Seismicity induced by fluid injection in faulted geothermal reservoirs have been reported worldwide. Although it is clear that fluid injection plays a crucial role in triggering fault slip and seismicity, the correlation between fluid injection-induced seismicity and underlying fault structures remains an active topic of research. In this study, induced seismicity recorded surrounding five injection wells at Hellisheidi geothermal field in Iceland over a half-year fluid injection period was first analysed using the k-means clustering model, which identified two clusters associated with the injection wells studied. Based on the clustered seismic events, a statistical approach, using locations of three seismic events to identify one fracture plane, was utilised to identify the dominant fault structures with dip angle and dip azimuth attributes, which spatially correlate well with local fault structures dominating the fluid flow. To be specific, the fluid injection rate and wellhead pressure for the five wells were separately analysed against the seismic-derived fault structure attributes using a rolling time window method, which shows a good correlation between fluid injection and fault structures in both spatial and temporal domains. Such seismic observations and fault structures were then integrated into a numerical reservoir modelling, which was calibrated first by history matching of bottom hole pressure time series at five injection wells. In this modelling, a fracture criticality, which refers to the gradient of critical pore pressure change to trigger seismicity, was proposed to achieve the probabilistic distribution of injection-induced seismicity in both fault and off-fault zones. The results show that relatively high probability of seismic event occurrence was estimated for fault zones around five injection wells, which were consistent with seismically-active areas over the seismic monitoring period. It can be concluded that the integrated analysis of induced seismicity, fault structures and fluid injection in spatial-temporal domains is crucial to understand the correlation between fluid injection and induced seismicity in faulted geothermal reservoirs.

induced seismicity,fluid injection,faulted geothermal reservoir,numerical modelling,fracture criticality