Fractures within the earth control rock strength and fluid flow, but their dynamic nature is not well understood. As part of a series of underground chemical explosions in granite in Nevada, we collected and analyzed microfracture density data sets prior to, and following, individual explosions. Our work shows an ∼4-fold increase in both open and filled microfractures following the explosions. Based on the timing of core retrieval, filling of some new fractures occurs in as little as 6 wk after fracture opening under shallow (<100 m) crustal conditions. These results suggest that near-surface fractures may fill quite rapidly, potentially changing permeability on time scales relevant to oil, gas, and geothermal energy production; carbon sequestration; seismic cycles; and radionuclide migration from nuclear waste storage and underground nuclear explosions.
The time required for rock fractures to fill and seal is poorly constrained, particularly in shallow crustal environments. Fracture apertures, which are reduced by filling and sealing, are important to oil, gas, and geothermal energy production (Hippler, 1993; Boles et al., 2004; Dempsey et al., 2013), carbon sequestration (Shukla et al., 2010), and radionuclide migration from nuclear waste storage (Ticknor et al., 1989; Moreno and Neretnieks, 1993) and underground nuclear explosion detection, monitoring, and characterization (Jordan et al., 2015). While fracture filling is often assumed to be on “geologic time scales” (e.g., Lander and Laubach, 2015), some studies have suggested that sealing on the order of 101–104 yr could better explain the timing of seismic cycles in tectonically loaded faults (e.g., Blanpied et al., 1992; Byerlee, 1993; Chester et al., 1993). Other field-scale studies have suggested even faster sealing rates of months to years, based on transient changes in seismic velocities (Tadokoro and Ando, 2002; Hiramatsu et al., 2005; Li, 2006) and permeabilities (Rojstaczer and Wolf, 1992; Claesson et al., 2007; Xue et al., 2013; Wästeby et al., 2014) following earthquakes, but the mechanism by which these changes occur (i.e., how fractures might be closing) is not observed with these techniques.
Laboratory studies have shed light onto the potential processes by which fractures can seal, including crack filling due to mineral precipitation at temperatures of 150–1000 °C (Morrow et al., 1981, 2001; Moore et al., 1994; Brantley et al., 1990; Tenthorey et al., 2003), removal of asperities through dissolution/reprecipitation (Beeler and Hickman, 2004; Gratier et al., 2013; Aben et al., 2017), or asperity contact yielding without fluids (Dieterich and Kilgore, 1994), which can lead to reduced permeability (e.g., Morrow et al., 1981; Moore et al., 1994). However, the time scales and relative importance of each of these processes are not well quantified at crustal conditions, particularly in low-temperature, near-surface conditions.
Improving our understanding of fracture filling and sealing rate requires isolating a single deformation event. Earthquake prediction is so difficult that it precludes direct comparison of pre-earthquake fracture properties with a time series of postearthquake fracture properties. Thus, the Source Physics Experiment (SPE), a series of underground chemical explosions in granite in Nevada, provided a unique opportunity to collect core before and after a (nontectonic) damage event.
As part of an effort to understand the role of damage on seismic signatures of underground chemical explosions (Snelson et al., 2013), core was collected prior to any chemical explosions (“pre-ex,” U15n) and after each of the larger two explosions (“post-ex,” U15n#10, U15n#12, and U15n#13; Table 1; locations in Fig. 1). These explosions were conducted in the quartz monzonite member of the Climax Stock granite, located on the arid Nevada National Security Site. Air foam, and not drilling mud, was used during drilling. U15n core was collected first, and then the borehole was enlarged to enable emplacement of the explosives. U15n#10 core was collected 6 wk after the SPE-2 explosion; U15n#12 and U15n#13 cores were collected 18 and 21 wk, respectively, after the SPE-3 explosion, and more than a year after SPE-2. Cores U15n#10 and U15n#12 were drilled at an angle of ∼30° from vertical (to ensure intersection with the explosive source region), while U15n and U15n#13 were drilled vertically (Fig. 1). All core was delivered to and stored at the U.S. Geological Survey (USGS) Data Center and Core Library in Mercury, Nevada.
Samples for microstructural analyses were selected from intervals approximately every 6 m along each core, and three mutually perpendicular thin sections were made from each sample. These are the same thin sections used to define the extent of damage by Swanson et al. (2018), and more information about thin section preparation and orientations can be found there and in the GSA Data Repository1.
For each thin section, transgranular microfractures were characterized as open or filled (see Data Repository Table DR1, Figure DR1, and text for classification criteria), and their respective linear densities were measured using the same technique utilized by Swanson et al. (2018) and described in detail in the Data Repository. For each sample (i.e., depth), the microfracture density values for each of the three orientations of thin sections were compared, and the largest microfracture density was chosen for comparison, to facilitate identification of potentially oriented damage. The core was not oriented, but a discussion of orientations of mesoscale fractures can be found in Swanson et al. (2018).
To determine if fracture fill was pulverized host rock or newly precipitated material, we collected images on a few select samples using an FEI Inspect scanning electron microscope (SEM), using a 15 kV beam. Preparation included sputter coating the samples with gold as the conductive layer. Elemental analyses were performed using energy dispersive spectra (EDS). Elements were identified using EDAX EDS software and quantified using a standardless ZAF quantification, which accounts for effects due to atomic number (Z), absorption (A), and fluorescence (F).
We use the term “filled” to describe fractures with visible amounts of material in them, in contrast with the term “sealed” used in previous studies (e.g., Wilson et al., 2003; Mitchell and Faulkner, 2009), to avoid the implication that these fractures are impermeable or have recovered their strength. However, fracture fill is a step toward sealing, and some permeability loss and/or strength recovery may have occurred (relative to open fractures), and with time, these fractures may seal completely.
We compared open microfracture density to depth for all samples (Fig. 2A). In the pre-ex core, U15n, open microfracture densities were all relatively low, with fewer than 1 microfracture per millimeter (mf/mm) for every sample. The pre-ex sample containing the most fractures was collected near a previously identified minor fault at 26.5 m depth (Townsend et al., 2012). The post-ex cores U15n#10, U15n#12, and U15n#13 showed higher and more variable open microfracture densities, ranging from 0.2 to 6.7 mf/mm. Among post-ex samples, we found the highest microfracture densities (>2 mf/mm) in samples within 10 m of the explosion center, and within 15 m of the surface. We expected this pattern for explosion-induced processes, because shock waves should damage the nearest samples before being attenuated. Near the surface, additional damage results from spallation, i.e., the reflection of the upward explosion-induced seismic waves into a downward-traveling tensile wave, enhancing damage (Eisler and Chilton, 1964).
Filled microfracture densities for the pre-ex core U15n were <1.1 mf/mm and highest near a fault zone (Fig. 2B). Filled microfracture densities from post-ex samples (cores U15n#10, U15n#12, and U15n#13) ranged from 0 to 4.7 mf/mm. The highest post-ex filled microfracture densities were over a factor of 4 higher than the pre-ex core. In addition, 25 post-ex samples (out of 55) showed more filled microfractures than any pre-ex sample. The majority of filled microfractures contained clay. Sixteen percent of pre-ex filled fractures contained calcite, but less than 5% of post-ex filled fractures did (Table DR1) because of the increase in number of clay-filled fractures.
Most fracture-fill material consisted of montmorillonite, and not broken pieces of host rock (Fig. 3). The texture of the clay, with no laminations or alignment of clay grains that would indicate flow or disruption (Figs. 3A, 3B, and 3D), suggests in situ precipitation of the montmorillonite, rather than the displacement of pulverized material into fractures during the shock wave propagation.
The spatial distribution of regions with large numbers of open microfractures coincided with that of filled fractures, suggesting a similar formation process (Figs. 2A and 2B). The regions of enhanced damage were consistent with shock fracturing near the source, and spallation near the surface, matching our expectation for explosion-induced fracturing (Swanson et al., 2018). We would not expect this pattern if all filled microfractures reflect preexisting faults. While there is a small fault zone that may explain the increase in filled fractures within U15n#10 and U15n#12 near the source (Fig. 1C), there is no core-scale evidence of faulting in other areas with large densities of filled microfractures (Townsend et al., 2012).
Although the data are limited, host-rock saturation appears to have influenced the time at which an increase in filled microfractures was measurable (Fig. 2B). The top of the perched water table ranged between 22.4 and 24.1 m depth during the time of study, with brief excursions following the explosions and/or drilling (Townsend et al., 2012). Below the water table, all post-ex cores (U15n#10, U15n#12, and U15n#13) contained samples that showed higher filled microfracture densities than the pre-ex samples. Above the water table, samples from U15n#12 and U15n#13, but notably not U15n#10, showed more filled microfractures than pre-ex samples (Fig. 2B). We examined samples of rock core from U15n#10, which was collected 6 wk after SPE-2, and samples of rock core from U15n#12 and U15n#13, which were collected 18 and 21 wk after SPE-3, and more than a year after SPE-2 (Table 1). This examination suggested that explosion-induced microfractures took less than 6 wk to begin to fill (but not necessarily completely seal; see Fig. DR2) under saturated conditions, but they took more time to begin to fill above the water table. The exact timing of detectable filling above the water table is more poorly constrained because it was not possible to determine if the filled microfractures in U15n#12 and U15n#13 opened from SPE-2 or SPE-3.
Modeling of the groundwater at the Climax stock suggests that it is out of equilibrium with the host granite body, and likely to precipitate montmorillonite and/or kaolinite, although the rate of precipitation was not quantified (Isherwood et al., 1982). A more complex model for a generic granite body predicted the precipitation of montmorillonite would become substantial within 5 wk of weathering (Fritz et al., 2009), although this has never been demonstrated in situ.
Based on these results, we suggest that explosion-induced fracturing creates fresh, unaltered mineral faces that readily react with existing groundwater and precipitate alteration minerals. Above the groundwater table, the lower abundance of water is likely responsible for a slower fracture filling rate. At this study area, in a granite less than 120 m deep, these reactions facilitated precipitation of clay minerals, especially montmorillonite. Given that these reactions can occur even in shallow, relatively cold conditions in an arid area, it seems likely that this fracture-filling process could occur in any setting where fresh, open fractures come in contact with chemically reactive fluids, even at very shallow depths (< 1 km), although the mineralogy would likely differ in different settings.
Our results show that fractures can fill within 6 wk. Given the different mechanical properties that filled and open fractures might have, time has the potential to affect rock strength and would affect damage patterns from any subsequent explosions. In addition, these rapidly filling fractures indicate the need to treat permeability as a physical property that varies with time. In terms of explosion monitoring, rapidly filled fractures would decrease permeability, delaying gas migration to the surface, and could impact nuclear explosion characterization. Beyond explosion applications, the inferred strength and permeability changes, as discussed in this investigation, influence how we apply rock properties and fluid flow to other topics such as earthquake cycling, mine stability, energy production, and waste storage. This study highlights the importance of considering interactions with water content and chemistry in evaluating the effect of fractures on rock properties.
The Source Physics Experiment (SPE) would not have been possible without the support of many people from several organizations: National Nuclear Security Administration, Defense Nuclear Nonproliferation Research and Development (DNN R&D), and the SPE working group, a multi-institutional and interdisciplinary group of scientists and engineers. This work was done by Los Alamos National Laboratory, under award number DE-AC52–06NA25396, LA-UR-17–21442 (Version 4). Sandia National Laboratories is a multimission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525, SAND2019–10161J. The views expressed here do not necessarily reflect the views of the U.S. government, the U.S. Department of Energy, Sandia National Laboratories, or Los Alamos National Laboratory.