Monday, 1 April 2013

Induced Earthquakes in the USA, and some implications for CCS

Here's a recent BBC report on earthquakes induced by oil and gas activities in the USA. As can be expected, the twitter/blogo-sphere has been lighting up over this in the last few days. For me the biggest surprise is that this has only come up in the wider media in last few days: induced earthquakes have been a key topic of discussion among geophysicists for a couple of years now. The USGS has noted an increase in medium-sized earthquakes in the last decade:
The black line shows the total number of earthquakes in the midcontinent USA (excluding the very active San Andreas fault and other active parts on the west coast) greater than M3 since 1970: you can see the increases as the line gets steeper.

The oil industry likes to dispose of waste-water by injecting it into deep-lying saline aquifers. However, it has been well known since the Rocky Mountain Arsenal in the 1960s that deep fluid injection can trigger earthquakes. It is argued that the increase in oil industry injection activities in the last decade has been the cause of the increase in the numbers of earthquakes.

This remains under debate - could the increase be simply that, as more (and better) seismic monitoring networks are installed, we are detecting more earthquakes than we did in the past. The latest news story is a case in point. The paper in Geology attributes an M5.7 earthquake in Oklahoma to injection of waste-water. The Oklahoma Geological Survey has subsequently released a rebuttal stating that as far as it is concerned, there is not enough evidence to tie the quake to injection activities (strangely enough, the OGS rebuttal hasn't been given much of a look-in from the media).

Nevertheless, I think that it inarguable that, in certain cases at least, fluid injection has triggered earthquakes with magnitudes from about M3 to M6.

This brings me to a couple of asides. Firstly, following on from my last post about bad media reporting of these issues, many reports attributed the quake to injection of waste-water from fracking. This is not the case - the waste water in this case came from conventional oil production. This harks back to an older post I made about the relative risk profiles from fracking in comparison to conventional oil and gas. The need to dispose of large quantities of contaminated waste water is not a new, fracking-related problem in the oil industry. If you are opposed to fracking, you must presumably be opposed to all oil and gas related activity.

Secondly, M5.7 is a large earthquake. It is about 100,000 times larger than the quake induced in Blackpool by fracking. It is larger than any earthquake ever recorded in the UK. Perhaps only a few historical earthquakes in the UK have been of a similar size. An M5.7 triggered earthquake here would be serious news.

So, can we get an estimate of what earthquake magnitude might be triggered by our various activities? Art McGarr, a venerable (and venerated) and highly experienced geophysicist with the USGS has made an effort to do this. McGarr cut his teeth in the 1970s looking at mining induced seismicity, where he noticed a correlation between the total energy released during rock extraction and the volume of rock extracted. He developed the so-called McGarr equation:

Sum(Moment) = G dV

The sum of the released seismic moment equals the volume change (dV) multiplied by the shear modulus (G). It should be noted that this equation is based on empirical observation only. It has subsequently been applied to fluid injection (or mis-applied, some would say, as there is no obvious basis for arguing that physical processes during fluid injection should match those during rock removal (mining)), where dV becomes the volume of fluid injected.

More recently, McGarr has been looking at earthquakes attributed to fluid injection. This includes waste-water injection as discussed above, as well as geothermal activities and, of course, fracking. He has developed the following plot:
Each + represents an injection-induced seismic event. Unfortunately for any non-geophysicist readers, McGarr has given the earthquake sizes in moment, rather than magnitude, but 10^12 is about M2, 10^15 is about M4, 10^18 is M6. I've not found out what all of McGarr's abbreviations are, but
  • BUK is the Blackpool earthquake
  • RMA is the quake induced by fluid disposal at the Rocky Mountain Arsenal
  • BAS is the Basel (Switzerland) earthquake caused by geothermal activity
  • STZ is an earthquake caused by geothermal activity at Soultz, France
  • RAT (several of them) are earthquakes in the Raton Basin (Colorado) associated with waste water injection
  • POK is the Oklahoma earthquake discussed in this blog
You can see a general correlation between the maximum magnitude and the injection volume, following a McGarr-esque equation, replacing the sum of the moment by a maximum magnitude: Mmax = GdV. It should be remembered that this line appears to be describing the MAXIMUM POSSIBLE magnitude. There are over 150,000 waste-water injection wells in the USA, only a tiny fraction of them have caused detectable earthquakes.

So how does this apply to the UK? The first thing to note is that deep injection of waste fluids is not allowed in this country, so we can strike this risk off immediately. What about fracking? A typical frack stimulation uses about 1000 - 5000 metres cubed of water - that's ~10^3. This leaves us with a maximum induce-able moment of ~10^13 (or a magnitude of about M3). We get 30 or so M3 events in the UK every year, so inducing a few more due to fracking isn't going to make much difference.

What about CCS? Carbon capture and storage is a key plank in the UK's CO2 emissions reductions plan. All well and good, but CCS involves the injection of very large volumes of fluid into subsurface aquifers. Could this trigger earthquakes?

I've modified McGarr's plot to add the injection volumes of Sleipner and In Salah, two of the foremost CCS projects currently in operation (as well as changing the scale from moment to magnitude to make life a little easier for non-geophysicists):

You can see that, following the McGarr plot, Sleipner and In Salah have the potential to trigger earthquakes of M5 or larger! Of course, they haven't: Sleipner has barely done anything, while In Salah has triggered at most an M1 event (so small you can't feel it without the aid of sensitive seismometers). The McGarr plot tells you the maximum possible magnitude, not what magnitude you will get. Hence why I have shaded in the area under the line: you could get an event on the line, or anywhere under the line.

Still, I find the potential for induced earthquakes from CCS to be worrying. I think this has been under-appreciated by the UK CCS community. There is a clear need for further study on why most injection sites do not produce seismicity, but a few do? What is it that is different about these sites, and how can we identify this in advance, and only select sites that won't trigger events during CO2 injection. At the same time, we can quickly see that the earthquake risk from fracking has been hugely overplayed in comparison to the risks posed by other activites (geothermal, CCS, waste-water injection, mining, and even hydroelectric energy).

1 comment: