- The basin is a bit of a Swiss cheese: besides a fair amount of oil wells (though it ain't no Texas), there are water wells and geothermal doublets, a particular feature of the region. Each doublet is made of a producer and an injector (hence the name) of Dogger brine, which can reach a temperature of 80 degC at its deepest; at that temperature you can only use it for district heating, since the brine is too cold to generate meaningful power. Most doublets were drilled in the '70s and early '80s, although there has been some resurgence in this century.
Above: temperature map of the Dogger formation, from Lopez et al., 2012. The tiny white blob at the center is the city of Paris.
The risk of contamination thus seemed real, especially because of the abundance of wells coming from industries with more dubious safety records than oil & gas (don't laugh: this is actually true!). And yet: in spite of the potential, actual leaks have been rare, and all of them flowed through the casing and not the annulus. So the search was on for the hidden helper that was keeping a lid on risk.
To cut a long story short, we found a number of clay layers of Lower Cretaceous and Upper Jurassic age that exhibit reliable and effective creep, helping cement seal the annulus and controlling fluid migration into the fresh water aquifers. Et voilà : our geological barrier. Evidence for the sealing mechanism and the barrier's failure mode is thin (who would log shallow shales?) but convincing: drilling hazards, a few logs and a serendipitous pressure test.
Creeping shales are likely to fail through debonding, here represented by the two thin red lines, rather than through the opening of a vertical hydraulic fracture.
Looking around in other geological provinces, including the Molasse Basin in Germany, unearthed other such formations that share a key trait: they all contain a plastic clay mineral called smectite (a.k.a. montmorillonite) as the principal constituent of their matrix. Of course some rock salt layers also creep, and so might ice and permafrost. Coal layers swell too when exposed to carbon dioxide (CO2), although this is of passing interest to anybody outside the carbon geological storage cabal.
There is also additional evidence of geological barriers in the literature, among others in the SPE paper 11932 by Williams et al. that described the behavior of the Hordaland Green Clay offshore Norway. Even the latest version of NORSOK D-010 allows the use of “creeping formations”, albeit under puzzlingly strict conditions.
To be precise "creep" is a catch-all term that lumps together complex viscoplasticity and slightly simpler viscoelasticity. Luckily, as pointed out by Brice Lecampion, you can use the latter provided you fit measured data (the physics boilerplate is longer, but the essence is the same). Finite element models can help figure out how fast the formation closes the annulus, or squeezes a microannulus shut, as well as the closure stress that prevents (and mitigates) leaks. Alternatively, as Brice and Sofia Mogilevskaya have done, you can solve a simple elastic problem in the Laplace space (simple if you're black belt in mathematics, that is). Analytical solutions, even if they cannot tackle all cases, are fast and help validate finite element approaches.
Of course you cannot "engineer" geological barriers directly: either you have them or you don't. Some places, such as Texas, seem to have them; others, such as Colorado, may not. But if they are present, they can and should be part of your well design. After all they are robust: they control a leak even after they fail, the way cement does. Better than cement alone, though, they are actually self-healing: once the cause of debonding is eliminated, they will seal the annulus again and again.
There are a few things you can do to understand creeping shales and predict their behavior. Logs play a role, and so the judicious use of experiments. Risk assessment should drive the selection of candidate formations, and an optimized characterization program can reduce the cost and time required to gather data.
If you want to know more, you can download the paper we presented at the GHGT-13 from ResearchGate, or you can drop me a message and I'll send it to you: I'll be glad to have your opinion and discuss further this passion of mine.