Estimation of Induced Seismicity Frequency-Magnitude Distributions Related to Fluid Injection
Abstract
Risks associated with induced seismicity (IS) are a significant factor in the design, permitting and operation of enhanced geothermal, geological CO2 sequestration and other fluid injection projects. The probabilistic seismic hazard analysis (PSHA) framework that is in widespread use to estimate hazard from naturally occurring earthquakes cannot be applied to IS without adaptations to address the differences between induced and natural seismicity. In particular, the normal PSHA assumption of stationary Poissonian statistics to describe seismic event occurrence are often shown to be invalid in engineered systems, as evidenced both by field observations and by theoretical considerations of the time- and space-dependent pore pressure perturbation of the reservoir. Here we discuss a physics-based approach to adapt conventional PSHA to provide the basis for IS risk analysis before the start of injection and to enable risk estimates to be progressively updated as data are gathered during and after injection. For the pre-injection analysis, earthquake frequency-magnitude distributions are generated by a numerical model that explicitly represents the known geologic structure within the vicinity of the reservoir. To illustrate the approach, we focus our analysis on CO2 sequestration applications. Using a realistic fault geometry, a von Karman spatial distribution of the coefficient of friction is generated along the fault surface using a flexible, hierarchical (quad-tree) representation. The fault geometry is then meshed and a far-field, constant strain rate boundary condition is applied to mimic regional tectonic loading. Pore pressure perturbations are derived from multi-phase simulations of CO2 plume evolution in the reservoir and used as a source function. The fault geometry, boundary conditions and pore pressure source term are then used in an alternating event-driven, explicit numerical simulation framework to generate the seismic sources and to evolve the stress distribution over the fault plane. In such an approach, the highest observable frequency is limited by the mesh element size. We evaluate parameter sensitivity by sampling multiple model realizations in order to explore reduced order models that enable fast calculation of the system response. The framework as well as the preliminary illustration of its application will be presented along with commentary on the benefits and drawbacks of the approach. This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFM.S41C2195J
- Keywords:
-
- 7230 SEISMOLOGY / Seismicity and tectonics;
- 7290 SEISMOLOGY / Computational seismology;
- 4302 NATURAL HAZARDS / Geological;
- 4314 NATURAL HAZARDS / Mathematical and computer modeling