Deriving Geomechanical Constraints from Microseismic Monitoring Demonstrated with Data from the Decatur CO2 Sequestration Site

Abstract:

The occurrence of induced and triggered microseismicity is of increasing concern to the general public. The underlying human causes are numerous and include hydrocarbon production and geological storage of CO₂. The concerns of induced seismicity are the potential hazards from large seismic events and the creation of fluid pathways. However, microseismicity is also a unique tool to gather information about real-time changes in the subsurface, a fact generally ignored by the public. The ability to detect, locate and characterize microseismic events, provides a snapshot of the stress conditions within and around a geological reservoir. In addition, data on rapid stress changes (i.e. microseismic events) can be used as input to hydro-mechanical models, often used to map fluid propagation.

In this study we investigate the impact of microseismic event location accuracy using surface seismic stations in addition to downhole geophones. Due to signal-to-noise conditions and the small magnitudes inherent in microseismicity, downhole systems detect significantly more events with better precision of phase arrival times than surface networks. However, downhole systems are often limited in their ability to obtain large enough observational apertures required for accurate locations. We therefore jointly locate the largest microseismic events using surface and downhole data. This requires careful evaluation in the weighting of input data when inverting for the event location. For the smaller events only observed on the downhole geophones, we define event clusters using waveform cross-correlation methods.

We apply this methodology to microseismic data collected in the Illinois Basin-Decatur Project. A previous study revealed over 10,000 events detected by the downhole sensors. In our analysis, we include up to 12 surface sensors, installed by the USGS. The weighting scheme for this assembly of data needs to take into account significant uncertainties in the near-surface velocity structure. The re-located event clusters allow an investigation of systematic spatio-temporal variations of source parameters (e.g. stress drop) and statistical parameters (e.g. b-value). We examine these observations together with injection parameters to deduce constraints on the long-term stability of the injection site.