Recent scientific advances in the understanding of induced seismicity from hydraulic fracturing of shales

The Secretary of State for Business, Energy & lndustrial Strategy has commissioned the British Geological Survey to write a short report about seismic activity associated with hydraulic fracturing (HF) of shales to extract hydrocarbons. The specific terms of reference are available at https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/fi le/1066525/BGS_Letter.pdf. These ask six questions related to recent scientific research on the hazard and risk from induced seismicity during hydraulic fracturing of shale rocks. Our report considers the scientific advances in this area since 2019 that have been published in peer reviewed scientific journals as well as other recent studies commissioned by regulatory authorities. The main conclusions of our report in relation to each of the questions in the terms of reference are as follows: Forecasting the occurrence of large earthquakes and their expected magnitude remains a scientific challenge for the geoscience community. This is the case for both tectonic and induced earthquakes. (Questions 1 and 2) Methods to estimate the maximum magnitudes of induced earthquakes based on operational parameters and observed seismicity have been tested using data from both Hydraulic Fracturing (HF) operations and data from other industries. These methods have shown some applicability to guide operational decisions using real-time data. However, they do not currently account for the possibility of events that occur after operations have stopped or earthquakes on faults that extend outside the stimulated volume whose magnitude is not controlled by operational parameters alone. (Questions 1 and 2) Probabilistic methods widely applied to model and forecast tectonic earthquake sequences show some promise when modified to incorporate information about HF operations and appear capable of providing informative forecasts of the observed earthquake patterns. Operators could make forecasts for operations in new wells using either generic parameters or ones calibrated for operations in adjacent wells. Further testing of these methods may allow them to be further developed for operational scenarios. (Questions 1 and 2) Enhanced seismicity monitoring and measurement based on machine learning (ML) has been shown to reveal previously undetected earthquakes and hidden faults, essential for both more reliable earthquake forecasts and characterisation of fault reactivation potential. This can compensate for both limited numbers of seismic stations and faults that remain unmapped even by 3D exploration seismic data. (Questions 1 and 2) Widely used probabilistic methods to assess hazards and risks for tectonic earthquakes can also be applied to induced seismicity. However, there are important differences between how tectonic and induced seismicity evolves in space and time. Recent studies have suggested possible solutions, but further work is needed to develop these models and incorporate them in risk assessments. (Questions 1 and 2) Traffic light systems remain a useful tool for the mitigation of risks from induced seismicity. New research shows how red-light thresholds can be chosen to reduce the probability of the scenario to be avoided to a required level. This research recommends that there should be sufficient space between the amber and red-light thresholds to ensure that operators have an opportunity to modify operations to mitigate risks. (Questions 1 and 2) Induced seismicity has been observed in other industries related to underground energy production both in the UK and elsewhere. In the absence of a seismic building code in the UK, consistent risk targets, i.e., scenarios to be avoided, could be considered for all energy related industries that present a risk of induced earthquakes. (Question 3) Recent research using high quality exploration data that is available for some parts of the UK reveals localised structural and stress heterogeneity that could influence fault reactivation. However, it is not possible to identify all faults that could host earthquakes with magnitudes of up to 3 prior to operations, even with the best available data. (Questions 4 and 5). Recent research from the USA demonstrates the importance of geomechanical modelling to identify faults that are most likely to rupture during operations. This information can be used to assess risks prior to and during operations. However, these models require accurate mapping of sub-surface faults, robust estimates of stress state, and knowledge of formation pore pressures and the mechanical properties of sub-surface rocks. While this information is available in areas with unconventional hydrocarbon potential such as the Bowland Basin, more data is needed from other basins to apply this more widely (Questions 4 and 5). Limited exploration data from other basins with unconventional hydrocarbon potential of the UK means that there are significant gaps in our knowledge of sub-surface structure of potential shale resources in these places. (Questions 4 and 5) The rates of HF-induced seismicity in other countries where shale gas production has been ongoing for many years are observed to vary widely. The limited number of HF operations in the UK means that it is difficult to make a valid comparison of the rates of occurrence of induced seismicity with elsewhere. This underlines the importance of knowledge exchange in monitoring and operational practices. (Question 6) Our review focusses on recently published geoscience related to induced seismicity caused by HF of shales. Ongoing and future research may bring new insights that may reduce uncertainties and improve mitigation of risks. We did not consider socio-economic research on perception of risks or the benefits of shale gas. Similarly, we do not consider technological advances in hydraulic fracturing.