Induced Seismicity in the UK – Part 2: What is done to manage the magnitude of earthquakes?
The public, regulators and policy makers are understandably concerned about the magnitude of earthquakes induced by fracking operations and the risk of any consequent damage that might occur. When a M2.3 earthquake was induced 7 years ago by fracking operations in the Presse Hall project near Blackpool, the authorities responded by imposing tight limits on the magnitudes of induced earthquakes that they considered acceptable. Threshold magnitudes were set to determine specific actions in a ‘traffic light’ (green, amber or red) system, which I will described in more detail below. The threshold magnitudes were intended to provide a wide safety margin in managing the risk of felt or damaging earthquakes.
In this post, I thought I would take you through some of the information that feeds into how the fracking industry is required to manage the size of induced earthquakes. This blog post is a bit more technical than I would ideally like – but there is a lot of information that I need to cover to provide an appropriate context.
I would like to note that I am not funded by the fracking industry and the opinions expressed here are my own.
1. Site Selection
There are many factors that feed into the selection of a site for fracking operations, but the bottom line is whether they stack up for a site to be commercially viable. These factors include there being a shale of sufficient volume and quality to produce commercial quantities of oil and/or gas, the operator is likely to obtain the necessary permissions, and that it is likely that the fracking and production can be carried out safely and successfully.
Can the choice of site change the seismic hazard? The short answer is yes – alarge amount of fracking has been carried out in the US and Canada without inducing earthquakes above M=0. This is well below the level that may even be felt by the local population. The hope of the fracking industry is that this is the norm, so that fracking should be able to be carried out pretty much anywhere in the UK with similar levels of induced seismicity. However, this track record has been established in regions with very large basins with very simple geology that are continuous in all directions – mostly very boring geology. In contrast, the geology of the UK is complicated on short length-scales because it has undergone a long and complex history. There are faults all over the place – which I like to think as the laughter lines of old age. It is therefore not obvious that the lessons learned in the regions of the Mid-Western US and Canada where the risk of felt earthquakes is low will translate simply to the UK. Hence, I would not be surprised if it was more common for UK sites to experience higher levels of induced seismicity than these areas.
Just to be clear – I do not think that there is a high likelihood of a damaging earthquake occurring at the Preston New Road project. The seismicity so far recorded is clearly lower than that of the natural seismicity in the UK. But, I do not think this is the point. The industry needs to demonstrate skill in managing the risk of induced seismicity within the regulatory framework. The first approach to achieving this is by the operators carefully choosing a site where the risk of inducing felt earthquakes is assessed to be low and by demonstrating that they can limit the size of the earthquakes that occur by how they design their injection strategy. I think of this as trying to engineer the maximum magnitude that could occur during operations. I elaborate on these two topics below.
2. Seismic Hazard Assessment an Planning
Cuadrilla’s Environmental Report (Chapter 12) describes various aspects of the expected seismic hazard posed by the fracking operations at the Preston New Road site. This Chapter is quite technical – so I will translate some of the key points. This report was published in 2015 – so it is a public record of how the site was expected to perform before operations started. Now that we have data coming in, we can compare the expected risks anticipated at the planning stage with reality.
- The injection strategy for the current Preston New Road project has been modified compared to that undertaken in the earlier Presse Hall project in 2011, with the aim of reducing the risk posed by induced seismicity
- Mitigation measures have been proposed to reduce the risk of felt magnitude seismic events occurring (generally greater than 1.5ML), rather than preventing very low magnitude seismic events (less than 0.5ML) occurring altogether.
- The report classifies the likelihood of events of a given size being either ‘High’, ‘Medium’, ‘Low’ or ‘Very Low’ (Table 12.2). However, I think these descriptions are a bit weak because there is no explicit quantification as to what these levels mean in the reports. For example, does a ‘Low’ likelihood mean that you still expect say 1 in 100 fracking operations to produce an earthquake of a given magnitude? This is important because the definition needs to be understandable in the context of how frequent larger event might be if the industry were scaled up to several hundred wells.
- In Appendix L, there is a little more detail on the definition of ‘Low’ in Table 18 on page L98, “Seismic source, pathway and receptor exist. Established mechanism of seismicity, however the linkage is not certain and events anticipated to occur in frequently per duration of exploration operations in the area, industry examples, or has occurred in previous Cuadrilla operations.” This definition of ‘low’ is linked to uncertainty rather than evidence of low rates and again does not provide quantification.
