18.05.2021: Pre-existing fracture corridors as a permeable pathway leading to fault reactivation in the Duvernay Shale

In a new paper released in JGR, we have assessed the mechanisms for fault reactivation and induced seismicity in the Duvernay Shale, Alberta. Microseismic monitoring with a dense surface array revealed the reactivation of pre-existing fracture corridors that intersected the wells as hydraulic fracturing took place. As fracturing continued, induced seismicity began to occur on a fault that did not intersect the wells, which leads to the question - how did the fracturing reactivate the fault? The timing of the microseismicity and induced seismicity was crucial: we found the a time delay of 40 - 90 hours between the reactivation of each fracture corridor, and induced seismicity on the portions of the fault aligned with each corridor. Comparisons with fluid flow simulations showed that this delay time is entirely consistent with the propagation of fluids pressures along the fracture corridors to intersect the more-distant fault.

These observations show the importance of understanding the pre-existing permeability structures in a reservoir when attempting to understand the impacts of hydraulic fracturing on the subsurface.

29.10.2020: Microseismic imaging of fault reactivation at the Preston New Road, Lancashire, site in August 2019

On the 26th August 2019, a magnitude 2.9 earthquake was triggered by hydraulic fracturing at the Preston New Road site in Lancashire. This event was felt in the nearby towns of Blackpool and Preston, and led to a decision by the UK government to place a moratorium on further shale gas hydraulic fracturing. In a new paper released this week in Seismological Research Letters, we provide the first detailed study of how the hydraulic fracturing operations led to the reactivation of a fault that generated this event.

We use detailed microseismic monitoring observations made using a downhole geophone array. The microseismic events reveal the propagation of hydraulic fractures from the well. During the early stages of stimulation, fractures propagated outwards from the well - while they intersected some pre-existing fractures, these did not generate any appreciable seismic activity. However, as stimulation proceeded along the well, a new set of hydraulic fractures developed. As these propagated southwards, they encountered a pre-existing fault on which the largest seismic events occurred. This fault was about 200 m from the well. Because of the time required for pressures to travel across this distance, most of the seismicity occurred after each injection stage had been completed. The M 2.9 earthquake occurred 72 hours after injection of Stage 7 had been completed on the 23rd August

We used the "seismic efficiency" method to forecast earthquake magnitudes during the PNR operations. This method compares the rate at which seismicity is generated to the injected volume, from which we can extrapolate a "final" event population. However, the delays between injection and the occurence of seismicity posed a challenge for this approach. After Stage 6 on the 21st August, our forecast for the largest expected event was M 2.5. In response to these forecasts, the operators reduced the injection volume, and increased the fluid viscosity, in order to reduce the size of the hydraulic fractures generated, and thereby prevent further reactivation of the fault. However, this change was clearly unsuccessful, as further reactivation did take place after the reduced-volume Stage 7, leading to the M 2.9 event. Evidently, this event was slightly larger than our forecast, but the actions of the operator nevertheless show these forecasting methods can be used to guide decision-making during hydraulic fracturing, albeit that the modifications made by the operator did not in this case prevent further seismicity from occurring.

We also present some other observations of interest. Stimulation of the adjacent PNR-1z also caused induced seismicity in October/December 2018. The two wells are about 200 m apart. However, the microseismicity shows that the two wells reactivated completely different faults - there is no overlap between the volumes of rock stimulated by the two wells. We compared the stress conditions acting on the two faults. The PNR-1z fault received a much higher volume of fluid, it directly intersected the PNR-1z well, and yet it produced an order of magnitude less seismicity, and it reactivated very gradually. In contrast, the PNR-2 fault received less fluid volume, the fault is 200 m away from the well, and yet it produced more seismicity, and the reactivation took place relatively rapidly. These different behaviours can be understood in terms of their orientations in the stress field. The PNR-1z fault is orientated such that it has a moderate level of shear stress acting on it, whereas the PNR-2 fault is optimally orientated in the stress field such that shear stress is maximised. This may explain why it was able to reactivate so readily, and shows the importance of the in situ stress field in controlling fault reactivation behaviour.

03.05.2020: Three new papers examining the use of fibre-optic DAS sensors for microseismic monitoring

The use of fibre-optic cables as distributed acoustic sensor (DAS) arrays offers significant potential as an alternative to borehole geophones for microseismic monitoring. Cables are installed in boreholes, and light pulses fired down the cables. Back-scattered light is recorded at the interrogator, and phase changes in this light provide a measurement of the change in length, or strain, along the fibre. DAS arrays are cheaper to install and more robust than geophones, and can be permanently installed behind well casing. Their dense channel spacing across a long cable provides a high fold, wide aperture sensing system. However, the use of DAS arrays also presents challenges. Very high data volumes are generated, which can pose a challenge for real-time microseismic processing. DAS arrays record a single component of motion (strain-rate along the cable axis), which can create angular ambiguities in event location estimates, and may make inversions for source mechanisms more challenging.

