Potential risks of induced seismicity from high volume hydraulic fracturing of shales in Northern Ireland

Hydraulic fracturing (HF) has made it possible to economically produce hydrocarbons directly from low‐permeability reservoirs such as shales by injecting high pressure fluids to create fracture networks. However, over the last decade the number of observations of induced earthquakes caused by HF operations around the world has increased as the shale gas industry has developed. Data from the US and Canada suggest that on average around 1% of HF wells can be linked to earthquakes with magnitudes of 3 or greater. Earthquakes of this size are large enough to be felt by people. However, in some areas of the US and Canada the percentage of wells associated with induced earthquakes is much higher (>30%). This variability is often explained in terms of geological factors such as proximity to existing faults. In a small number of cases, HF operations have triggered earthquakes large enough to cause potentially damaging ground motions. Such earthquakes cannot be confidently predicted in advance of operations. These observations suggest that the risk from induced seismicity during HF operations is not negligible. Earthquakes with magnitudes greater than around 2 result from slip on existing faults that is triggered by stress changes caused by the injection of fluid during the HF process. The size of the earthquake will depend on both the area of the ruptured part of the fault and the amount of slip. Since such faults may extend outside the hydraulically fractured zone, the maximum magnitude will be controlled by local geology and tectonics, not operational parameters such as the amount of injected fluid. As a result, the maximum magnitude is highly uncertain. Induced earthquakes have been observed in wide variety of geological settings and in areas where there are relatively few tectonic earthquakes. In some areas, the resulting hazard from induced earthquakes due to HF operations is significantly greater than the hazard from tectonic earthquakes. As a result, the low hazard from tectonic earthquakes in Northern Ireland does not guarantee that the hazard from induced seismicity will also be low. Induced earthquakes are likely to be clustered in space and time around the locus of HF operations. Hazard is likely to increase with the number of wells and will be highest during or shortly after HF operations. Hazard may also be a function of total injected volume, with larger injected volumes leading to more earthquakes and increasing the probability of larger events. Operations that target shallow formations may pose a higher hazard, since for a given magnitude, the intensity of ground motions at the surface will be greater. The potential for actual damage depends on the intensity of motions and both the number and vulnerability of buildings exposed to ground shaking. As a result, the risk of damage to buildings will be higher in densely populated urban areas than in rural areas. Risk studies for the UK have shown that cosmetic and minor structural damage may occur for earthquakes with magnitudes as low as 3. Higher resolution geophysical data is needed to identify fault structures and depth to basement in sedimentary basins with hydrocarbon potential in Northern Ireland in order to help mitigate risk of induced seismicity from hydraulic fracturing. Improved regional seismic monitoring should also be considered. Similarly, the present-day stress regime and stress state of faults in both the Lough Allen and Rathlin basins is poorly known. Further work is needed to address this. Current risk-mitigation strategies have had limited success. There may be insufficient data to identify geological faults prior to operations and even where high resolution data are available, there may still be hidden faults. Similarly, traffic light systems based on specific earthquake magnitude thresholds have often failed. Statistical methods that relate the volume of injected fluid or the injection rate to induced earthquake activity may allow useful probabilistic forecasts in the future but may be associated with considerable uncertainties without calibration for local conditions.