INTRODUCTION

As stated in Hennings and Smye (2024, this issue, p. 2202) “When you produce fluids from the subsurface you learn things, when you inject fluids into the subsurface you learn much more”. This is an important aphorism that petroleum geoscientists and engineers learn as they manage subsurface reservoirs. In this regard, over the last approximately 15 yr the petroleum industry has been conducting what amounts to a grand geomechanical experiment in the United States midcontinent and Permian Basin region by injecting approximately 50 billion bbl (∼8 billion m³) of produced wastewater—mainly the byproduct of unconventional oil production from tight reservoirs. Injection of this produced water into formations between the tight oil-productive zones and geologic basement (deep injection) has caused existing faults to slip producing more than 4,000 magnitude (M) ≥ 3.0 earthquakes with 10 M ≥ 5.0 as of October 2024 (https://www.usgs.gov/programs/earthquake-hazards). In the Permian Basin, the petroleum industry now handles an estimated 15 million bbl/day (2.4 million m³/day) of produced water from unconventional production (Smye et al., 2024b, this issue). In addition to high rates of problematic induced seismicity stemming from deep injection, injection above oil-productive zones (shallow injection) has pressurized those reservoirs causing additional cases of induced seismicity, uplifted the ground surface, elevated the drilling hazard, and created saline surface flows at old wellbores (Hennings et al., 2023; Smye et al., 2024b, this issue). It appears that recycling of produced water for hydraulic fracturing (HF) has plateaued at only 30% of that daily total and broad implementation of produced water treatment for beneficial reuse lies many years in the future. The impacts of produced water injection (SWD) have become acute, especially given concerns that the subsurface capacity for injection is finite and has been exceeded at current injection rates in many areas (Smye et al., 2024b, this issue; Hennings and Smye, 2024, this issue). Total SWD in the entire Permian Basin from 1983 through to present totals >60 billion bbl (>9.5 billion m³), yet one estimate of future water production exceeds an additional 310 billion bbl (∼50 billion m³) as the Permian Basin region is fully produced in the decades ahead (Scanlon et al., 2020). Therefore, we are much closer to the beginning than the end of the produced water challenge in the region.

Understanding the subsurface impacts of large-scale injection, and the evolving hazard first requires an understanding of the native in situ condition – the equilibrium state prior to the perturbance introduced by injection. At this stage, one seeks to know the flow characteristics and extent of the injection reservoir(s); the pore pressure condition of the reservoir(s); the full tensor describing the stress state; the existence, characteristics, three-dimensional (3-D) extent, and reactivation potential of faults; and knowledge of prior seismicity. From the standpoint of technical capabilities, these are subsurface analysis tasks that the industry can perform with ease and confidence and are typically performed iteratively and methodically. Once large-scale injection begins, and the hazard becomes apparent from induced earthquakes or excessive reservoir pressurization, the technical endeavor quickly transitions to needing to understand and predict the disequilibrium state, and gain quantitative insight into the dynamic interrelationship of the magnitude and distribution of pore pressure change (ΔPp), how ΔPp alters in situ stress, how faults experience the stress change and contribute to the distribution of ΔPp, the location and characteristics of earthquakes that are anomalous as compared to historic norms, and to identify challenges to ongoing subsurface operations and the near surface environment. Given a new and local risk from earthquake shaking, business interruption and investment loss stemming from regulatory action, and public concern impacting social license to operate, operators find themselves with short timelines within which to implement effective mitigative actions. Publicly accessible research into the causes, characteristics, and trajectory of induced seismicity plays a vital role in assisting hazard characterization and the implementation of effective mitigation strategies.

Working on the topic of induced seismicity in the space shared between the upstream and midstream petroleum industry, academic research, regulatory entities, and the public is essential, yet cumbersome. Within this shared space, the Center for Injection and Seismicity Research (CISR) at the Bureau of Economic Geology at The University of Texas at Austin has been researching the impacts of SWD in Texas since 2016. During this same period, the State of Texas-funded Texas Seismological Network (TexNet) has been developed to provide essential data on earthquakes in Texas. In this issue of the AAPG Bulletin titled “The Geology of Injection-Induced Earthquakes in the Midland Basin Region,” CISR contributes six papers on the topic of wastewater production and injection and its impact in the Permian Basin region with an additional focus on the Midland Basin. Collectively, these papers provide an understanding of geological, reservoir engineering, and geomechanical aspects of the issue, locally and regionally. The papers in this special issue technically build on each other to leverage important aspects of integration.

SYNOPSIS OF PAPERS IN THE SPECIAL ISSUE

Two papers in the special issue focus on issues of SWD and induced seismicity across the Permian Basin region. Hennings and Smye (2024, this issue) provide a summary of the current understanding of causes and the mechanistic nature of induced seismicity in the Permian Basin. They review the long-term trends of natural and induced seismicity in the region and describe the characteristics of recent earthquakes as cataloged by a variety of seismic networks. They provide a synthesis of earthquake causal mechanisms and then partition those mechanisms and the associated earthquake clusters into seven distinct induced earthquake systems. They discuss the controls on maximum cataloged earthquake magnitude as a function of the hosting earthquake system. Before posing a series of questions and providing suggestions than can assist in framing future technical analysis and research, Hennings and Smye review the actions taken by petroleum regulators and injection well operators which has led to significant reductions in the rate of problematic earthquakes in key areas of the Permian Basin. They note that the rate of problematic earthquakes remains stubbornly high in some areas of the basin.

Understanding and addressing the impacts of large-scale SWD begins with knowing the source of produced water and trends in SWD. Smye et al. (2024b, this issue) address the systematics of wastewater production and SWD injection and update our understanding of how “wet” production is in the Permian Basin as compared to other North American unconventional plays. Smye et al. provide an analysis of the distribution of coproduced water from both legacy conventional production and unconventional production and highlights the heightened challenge experienced by unconventional operators in the Delaware Basin where the ratio of water to oil volumes produced can exceed 4.0. Smye et al. describe the geologic zones used for SWD and the volumes injected into each. In simplest terms, wastewater is injected into four regional rock volumes (injection systems): deep reservoirs in the Delaware and Midland basins and shallow reservoirs in those basins. Carbonate-dominated reservoirs used for deep SWD have properties highly favorable for injection with rates that can have permit allowances of 75,000 bbl/day (∼12,000 m³/day). Shallow SWD in the Midland Basin also utilizes injection zones dominated by carbonate reservoirs with properties similar to the zones used for deep SWD. Shallow SWD in the Delaware Basin utilizes injection zones dominated by clastic reservoirs that have properties that are somewhat less favorable for injection as compared to the carbonates. Smye et al. discuss the implications of published models of SWD reservoir ΔPp and show that increases of up to 500 psi (∼3.5 MPa) from injection are now common in many areas with large-scale SWD operations. This represents an increase in native pore pressure of up to 10% for the deep carbonates and up to 20% for the shallow clastic assemblages. Smye et al. continue with an analysis of how ΔPp has pushed these injection systems into disequilibrium states, causing anomalous rates of seismicity and concludes with the observation that the relationship between ΔPp and earthquake rate is complex when systems are compared against each other. Deep injection in the northern Delaware Basin is able to generate the highest earthquake rate and is therefore the most sensitive injection system in the Permian Basin region. Adverse impacts of shallow injection include drilling hazards related to overpressuring and compromised integrity of older vertical wellbores resulting in surface wastewater discharges.

Four papers in the special issue focus specifically on the Midland Basin. Mechanistic analyses of induced seismicity hazard as well as causal assessments begin with a detailed understanding of the geologic characteristics of the reservoirs used for injection. Calle et al. (2024, this issue) provide a detailed characterization of the formations in the Midland Basin used for deep injection. The stratigraphic correlation and mapping; petrophysical analysis of facies, porosity and matrix permeability; and 3-D geomodeling integration results in maps of general structure and lithology-dependent thickness for the Ellenburger Group, top of Thirtyone Formation to base of Montoya Group, and units of Pennsylvanian age. For consideration of deep injection capacity, Calle provides maps of aggregate matrix porosity and pore volume for those three gross rock sections. Collectively, these intervals are the deep SWD storage and flow units. Calle also provides interpretations of the thickness and characteristics of the low permeability, shale-dominated units including the Simpson Group, Woodford, and Barnett Formations that lie in between the flow units. This provides important constraints on the limits of vertical connectivity within the overall injection system. By inspection of the pore volume maps in relation to the location and volumes of wastewater injected thus far, there are enormous volumes of injection reservoir capacity that lie outside of the current region of large-scale deep SWD and outside of the corridor of induced earthquakes. Given the future demand for SWD injection capacity, a question that must be posed is the degree to which this enormous deep injection resource in the Midland Basin can be utilized for SWD in ways that do not elevate the induced seismicity hazard.

Ge et al. (2024, this issue) directly employ the 3-D reservoir property model from Calle et al. along with a thorough reservoir engineering analysis of SWD injection rate, volume, and pressure to develop a pore pressure evolution model for deep SWD in the Midland Basin spanning 1983-2023. The approach taken to calibrate the model, using mainly wellhead pressure data, follows the approaches described in Gao et al. (2021) and Smye et al. (2024a) who also modeled ΔPp from SWD into carbonates above geologic basement. The layers of the Ge et al. pore pressure model representing the Ellenburger Group are the most salient to consider because, as discussed in Horne et al. (2024, this issue) and Hennings et al. (2024, this issue), these layers lie just above the seismogenic basement and there are many basement-rooted faults in the basin that cut from basement into the Ellenburger Group providing ready conduits for ΔPp. Ge et al. demonstrate that local areas of notable ΔPp increase of >300 psi (∼2 MPa) from SWD were developed in some areas of the Midland Basin by 2016 but, by the end of 2023, the region impacted by a ΔPp had expanded significantly. At the end of 2023, areas of ΔPp >300 psi (∼2 MPa) extended throughout Andrews, Martin, Howard, Scurry, Midland, Glasscock Counties and elsewhere, and some areas of Martin and Howard Counties have ΔPp >700 psi (4.8 MPa). These areas have all been recently impacted by injection-induced seismicity.

Induced earthquakes occur on preexisting faults that are destabilized by an increase in pore pressure and a commensurate local change in stress. This can cause fault slip and the release of seismic energy. Therefore, having the most complete and accurate interpretation of the faults in 3-D is essential for site characterization, hazard assessment, and mechanistic analysis (see Hennings et al., 2024, this issue for a discussion of mechanistic analysis approaches). Horne et al. (2024, this issue) provide a new interpretation of faults and slip hazard in the Midland Basin. This new fault interpretation originates from an integration of two-dimensional and 3-D reflection seismic data, formation tops from vertical wells, landing zone structure from more than 17,000 horizontal wells, fault indicators from linear earthquake clusters, and extensive information from CISR sponsors and other industry sources. This final interpretation is expressed in terms of mapping confidence following the methods described in Horne et al. (2020). Horne interprets 795 faults in the Midland Basin that sum to more than 3,200 mi (>5,200 km) of fault trace length. These faults are generally rooted in the geologic basement and extend upward through lower Paleozoic strata to levels as shallow as the lowermost levels of the Wolfcamp Group. The faults are generally steeply dipping and have a dominant fabric that strikes to the northeast throughout the basin. Faults that strike to the northwest are also common but occur in aerially confined domains. Approximately 370 mi (∼600 km) of the cumulative fault trace length have been seismogenic in the 2017-2023 period. Shallow, strata-bound faults such as extensively developed in the Delaware Basin (e.g., Horne et al., 2022) do not appear to exist in the Midland Basin. Using this fault interpretation, Horne provides a fault slip hazard assessment, finding that faults that strike to the northeast and to the northwest are highly sensitive to slip with only a modest increase in injection reservoir ΔPp. The extent of faulting as reported by Horne et al. has important implications to the mechanistics of induced seismicity in the basin including (1) the faults provide direct hydrogeologic connection from the deep injection strata to the seismogenic basement throughout the basin; (2) the orientation characteristics of the faults within the in situ stress field clearly describes the fault slip hazard; (3) the distribution of faults that have become seismogenic closely follows the published trend of Grenville tectonic front (e.g., Mosher, 1998), which is important for future induced seismicity site characterization and hazard assessment; and (4) the upper tips of the faults do not extend into reservoirs typically used for unconventional development, reducing the induced seismicity hazard associated with HF in formations of Permian age but deeper formations will therefore carry an elevated risk of HF-induced seismicity when operations are conducted in proximity to sensitive faults (e.g., McKeighan et al., 2022). The fault interpretation of Horne combined with Horne et al. (2021 and 2022), Hennings et al., (2021), and Morris et al. (2021) provides new fault interpretations and assessments of fault slip hazard throughout the Permian Basin region.

Hennings et al. (2024, this issue) incorporates the data, interpretations, and model results from the other studies in this special issue to provide analyses of the ΔPp associated with the onset of 21 earthquake clusters in the Midland Basin. They conclude that local cluster onset is associated with as small a ΔPp of 37 psi (0.26 MPa) but ranges up to 529 psi (3.65 MPa) and the mean ΔPp associated with earthquake cluster onset is 216 psi (1.5 MPa). Onset is associated with a variety of injection rate histories, especially strong month-to-month variability. Hennings et al. found that there is a close spatiotemporal relationship between the rates of injection, ΔPp, change in critically stressed fault segment length, and seismicity. There is a loose coupling of increased ΔPp and earthquake cluster development, but there is a tight spatiotemporal coupling of decreased ΔPp and earthquake rate decrease, which is beneficial to the effectiveness of earthquake rate mitigation via reduced injection rate. There is a distinct contrast in the rupture sensitivity of the fault systems in the basin as subject to ΔPp, with areas in the south and southwest near the Midland-Odessa population center being considerably more sensitive as compared to areas to the north and northeast including Martin, Howard, Scurry and Fisher Counties. This indicates that the dynamic injection capacity for deep SWD reservoirs varies significantly across the earthquake-prone corridor of the Midland Basin.