Induced Earthquake Sequences in Geothermal Settings: Data Trends and Modeling Approaches

Abstract

Earthquakes induced by fluid injection are common in the context of reinjection into a producing hydrothermal system and well stimulation associated with creation of an Enhanced Geothermal System (EGS) as well as in other subsurface fluid activities including wastewater disposal, hydraulic fracturing, natural gas storage. Induced microseismicity can be a useful tool for characterizing the subsurface, but induced seismicity at moderate magnitudes may pose a significant hazard. Understanding the connection between fluid injection, subsurface pressure rise, and induced earthquakes is important for operating safely as well as minimizing financial risk due to project delays or cancellation. In this paper, we compare and contrast seismicity and injection rate data for five producing hydrothermal fields and five EGS stimulations. We show that reinjection operations in hydrothermal fields produce several orders of magnitude fewer earthquakes per mass of injected fluid than EGS stimulations. Our analysis also indicates that mean injection rate and total injected mass are poor predictors for the largest earthquake in the sequence. The number of earthquakes preceding the largest event, on the other hand, is a moderately good predictor, consistent with the assumption of a Gutenberg-Richter distribution of event magnitudes. On the basis of this analysis, we propose that numerical modeling of earthquake sequences – sets of events of varying magnitude, location and timing – is a potential method for constraining the likelihood of large magnitude events. To this end, we describe initial development of a quasidynamic earthquake simulator suitable for inclusion within a reservoir simulation framework. We model evolving rupture on a 2D planar fault using a rate-and-state friction law and a spectral method to calculate slip-induced stress changes. To demonstrate the approach, we simulate effective normal stress reduction due to injection-induced overpressure on a fault with arbitrary spatial heterogeneity in friction parameters. For a six day injection period, seven discrete episodes of fault slip are produced.