Seismicity induced by fluid injection in faulted geothermal reservoirs have been widely documented worldwide, where the fracture networks play a crucial role in bridging fluid flow and induced seismicity. The use of seismic monitoring data for the inversion of fracture networks remains an active topic of research to understand the correlation between fluid injection and induced seismicity. In this paper, a robust random sample consensus (RANSAC) method integrated with the alpha-Shape model has been developed to construct the discrete fracture networks (DFN) based upon seismic monitoring data. To calibrate the capacity of this method, a series of numerical simulation tests were carried out using Monte Carlo method to generate seismic events in the randomly simulated fracture planes, along with a certain number of noise events outside the fracture planes. Such seismic and noise events were then utilized to inverse the random fracture planes by the robust RANSAC model. To further validate the method, induced seismicity recorded at the Hellisheidi geothermal field in Iceland was fed into the RANSAC model to identify the fracture networks, which spatially correlate well with local fault structures dominating the fluid flow. Based on the seismic-derived DFN abovementioned, the seepage pipe was established by connecting the centroid of each fracture and the centre of the intersection line with another adjacent fracture. In this context, the distribution of head pressure in the DFN was simulated using Darcy’s law under the realistic in-site fluid injection locations and behaviours. The results show that this seismic-derived DFN method could be a prevailed proxy to investigate and understand the correlation between fluid injection and induced seismicity in faulted geothermal reservoirs.

induced seismicity,fluid injection,discrete fracture network,faulted geothermal reservoirs,RANSAC method