Revisiting 2013–2014 Azle seismicity to understand the role of Barnett production on stress propagation and fault stability

Abstract

Propagation of production- and injection-induced stresses from reservoir layers to basement faults is not well understood, especially in reservoirs with hydraulically segregated production and injection units situated across low-permeability structural or stratigraphic boundaries. This is one of the challenges in understanding the 2013–2014 Azle-Reno earthquake sequence (maximum magnitude M w 3.8) in the Fort Worth Basin of Texas, which occurred after several billion cubic meters of gas and several million cubic meters of brine were produced from the Barnett Shale and several million cubic meters of water was injected into the Ellenburger Formation below the Barnett. The effect of Barnett production on induced stress and reactivation of the synthetic-antithetic fault pair hosting the seismicity is not well understood from prior studies. The interaction among production-induced contraction, injection-induced expansion, and decrease in fault compression due to fluid diffusing from the injection layer strengthens two-way coupling among fluid flow, deformation, and fault failure processes. Using a high-resolution coupled flow-geomechanics-fault reactivation model of the Azle region and a modified corner point meshing algorithm, we integrate structural, seismic, and well data to model novel mechanisms of stress interaction and propagation. The model indicates how Barnett’s production-induced contraction interacts with Ellenburger’s injection-induced expansion in the footwall block of the antithetic fault to cause updip shear on the fault. In the hanging-wall block of the synthetic fault below Ellenburger, the model finds that the downdip shear from Ellenburger expansion colludes with the decrease in fault’s effective compression due to downdip fluid diffusion. The collusion induces normal faulting in the model. Modeling also reveals that the reorientation of the principal stresses is more severe in the Barnett compared with that in the Ellenburger. The model quantifies the impact of stress reorientation on the magnitude and location of fault reactivation events.