Affiliation:
1. University of Southern California
Abstract
AbstractThe activation mechanism of Irving-Dallas events is not well understood as it is shrouded in ambiguity due to many earthquakes located relatively far (>15 km) from production and injection wells. This requires a modeling approach that can quantify spatiotemporal propagation of production- and injection-induced stresses from wells to the faults while resolving fault geometry, stratigraphy, and well activity. However, constructing one such detailed model for the entire basin is computationally prohibitive due to the millions of grid cells needed to discretize the basin at that resolution. Based on our analysis of the data on well activity and fault position, we employed a novel two-model approach that exploits the disparity in scales between the basin-scale injection analysis and the well-scale fault reactivation analysis. We construct a coarse-scale model of Ellenburger injection in the Fort Worth basin and a fine-scale flow-geomechanics model of the Dallas-Irving region containing the faults that hosted the seismicity and the production/injection wells in the region. We use the coarse model to provide time-dependent pressure boundary conditions to the fine-scale model. We analyze the spatiotemporal evolution of pressure fields at both basin and reservoir scales. Analysis of the results provides evidence for interaction between Barnett's production and Ellenburger's injection as well as pressure diffusion from Ellenburger into the basement along the through-going faults. It allows us to test the hypothesis of injection-induced reactivation as the causative mechanism for the Irving seismic events. Almost all injection-induced seismicity studies in the literature show how injection near a fault (well-to-fault distance < 10 km) can induce seismicity. We provide evidence of far-field injection-induced seismicity (well-to-fault distance > 80 km) by coupling basin-scale and reservoir-scale models and a multi-physics approach.