Topic: Scaling Permeability Evolution Driven by Seismicity
Speaker: Derek Elsworth
Triggered seismicity is a key hazard where fluids are injected or withdrawn from the subsurface and may also impact permeability. We constrain maximum event magnitudes in triggered earthquakes by relating pre-existing critical stresses to fluid injection volume – to explain why some recorded events are significantly larger than anticipated seismic moment thresholds. This formalism is shown consistent with a number of uncharacteristically large fluid-injection-triggered earthquakes. Such methods of reactivation of fractures and faults through hydraulic stimulation in shear or in tensile fracturing are used routinely as a method to create permeability in the subsurface. Microearthquakes (MEQs) generated in such stimulations may be used as diagnostic of permeability evolution. Although high fidelity datasets are meager, the EGS-Collab (@SURF) and Utah FORGE demonstration projects both provide high fidelity data sets that concurrently track permeability evolution and triggered seismicity. Machine learning deciphers the principal features of MEQs and the resulting permeability evolution that best track changes in permeability – with transfer learning methods allowing robust predictions across multiple geological settings. Changes in permeability on reactivated fractures in both Mode I and II suggest that permeability () scales with the seismic moment (
) of individual MEQs as
. This scaling relation is exact at early time but degrades with successive MEQs but presents a method for characterizing crustal permeability evolution using MEQsTriggered seismicity is a key hazard where fluids are injected or withdrawn from the subsurface and may also impact permeability. We constrain maximum event magnitudes in triggered earthquakes by relating pre-existing critical stresses to fluid injection volume – to explain why some recorded events are significantly larger than anticipated seismic moment thresholds. This formalism is shown consistent with a number of uncharacteristically large fluid-injection-triggered earthquakes. Such methods of reactivation of fractures and faults through hydraulic stimulation in shear or in tensile fracturing are used routinely as a method to create permeability in the subsurface. Microearthquakes (MEQs) generated in such stimulations may be used as diagnostic of permeability evolution. Although high fidelity datasets are meager, the EGS-Collab (@SURF) and Utah FORGE demonstration projects both provide high fidelity data sets that concurrently track permeability evolution and triggered seismicity. Machine learning deciphers the principal features of MEQs and the resulting permeability evolution that best track changes in permeability – with transfer learning methods allowing robust predictions across multiple geological settings. Changes in permeability on reactivated fractures in both Mode I and II suggest that permeability (
) scales with the seismic moment (
M) of individual MEQs as
. This scaling relation is exact at early time but degrades with successive MEQs but presents a method for characterizing crustal permeability evolution using MEQs.
Derek Elsworth
Derek Elsworth is G. Albert Shoemaker Chair, Professor of Energy and Mineral Engineering and Geosciences and co-director of the Center for Geomechanics, Geofluids, and Geohazards at Penn State. His interests are in the areas of computational mechanics, rock mechanics, and in the mechanical and transport characteristics of fractured rocks, with application to geothermal energy, the deep geological sequestration of radioactive wastes and of CO2, unconventional hydrocarbons including coal-gas, tight-gas-shales and hydrates, and instability and eruption dynamics of volcanoes.
