Modeling Pore Pressure Increase in a Class II Underground Injection Well

Numerical Modeling of Fluid Injection in Northstar #1 UIC Class II Well, Youngstown, Ohio

From January 2011 to February 2012 over a hundred small earthquakes occurred in a formerly seismically quiescent area near Youngstown, Ohio, within 1000 m of a deep salt water injection well. This case of induced seismicity has been studied by ODNR (2012), Kim (2013) and Raziperchikolaee and Miller (2015). The latter authors used numerical modeling to test the hypothesis that highly conductive vertical faults connect the target horizons to the deeper hypocentral zone in the Precambrian basement. This brief describes a similar modeling effort using the MATLAB Reservoir Simulation Toolbox (MRST), an open source software application developed by SINTEF Digital (Lie, 2016).

The purpose of this simulation is to model the flow of fluid into the target reservoir and surrounding porous media to correlate the increase in pore pressure with the spatial and temporal distribution of recorded seismicity. The information on well construction, borehole lithology, formation permeability and porosity, daily surface injection pressure and volume are taken from ODNR (2012). The timing and distribution of earthquakes in this sequence has been well-studied by Kim (2013). The unknowns in the simulation are the structure and hydraulic characteristics of the fault/fracture network connecting the target reservoir to the hypocentral zone in the Precambrian crystalline basement. In modeling the hydraulic characteristics of the faults, we rely on the empirical results of the simulation by Raziperchikolaee and Miller (2015). By repeating the numerical simulation of Raziperchikolaee and Miller (2015) we are bench marking the performance of our MRST application.

Earthquakes Caused by Injection into Northstar #1

Northstar #1 became operational in late December 2010. On 17 March 2011, residents in Youngstown felt a M_w 2.3 earthquake. By 13 January 2012, 12 earthquakes (M_w ~1.8–3.9) had been recorded by regional seismographs and six of these had been felt by local residents. At the request of ODNR, personnel from Lamont-Doherty Earth Observatory (LDEO) installed four portable seismographs on 1 December 2011 to monitor the seismicity and provide more precise hypocentral locations. On 24 December 2011 a M_w 2.7 earthquake occurred near the injection well. In response to the increase in seismic activity ODNR requested the operator to shut down the Northstar #1 well on 30 December 2011. Within 24 hours of the order the largest event in the sequence occurred on 31 December 2011 (M_w 3.9). Seismicity continued into February 2012 with a series of very small magnitude events and the LDEO stations were removed (Kim, 2013). After further analysis, ODNR concluded that, based on the spatiotemporal distribution of the events relative to the disposal operation, the earthquakes were induced by fluid injection into Northstar #1 (ODNR 2012).

The MRST Simulation

This MRST simulation requires a grid of cells covering the model space, a rock model that assigns permeability and porosity to each cell, a well model that describes the location of the injection well and its properties, an injection schedule that specifies the volume of fluid injected over time, a fluid model characterizing the physical properties of the injectate (brine in this case), a description of the initial reservoir state and a flow solver to run a simulation through all the time steps given a non-linear physical model, schedule and initial state. The following figures describe how the model was set up and the results obtained, which indicate that the deep east-west striking fault hypothesized by Kim (2013) can experience an increase in pore pressure sufficient to cause slip by hydraulically connecting the borehole to the hypocentral region via highly conductive vertical fractures in the manner suggested by Raziperchikolaee and Miller (2015).