In a series of papers published this month in Geophysics, we have addressed these challenges. Stork et al. (2020) develop a machine learning algorithm capable of rapidly detecting events amongst the high data volumes generated by DAS systems. Baird et al. (2020) examine the affect of seismic anisotropy and source mechanisms on the signals that will be recorded on a 1C DAS array, and in particular show that shear-wave splitting may prove to be a useful tool for resolving angular ambiguities. Finally, Verdon et al. (2020) provide a case study, showing the use of DAS arrays in action to monitor microseismicity during hydraulic fracturing.

11.01.2020: New paper examining the geomechanical controls on fault reactivation at Preston New Road

Following on from our preliminary study of the PNR-1 hydraulic fracturing, today our study into the geomechanical controls on induced seismicity was published in the Journal of Geophysical Research. We used the high-quality microseismic data to examine the processes by which the hydraulic fracturing caused the faults identified in our 2019 SRL paper to reactivate. Typically, the mechanism by which faults are reactivated is because of the increased fluid pressure in the system produced by the stimulation. However, the tensile opening of the hydraulic fractures themselves will also cause deformation in the surrounding rocks - producing a change in the stress field. This change in stress field can also help push the fault towards failure, allowing it to slip more easily. It can also act in the opposite way, clamping the fault and preventing it from slipping. The effect it has will depend on the relative orientations and positions of the fault and the hydraulic fracturing. At PNR we found that the majority of induced events occurred in the parts where the tensile opening was pushing the fault towards failure, and did not occur in the parts where tensile opening was clamping the fault. From this, we infer that the deformation produced by tensile fracture opening, rather than the increased fluid pressure, was playing the dominant role in reactivating the fault.

This matters because we need to understand how faults are reactivated during hydraulic fracturing, so that operators can develop ways to prevent it happening as their stimulations progress. The type of modelling we used in this paper is relatively simple and can be applied in real time while the well is being fractured, potentially allowing operators to redesign their stimulation program as the data and observations come in.

A copy of the paper is available here

14.08.2019: New paper forecasting induced seismicity at the Preston New Road hydraulic fracturing site

The Preston New Road PNR-1 well is the first to be hydraulically fractured since the government review into induced seismicity and hydraulic fracturing in 2012. As a site of high public interest, the well was extensively monitored, including for induced seismicity. The fracturing was subject to the UK's Traffic Light Scheme for induced seismicity, where the red-light threshold is set at a magnitude of 0.5, at which point operations must pause for 18 hours.

The TLS approach to mitigating seismicity is retroactive, because actions are taken after an earthquake occurs. Instead, if seismicity can be forecasted, then a more proactive approach can be taken based on model forecasts. In our new paper published in SRL, we demonstrate such an approach, which we applied in real time at the PNR site.

We used data from the downhole microseismic array, which recorded over 38,000 events with magnitudes down to -2. Analysis of the spatio-temporal evolution of the events revealed the interaction between the hydraulic fractures and a pre-existing fault. We populated a statistical model for the seismicity based on the observed events, and extrapolated this model to make forecasts of the expected earthquake size as injection proceeded. This approach was successful in quantifying the magnitudes of events that did in fact occur during stimulation. The successful application of this method demonstrates that a more proactive approach to mitigating induced seismicity should be possible

A copy of the paper is available here

17.04.2019: New paper discriminating natural and induced earthquakes

As the number of cases of induced seismicity grows, and as public awareness of the issue rises, it is clear that objective methods are needed to determine whether a particular sequence of earthquakes is natural, or has been induced by industrial activity. These schemes must provide a rapid assessment that is comprehensible to stakeholders with a range of knowledge levels. These schemes must also be capable of dealing with evidence that is less than certain, and providing a measure of how robust an assessment is. In a new paper accepted today in Seismological Research Letters we do just that.

Our new scheme asks a series of questions about the observed seismicity and its relationship with nearby activities. Each question may produce an answer that points to a natural cause, or an induced cause. We score minus points for a natural cause, and positive points for an induced cause, with the number of points scored being scaled by the significance of the answer. Additionally, if the evidence for a given question is uncertain, we reduce the number of points scored accordingly. If no evidence is available, the question is not answered, scoring 0 points. Summing the overall points from all the questions, gives us the Induced Assessment Ratio (IAR), which is positive if the event is induced, and negative if natural. Additionally, the number of points that can be scored given the available evidence is compared with the number of points that could be scored if perfect evidence was available. This gives us our second parameter, the Evidence Strength Ratio (ESR) that classifies how well-evidenced the assessment is.

We have applied this scheme to two UK case studies: Preese Hall in 2011, and the Newdigate, Surrey sequence in 2018. To demonstrate the strengths of the scheme we apply it at two times for each case - at an early time when data and information was limited, and at a second time when data was more robust.

