Quantifying the variability in fault density across the UK Bowland Shale, with implications for CCS-induced seismicity hazard in the UK Continental Shelf

The occurrence of induced seismicity during hydraulic fracturing led to the shut-down of the UK’s nascent shale gas industry. Various low carbon energy technologies that involve the injection of fluids into the subsurface, such as Carbon Capture and Storage (CCS), deep geothermal energy, and subsurface hydrogen storage also carry the potential to cause induced seismicity. Hence, there is a need to develop methods to characterize the expect seismic hazard that might be generated by the development and operation of such facilities.
Deterministic methods for characterization of induced seismicity have previously struggled to produce accurate a priori estimations of induced seismicity hazard. In particular, due to faults that can go undetected in 2D or 3D seismic surveys if they are shorter that the resolution retrievable from a seismic survey, or if they have low (and in some cases even zero) vertical displacement.
Instead, we develop a probabilistic method to characterize induced seismicity hazard from reflection seismic observations based on the seismogenic index, SI, which relates the number of induced earthquakes to the injected volume (Shapiro et al., TLE 2010): 10^(S_I )=bM+N_E⁄V. This relationship can be used to generate probabilistic forecasts of the maximum magnitude that will be induced after the injection of a given fluid volume. SI can also be related to the density of faulting within the volume perturbed by injection: 10^(S_I )=10^a.F/C.S, where a is the background tectonic Gutenberg-Richter a value, F is the fault density, C is the average critical stress change required to activate the faults, and S is the storativity of the formation. For a given formation or play, it is reasonable to assume that changes in a, C and S will be relatively minor, in which case variations in SI will be controlled by changes in fault density across the play.
We demonstrate this approach by application to hydraulic fracturing-induced seismicity in the Bowland Shale area in central Britain. We first obtained 3D reflection seismic cubes along an east-west axis across northern England and used an automated fault detection algorithm to map faults within lower Carboniferous strata. We then estimated the slip potential by resolving the formation’s stress and pore pressure conditions (with the Bowland shale being significantly over-pressured) onto each fault. The abundance of critically stressed faults varied significantly across the play, with a regional reduction in the intensity of mapped faults from west to east by as much as an order of magnitude.
We use these observations to inform a probabilistic assessment of seismic hazard, beginning by using the observed HF-IS in the Bowland-12 area to create an empirically-constrained baseline model, and then use the observed differences in fault densities to create an updated SI distribution that may be more appropriate for the less faulted regions to the east. This updated model shows that induced seismicity of sufficient magnitude to be felt could still be generated by hydraulic fracturing in these regions, however their likelihood of occurrence is reduced by an order of magnitude.
Finally, we follow a similar approach to characterise and update the induced seismicity hazard as activities such as CCS and geothermal energy are rolled out across the UK Continental shelf, particularly in the East Irish Sea and Mid North Sea located nearby the Bowland area.