Evaluating the effectiveness of induced seismicity mitigation: Numerical modeling of wastewater injection near Greeley, Colorado

Abstract

Mitigation of injection‐induced seismicity in Greeley, Colorado, is based largely on proximity of wastewater disposal wells to seismicity and consists of cementation of the bottom of wells to eliminate connection between the disposal interval and crystalline basement. Brief injection rate reductions followed felt events, but injection rates returned to high levels, >250,000 barrels/month, within 6 months. While brief rate reduction reduces seismicity in the short term, overall seismicity is not reduced. We examine contributions to pore pressure change by injection from 22 wells within 30 km of the center of seismicity. The combined injection rate of seven disposal wells within 15 km of the seismicity (Greeley Wells) is correlated with the seismicity rate. We find that injection from NGL‐C4A, the well previously suspected as the likely cause of the induced seismicity, is responsible for ~28% of pore pressure increase. The other six Greeley Wells contribute ~28% of pore pressure increase, and the 15 Far‐field Wells between 15 and 30 km from the seismicity contribute ~44% of pore pressure increase. Modeling results show that NGL‐C4A plays the largest role in increased pore pressure but shows that the six other Greeley Wells have approximately the same influence as NGL‐C4A. Furthermore, the 15 Far‐field Wells have significant influence on pore pressure near the seismicity. Since the main mitigation action of cementing the bottom of wells has not decreased seismicity, mitigation based on reduced injection rates and spacing wells farther apart would likely have a higher potential for success.

Small Earthquakes Matter in Injection‐Induced Seismicity

“…Maximum ΔP varies from 0.04 to 42.6 MPa. This is consistent with other site-specific pore pressure models of induced seismicity (e.g., Brown et al, 2017;Healy et al, 1968;Hornbach et al, 2015;Keranen et al, 2014;Nakai et al, 2017;Ogwari et al, 2018;Ogwari & Horton, 2016). Low permeability generates the largest pore pressure increase, but pore pressure does not diffuse very far into the basement or laterally (Figure 2b); high permeability generates much further diffusion of pore pressure increase, but the magnitude of the increase is much smaller (Figure 2d) than low and medium permeability simulations (Figures 2b and 2c).…”

“…The ΔCSS is shown in Figure 3 for average depth of the earthquakes, ranging from 3.8 to 4.3 km (Table S3), and in the direction of slip of the source fault. There are areas of positive ΔCSS in all scenarios that are as large as or larger than estimated triggering thresholds of pore pressure (~0.07 to 0.10 MPa) from site-specific studies (e.g., Brown et al, 2017;Keranen et al, 2014). Maximum ΔCSS at average depth varies from approximately 12 to 770 MPa in strike-slip fault scenarios (Figures 3a-3c) and approximately 18 to 62 MPa in normal fault scenarios (Figures 3d-3f).…”

Surface Deformation and Induced Seismicity Due to Fluid Injection and Oil and Gas Extraction in Western Texas

“…MODFLOW is the USGS's modular finite‐difference hydrologic model. It is commonly used for pore pressure simulation due to its flexibility and speed (e.g., Zhang et al, 2013, 2016; Keranen et al, 2014; Hornbach et al, 2015; Ogwari & Horton, 2016; Ogwari et al, 2018; Brown et al, 2017; Brown & Ge, 2018; Goebel et al, 2017; Nakai et al, 2017; Hearn et al, 2018). It solves the three‐dimensional movement of ground water of constant density in heterogenous and anisotropic porous media (from Harbaugh, 2005): where K xx , K yy , and K zz are the principal components of hydraulic conductivity, S s is the specific storage, W is the volumetric flux per unit volume representing sources of fluid, h is hydraulic head, and t is time.…”

High density oilfield wastewater disposal causes deeper, stronger, and more persistent earthquakes

“…One landmark modeling study found that several high-rate SWD wells in southeast Oklahoma City produced a fluid pressure front that accurately matched earthquake hypocenter locations leading up to the Jones earthquake swarm 8 . This history-matching approach was repeated in more recent studies linking SWD operations to earthquake occurrence, e.g., in Milan, Kansas 9 , Greeley, Colorado 10 , Fairview, Oklahoma 11 , Dallas-Fort Worth, Texas 12 , and Guthrie, Oklahoma 13 . Based on the success of this history-matching approach, groundwater models are now being incorporated into seismic hazard assessments to simulate fluid pressure decay following SWD volume reductions 14–17 .…”