Investigating physical and statistical correlations associated with induced seismicity in the Delaware Basin, Texas

Abstract:

Wastewater disposal, hydraulic fracturing and hydrocarbon production are all types of subsurface industrial processes that have been shown to induce seismicity across the globe. However, the geological controls that help to promote induced seismicity remain poorly understood. Furthermore, it is not known why the respective roles of different triggering processes, such as pore pressure diffusion, poro-elastic stress transfer, and static elastic stress transfer, vary from place-to-place.

To answer these questions, we need to integrate statistical and numerical subsurface models from a variety of regions where induced seismicity is observed, particularly in areas where detailed injection and production rates are recorded independently. We use a logistic regression approach to statistically assess some of the possible controlling factors of induced seismicity in the Delaware Basin, Texas, where there has been a sharp increase in recorded seismicity since 2009. We use a semi-analytical approach to determine the evolution of pore pressure and poroelastic stress based on monthly injection data. Our initial results show that whilst poroelastic stresses and pore pressures sometimes reach high values (~0.1 MPa) during some earthquake sequences, these modelled pressures fail to sufficiently correlate with induced seismicity over all areas in space and time, so we then incorporate geological parameters into our regression. These potential features include fault density, whether faults are optimally oriented for reactivation, and depth of injection relative to basement. We discuss the statistical significance for each of these features as possible predictors of seismicity.