- The report carries forward the maximum magnitude estimate of M=3.1 from the post-mortem geomechanical analysis of the Presse Hall site, but suggests that inducing an event of this size at the New Preston site would only be possible if no mitigation measures are taken (Paragraph 74). However, there is an important caveat on such a statement. An induced earthquake is, by definition, one that occurs primarily within the target formation – so it’s size is fundamentally limited by the dimensions of the shale and the amount of energy put into the system by fluid injection. This is acknowledged in the Environmental report, which defines a separate class of triggered earthquakes as including “A seismic event that is caused by only a small change in stress or by migration of fluids into a pre-stressed, pre-existing fault. Triggered events are sometimes referred to as fault reactivation. Triggered seismic events release more energy than is required to initiate them.
- The same paragraph repeats a statement that I find contentious “This is supported by the observation of maximum magnitudes of coal mining induced earthquakes in the UK (up to magnitude 3.0 ML), which is considered to provide a realistic upper limit of induced seismicity”. In a report on British Earthquakes by Roger Musson of the British Geological Survey it states on the section about Northern England , “The record of modern seismicity is contaminated by mining-induced events; it can be hard to distinguish, especially for older events, what is mining-related seismicity and what is natural seismicity in a mining district. One assumes, for instance, a natural origin for the 17 March 1816 Mansfield earthquake (4.2 ML), but recent low-magnitude seismicity in the Mansfield area has been of a mining character.” The attribution of 3.0 ML as an upper bound for mining induced seismicity is based on an informed but subjective choice, with a degree of circular logic, i.e. larger earthquakes were filtered out, as they were assumed not to be related to mining, leaving none above ML 3.0.
- In Table 12.10, the Cuadrilla report assesses the likelihood of a M=1.5 event as being ‘low’, but what does that mean? This vagueness is a bit annoying. If ‘low’ means one should expect to see this event at, say, only 1 in 100 fracking operations, and we now have close to two out of two operations exceeding this limit in the UK, then we could say clearly that the outcome is well above the initial assessment in the plan. This discrepancy highlights potential weaknesses in the methodology used to do the seismic hazard assessment and/or over confidence in the ability of the traffic light system to manage magnitude.
- The Cuadrilla report goes on to say “On this basis a 1.5 ML induced seismic event is considered to be the maximum magnitude event that could occur given the embedded mitigation measures that will be in place.” (Paragraph 147) This is a good statement because it is precise and therefore testable. Since the last frack stage on Monday, the current largest event has been M=1.1.
- Something that I think is key is in Paragraph 155 “If the monitoring indicates that a fault may be reacting to the hydraulic fracturing and showing signs of producing a seismic event greater than or equal to 0.5ML, then the pumping parameters may be amended (these parameters are constantly monitored) or the hydraulic fracture stage will be terminated early.” The rate and location of injection is the main thing that the operator has in its control to reduce the risk of induced seismicity. This is what they will be doing now following the M=1.1 earthquake on Monday. If/when operations restart we will be able to assess their skill in managing the magnitude of future events.
That is quite a lot more technical detail – but I think that the above are the important points for this discussion. There is also a lot more underling detail in Appendix L – but I have not ploughed through that yet.
My reading of the Cuadrilla Environmental Report is that is anticipated that the seismicity will be low, and that it considered operations would be rarely interrupted because few events would enter the Amber region, and very few in the Red under the traffic light system. Given several events in either category have occurred within a couple weeks of starting, the initial assessment was probably over-optimistic.
- The traffic light system implemented at Preston New Road
A few people have been asking me how the traffic light system works. This is the formal risk management system that must be used to manage the risks associated with inducing seismicity during the operation. This is in contrast to the pre-operation Environmental report quoted above, which sought to minimise the risks in advance of any operations.
In my previous post, I gave some of the history of where the traffic light system originates – namely to reduce the risk of roof collapses in coal mines. But how has it been adapted for projects that include fracking? Is the concept directly transferable? And how can we evaluate whether it is working in practice?
The flow chart from Appendix 4 of Cuadrilla a report titled “Hydraulic Fracture Plan PNR 1/1Z”is reproduced as a diagram below. This summarises how the traffic light management system works. In words, the traffic light system can be summarised as follows.
- GREEN: for seismicity with ML<0 operations can continue as normal
- AMBER: if an event occurs in the range 0<ML<0.5 while pumping the fracture stage, pumping can be completed for that stage. On completion of injection the flowback procedure will be initiated (where fluid is pumped back out from the subsurface to reduces fluid pressure with the aim of reducing the likelihood of further events).
- RED: if seismicity with ML>= 0.5 is observed during pumping operations a more complicated intervention is required. If the hypocenters are within the operational boundary, operators need to stop injection, reduce well pressure and pause injection for 18 hours. Once well integrity has been checked operations can recommence.
