COUPLED DEFORMATION AND FLOW MODELING IN THE STUDY OF INDUCED SEISMICITY: AN EXAMPLE FROM THE PARADOX BASIN, COLORADO

Subsurface fluid injection produces poroelastic stresses that may explain why induced earthquakes occur where pore fluid pressure is not elevated. We have used the finite-element software COMSOL to simulate excess pore fluid pressure and coupled poroelastic stress changes due to brine injection 4.3-4.8 km deep in the Paradox basin, Colorado.

Steady injection started in July 1996 and began inducing earthquakes four months later. Through 2012, 6.43 million m³ of brine had been injected at wellhead pressures up to 34.5 MPa, inducing thousands of microearthquakes recorded by the Paradox Valley Seismic Network, installed by the U.S. Bureau of Reclamation in 1985. Most microseismicity is near or within the limestone injection horizon, but many events also occur within two km above and below that.

The daily wellhead pressure and flow rate, which are the only data available to calibrate the model, constrain permeability of the injection horizon and reservoir geometry. A radial model with flat uniform layers, though oversimplified, can match the pressure and flow-rate data. After 2001, pressure rose approximately linearly with cumulative injected volume; this feature can be simulated by an impermeable boundary 14 to 18 km from the well.

Simulations show that, although forces induced by excess pore pressure dominate stress in the limestone, horizontal shear stress can become concentrated in the adjacent low-permeability confining layers and uppermost granitic basement where little injected brine penetrates. Both the steep pressure gradient near the well and the more distant impermeable boundary intensify these shear stresses. For example, shear stresses of 0.04 MPa occur in the upper two km of basement below the boundary. Stress concentrations at permeability contrasts near the injection horizon are likely present in other areas of injection-induced seismicity, and may account for events outside zones of high fluid pressure.

Although poroelastic moduli are not well constrained, the coupled model can calculate surface deformation. For example, simulated radial strains are of order 10^-6 on the surface at 4 km from the well, which would be measurable with borehole strainmeters. Deployment of suitable geodetic instrumentation around injection wells could yield observable deformation to improve model calibration.