Induced seismicity is a hazard associated with subsurface fluid injection and extraction. To advance understanding of the mechanism causing induced seismicity, a large laboratory-scale experiment was conducted in which fluid injection was used to activate slippage along a simulated fault. This work is part of a larger study conducted to understand the induced-fault motion and related effects. A 1 m3 Castlegate sandstone block was saw-cut diagonally ( 45°) into halves, simulating the fault. The two block halves were mounted in a stress frame, then varying stress was applied over time and fluid was injected via an injection borehole at 22 separate time points until frictional sliding occurred along the simulated fault plane. Pressure was monitored in three boreholes: two terminating close to the injection borehole and one terminating at the cut plane. Acoustic emissions (AEs) were monitored via 36 acoustic sensors set on the outside surface of the blocks. In addition, elevations of surfaces along the simulated fault plane were profiled by laser scanning pre- and post-experiment. Post-test, the fault surfaces were characterized, laser scan (LS) data were analyzed, and clusters of locatable AEs were analyzed relative to time, pressure, and the applied confining stresses. Geologic characterization revealed that a very fine grained sandstone material, described as "fault gouge," developed along the fault surfaces after frictional sliding. Analysis of grain sizes in the gouge relative to those in the pristine Castlegate sandstone showed a significant reduction (> 50%) in size. Topographic maps of the fault surfaces rendered using digital elevation data collected pre- and post-experiment portrayed rougher fault surfaces after slippage. Furthermore, analysis of AEs relative to injection time intervals, stress, and pressure indicated that the AEs were induced by both stress and injection. Time intervals of injections 1 to 20 showed good correlation between AEs (peak of 10 events/second) and differential stress over time, whereas AEs of intervals 21 and 22 showed good correlation between AEs (peak of 30 events/second) and injection. Clusters of AEs were observed to migrate along and across the fault through time, partly because the fault surfaces were uneven, as depicted in pre-experiment images rendered from the LS data.
|Original language||English (US)|
|Title of host publication||AGU Fall Meeting Abstracts|
|State||Published - 2018|