- TRAILING EVENTS: If an event occurs when pumping operations are not occurring, then it is defined to be a trailing event. Even if this event is greater than 0.5 – the operational plan does not require injection to be paused before restarting.
- In addition there is a criterion in the bottom right of the flow chart that states that if the peak velocity of ground shaking exceeds 15mm/s further interventions should be taken to reduce the risk of such shaking.
There are several points I will elaborate on here from my perspective as a statistical seismologist who works more commonly on natural seismicity.
- The traffic light system is mostly based on the size of the earthquakes that occur. In principle it provides a way to demonstrate skill in sustaining the size of events at a low level by changing the injection strategy and checking on the outcome. The only really hard limit imposed in the traffic light system below is if the recorded peak ground velocity exceeds 15mm/s. This threshold is related only indirectly to magnitude. The ground velocity is the speed at which the surface actually shakes and can be directly related to the chance of being felt or causing damage. The magnitude is a logarithmic measure of the energy released at depth – so you require modelling to convert magnitude and location to ground motion by propagating waves from the source of the earthquake to the surface.
- Trailing event are earthquakes that occur after well operations have ceased. I personally find this distinction a bit arbitrary because we know that after operations on a well have stopped – pressure will still diffuse and can initiate seismic events – in other words, we should expect some delays and lags between actions and consequences. If an earthquake occurs soon after operations have ceased it seems a bit arbitrary to suggest a different pathway through the traffic light system depending upon whether the event occurred just before or just after shutting the well.
- What is the justification for an 18 hour pause? I cannot pin this choice down exactly and it seems a bit unusual that it is quoted as a single magic number rather than a methodology that changes based on observable factors. If the estimation of the pause were related to modelling the physical processes, I would expect its duration to change as the physical conditions change. In part, it has likely arisen from the “Review and Recommendations for Induced Seismic Mitigation” produced by Green, Baptie and Styles following the Presse Hall earthquake. They state, “A reasonable period of time (12-24 hours) should also be allowed to elapse after the injection, to be sure that no seismic activity occurs as the fluid diffuses away from the wellbore. The monitored results should be fully considered, to allow determination of not only reservoir parameters, but also the in-situ stress, before the design of the main injection operation is finalised.” Whilst I have no evidence one way or the other, this number might have been chosen, at least in part, so that fluid injection/production could be carried out for 6 hours in a day and then restarted at the same time the following day. It would be great to see the integrated scientific case behind this choice.
- What risks is this system trying to manage? Risk management requires clearly defined metrics for success, defined in advance, so that we can test whether a risk management system is working. This transparent quantitative approach to risk management has led to the impressive safety record in the aviation industry for example. This is important because it is only through such ‘blind testing’ (i.e. making testable hypotheses before the outcome is known) that we can iteratively improve how risks are managed. I think my biggest criticism of the reporting in place around the traffic light system is that there is no clear statement regarding whether the intention is for it to (a) allow the operator to manage the event rate or (b) to allow the operator to manage the size of the largest events. Thus far, when fracking operations have recommenced both event rates and the size of the larger events have increased – so it is not yet achieving either. At the same time – the press releases state that the traffic light system is working ‘as intended’. This is true if you define working to mean that actions on the flow chart are being carried out as required by the regulator. However, I would argue that the traffic light system should only be considered working as intended if it is demonstrating some skill at reducing the risk. Since neither the rate nor the maximum magnitude are yet being limited – I think there is some way to go before it can be claimed that this approach has been proven to manage the risk.
4. What do I think?
The operators of Preston New Road have had to implement the traffic light system several times in the short lifetime of the project. This is down to two reasons. First, they have not been able to keep the magnitude of fracking induced earthquakes at the low levels observed in some parts of the US and Canada. Second, the thresholds for activating the traffic light interventions are conservative in the sense that they are lower than for comparable operations in other countries.
At the same time, the seismic hazard assessment carried out before injection started suggested the unfolding scenario had a ‘low likelihood’. The confidence implied by this statement highlights potentially weak assumptions in the assessment of the uncertainties involved. Vague statements have the advantage of allowing for wriggle room – but the industry needs to have a quantitative methodology with clear metrics for success, to assess whether the risk management plan is performing well. Just one more large event would mean that two out of two sites so far in the UK have performed poorly compared to expectations.
I believe it is necessary for the fracking industry to demonstrate skill in managing the magnitude of induced earthquakes against clearly-defined metrics of success, and that it is quite possible that the uncertainties introduced by the UK’s complicated geology could prove unmanageable.