| Title | Quantifying EGS Permeability Development Using Induced Seismicity |
|---|---|
| Authors | Jeremy RIFFAULT, David DEMPSEY, Rosalind ARCHER, Satish KARRA |
| Year | 2018 |
| Conference | Stanford Geothermal Workshop |
| Keywords | egs, stimulation, induced seismicity, Habanero, permeability |
| Abstract | In Enhanced Geothermal System (EGS) projects, high pressure injection of cold water is conducted with the aim to artificially enhance in-situ permeability. Assessing the spatial distribution of this permeability enhancement is critical in an EGS setting, as it indicates the success of these stimulation operations, and gives valuable information for future development of the project. Such stimulations typically induce large seismicity clouds, the origin of those micro-earthquakes being mainly attributed to sliding along preexisting cracks caused by pore pressure increase. Therefore, the distribution and intensity of the microseismicity cloud makes it possible to track fluid pressure evolution, which in turn reveals aspects of the permeability distribution and its enhancement. Using this idea, we develop a method to invert permeability development based on the distribution of seismicity in a 2D plane, allowing for simulation of non-azimuthal symmetry. We design a permeability assignment model where each node has its own semi-independent permeability regime. The density of earthquake hypocenters is derived from the pressure distribution obtained from a flow simulator. An inverse problem is set up in which a synthetic permeability enhancement regime is created and which we then attempt to recover from simulated microseismicity, accounting for both location and Poisson errors. While the inversion successfully recovers much of the true permeability regime, we find that the inclusion of post-stimulation seismicity is critical to constrain hydrological conditions in the final moments of the stimulation. We also note that the presence of uncertainties in the earthquake observations leads to a discrepancy between the best fitted model and the real parameter set the parameter set; the range of valid solutions should be chosen carefully. Overall the proposed method is able to constrain the spatial distribution of permeability enhancement. |