Our intention is that this scheme will become widely adpoted for cases where the attribution of induced or natural seismicity is contentious. A link to the paper is available here

02.02.2019: New paper on microseismic magnitudes

Many jurisdictions, including the UK, have regulations for induced seismicity that are based on event magnitudes. For example, current UK regulations state that hydraulic fracturing operations must pause if events exceed ML = 0.5. However, calculating event magnitudes is not a trivial exercise, especially for small-magnitude events, where signal-to-noise ratios are low, and monitoring stations must be placed in close proximity to the source location. In this paper we review some of the challenges for magnitude calculation of small events in industrial settings, and make recommendations that operators and regulators can follow to ensure that accurate magnitude estimates are used for regulatory decisions. The full paper is available here

03.12.2018: OGA Workshop on the Newdigate earthquake sequence, Surrey

Between April and August 2018, a series of small-magnitude earthquakes were felt in the area to the south of Dorking, Surrey. Public concerns were raised as to whether the events were caused by oil and gas activities in the area, specifically either the Brockham field, which has been under production for roughly 15 years, or the Horse Hill well in which flow testing was being conducted. Our group, alongside a range of other stakeholders and academics, was asked by the OGA to investigate these events, and any link to oil extraction activities in the area. Our conclusion was that the events were natural, and not induced. Our report is available here. The OGA convened a workshop in October 2018 where the various parties discussed their findings. The overall conclusions of the group match the conclusions made by our group, that the events were natural and not induced. The full details of the OGA Workshop are available via their website here.

18.07.2018: Reports on hydraulic fracturing, coal mining and respect distances for the OGA

In recent months, concerns have been raised about hydraulic fracturing in proximity to faults, and about hydraulic fracturing in areas that have experienced historic coal mining activities. In response, we were asked by the Oil and Gas Authority (the relevant regulator in this regard) to investigate these issues. We have produced two reports, examining the evidence for hydraulic fracture respect distances here, and investigating the impacts of longwall coal mining on stress conditions in underlying shale layers here.

29.06.2018: NERC Unconventional Hydrocarbons in the UK Energy System Funding Awards

Bristol scientists have been awarded a considerable chunk of funding in the recent NERC Strategic Program Area on Unconventional Hydrocarbons. This program aims to provide the scientific evidence base to evaluate potential economic, environmental and social impacts of unconventional gas exploration in the UK. The program set 5 "Challenges", and Bristol is involved in 3.

Challenge 2, lead by Al Fraser at Imperial College London, will investigate the resource potential of the Bowland Shale. James Verdon leads the geophysical component of Challenge 2, with Mike Kendall as Co-I, investigating how seismic techniques might be used to identify "sweet spots" in shale rocks.

Challenge 3, lead by Mike Kendall at Bristol University, with James Verdon as Co-I, will study coupled processes in the overburden between the shale layers and the near surface, with a particular focus on pathways for fluid migration.

Challenge 4, lead by the BGS, will address contaminant pathways and receptor impacts, including seismic hazards and risks. Max Werner leads the seismic hazard assessment element of this challenge.

Cumulatively these 3 research challenges comprise over £5 million of funding, with nearly £1 million coming to Bristol.

08.06.2018: TLE Special Section on Geomechanics

May's edition of The Leading Edge features a Special Section on Geomechanics coordinated by Yongyi Li, Doug Foster, Marisela Sanchez-Nagel, Mark Tingay and myself.

10.05.2018: Funding from RDPetro

We're delighted to announce the awarding of a new grant from RDPetro entitled "Passive seismic monitoring of Abu Dhabi reservoirs: detecting production-induced fracturing and hydraulic fracturing". We will work with Abu Dhabi's Khalifa University of Science and Technology to explore the use of microseismic monitoring in both conventional and unconventional gas fields in the emirate.

14.03.2018: Welcoming new students to the group

We're very pleased to welcome two new Ph.D students to the group. Jessica Arellano joins us from Mexico, funded by Conacyt, to study the potential for Carbon Capture and Storage in Mexico.

Nadine Igonin is visiting for 5 months from Calgary University to work on induced seismicity during hydraulic fracturing.

27.02.2018: New paper on forecasting earthquake magnitudes during hydraulic fracturing

This week in BSSA (Bulletin of the Seismological Society of America), a new paper by James Verdon, in collaboration with Jessica Budge of Nexen, examines the use of statistical models to forecast induced earthquake magnitudes during hydraulic fracturing.

We have found that, by parameterising seismicity with respect to injection parameters, microseismicity recorded during the early stages of injection can be used to forecast how the seismicity will develop as the stage progresses. This will provide a useful mitigation tool to operators looking to prevent induced seismicity from occurring during their operations.

19.02.2018: Cwmllynfell (Wales) Magnitude 4.4 Earthquake felt in Bristol

On Saturday 17th Feb, the magnitude 4.4 event in South Wales was felt widely across the Southwest and the Midlands. The UK experiences a quake of this size approximately once every 2 - years.

Fortunately, at the time of the event we had a number of seismometers recording for testing purposes, so here is what the Cwmllynfell earthquake looked like in the Wills Memorial Building: