Journal of Structural Geology 173 (2023) 104915

Contents lists available at ScienceDirect

Journal of Structural Geology

journal homepage: www.elsevier.com/locate/jsg

Fracture-pattern growth in the deep, chemically reactive subsurface

J.N. Hooker a, *, R.F. Katz b, S.E. Laubach c, J. Cartwright b, P. Eichhubl c, E. Ukar c,
D. Bloomfield d, T. Engelder d
a
School of Mathematics, Science, and Engineering, University of the Incarnate Word, 4301 Broadway, San Antonio, TX, 78209, USA
b
Department of Earth Sciences, University of Oxford, South Parks Road, Oxford, OX1 3AN, UK
c
Bureau of Economic Geology, The University of Texas at Austin, 10100 Burnet Road Bldg. 130, Austin, TX, 78757, USA
d
Department of Geosciences, Penn State University, 503 Deike Bldg., University Park, PA, 16802, USA

A B S T R A C T

Arrays of natural opening-mode fractures show systematic patterns in size and spatial arrangement. The controls on these factors are enigmatic, but in many cases the
depth of formation appears to be critical. Physical, potentially depth-dependent factors that could account for these variations include confining stress, fluid pressure,
and strain rate; these factors are common inputs to existing fracture models. However, temperature-dependent chemical processes likely exert an equally important
control on patterns, and such processes have not yet been rigorously incorporated into models of fracture formation. Here we present a spring-lattice model that
simulates fracturing in extending sedimentary rock beds, while explicitly accounting for cementation during opening of fractures, and for rock failure via both elastic
and time-dependent failure criteria. Results illustrate three distinct fracturing behaviors having documented natural analogs, which we here term fracture facies.
“Exclusionary macrofracturing” occurs at shallow levels and produces large, widely spaced, uncemented fractures; “multi-scale fracturing” occurs at moderate depth
and produces partially cemented fractures having a wide range of sizes and spacings; and “penetrative microfracturing” occurs at great depth and produces myriad
narrow, sealed fractures that are closely and regularly spaced. The effect of depth is primarily to accelerate both dissolution and precipitation reactions via increased
temperature and porewater salinity; the specific depth range of each fracture facies will vary by host-rock lithology, grain size, strain rate, and thermal history.

1. Introduction horizontal drilling (e.g., Li et al., 2018). A major advancement in this

regard was to incorporate subcritical crack propagation in numerical
Natural fracture growth and pattern-evolution in the Earth is affected models (Olson, 1993). In subcritical fracture growth, corrosive reactions
by the physical and chemical environment (e.g., Pollard and Aydin, in the high-tension region near fracture tips cause fractures to lengthen
1988; Eichhubl, 2004; Laubach et al., 2019). Previous conceptual and at stress intensities lower than those otherwise required to cause fracture
numerical modeling work has mostly focused on the physical environ­ growth (Atkinson, 1984). By incorporating this aspect of the chemical
ment, treating rocks as elastic bodies that fail in response to applied environment, more-realistic geological fracture patterns could be
loads—see especially the pioneering work on fracture spacing in bedded simulated numerically, supporting the importance of subcritical propa­
rocks (Price, 1966; Hobbs, 1967). This approach successfully explained gation and deepening our understanding of fracture interaction and
the often-observed positive correlation between stratabound fracture clustering (Olson, 1993, 2004; Savalli and Engelder, 2005).
spacing and layer thickness (e.g., Ladeira and Price, 1981) by showing Advances in microstructural imaging of cemented and partially
that a tension applied remotely becomes zero at traction-free fracture cemented opening-mode fractures containing characteristic crack-seal
walls, suppressing the development of fractures that are within some cement textures (Ramsay, 1980; Bons et al., 2012) indicate that frac­
distance proportional to the height of the fracture-hosting layer (Ji and ture growth and porosity partitioning are controlled in part by synki­
Suwatari, 1998; Bai and Pollard, 2001; Schöpfer et al., 2011), in beds nematic (during-opening) fracture cementation (e.g., Laubach et al.,
having meter-scale or smaller thickness (Ladeira and Price, 1981; 2004a). Based on size-dependent porosity preservation, whereby sparse,
Chemenda et al., 2021). large, partially open fractures are present amid abundant, fully cemen­
Such models of layer-bound fractures in elastic media do not account ted microfractures, Hooker et al. (2012) suggested that the primary
for the wide fractures found in some deep wells (e.g., Olson et al., 2009), mechanical effect of cement on fracture pattern evolution is to restore
or the irregularly spaced to clustered fractures found in some outcrops adhesion across fractures. This happens where cement deposition is fast
(e.g., de Joussineau and Petit, 2021) and revealed in the subsurface by enough to span the distance across fracture walls during fracture

* Corresponding author.
E-mail address: [email protected] (J.N. Hooker).

https://doi.org/10.1016/j.jsg.2023.104915
Received 30 December 2022; Received in revised form 6 July 2023; Accepted 8 July 2023
Available online 11 July 2023
0191-8141/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

opening, a quality called spanning potential (Lander and Laubach, 2015). 2. Natural examples of pattern attributes varying with fracture
Where the variably sized fractures in a growing population have cement
consequently variable spanning potential, more completely cemented
(and small) fractures will be less likely to grow larger, compared to 2.1. Contrasting size distributions, structural styles, and spanning
lightly cemented (and large) fractures. Thus a positive feedback loop potentials: an example from the Scottish Highlands
emerges between fracture size and propensity to grow larger, leading to
a power-law fracture size distribution (Hooker et al., 2012). Natural opening-mode fracture arrays in sedimentary rocks world­
With a solid cement bond spanning the intermittently disconnected wide have size distributions that vary by cement content: arrays con­
fracture walls, tension can then be transmitted across the fracture, in taining abundant synkinematic cement are present in power-law
violation of the original fracture-spacing models cited above (e.g., Price, aperture-size distributions, i.e.:
1966) that assume fractures, once opened, are acted upon only by fluid
w = aN − b
(1)
pressures. In the past decade, advances in numerical modeling have
enabled exploration of several consequences of fracture cementation.
where w is fracture kinematic aperture, meaning total distance between
Virgo et al. (2014) showed how the style of fracture reactivation,
fracture walls (Marrett et al., 1999), N is cumulative number (1 for the
including deflection and crosscutting, depends on the relative strength
largest aperture, 2 for the second-largest, and so on) and a and b are
of the cement and the host rock, as well as any rotation of the stress field.
constants. In contrast, barren or lightly cemented fractures commonly
Vass et al. (2014) came to a similar conclusion about the importance of
have narrow aperture-size ranges, often visible in outcrop. These trends
cement strength, and showed that cementation dynamically affects the
have been documented for sandstones (Hooker et al., 2009, 2014) and
porosity and permeability evolution of fractured layers. Hooker and Katz
limestones (Ortega et al., 2010; Hooker et al., 2012).
(2015) simulated the adhesive effects of cementation within fractures in
To some extent, contrasting patterns in sandstones of what could be
a spring-lattice numerical model; they showed how cementation rates
called “veins” and “joints”—referring to heavily and lightly cemented
that are fast, relative to layer-extension rates, produce smaller and more
fractures, respectively—can be attributed to physical, rather than ther­
closely spaced fractures.
mochemical, aspects of formation-depth. For example, Gillespie et al.
Moreover, cement accumulation within intergranular porosity in the
(2001) interpreted that joints in the Burren limestones of Ireland had
host rock contributes to mechanical stiffening (Laubach et al., 2009),
restricted fracture lengthening due to their shallow formation: low
and cementation of bedding planes can suppress sliding, a mechanism
overburden stress on bedding planes allowed for slip along them, which
proposed to halt fracture propagation and thus achieve layerbound
halted joint propagation. Late joints also abut against earlier, but still
fracturing (Price, 1966; Schöpfer et al., 2011). Thus the effects of the
uncemented, joints. The presence of these barriers to propagation pro­
chemical environment on fracture pattern development are diverse, and
duced a narrow range in fracture length, best fit by a lognormal distri­
can potentially explain enigmatic fracture-pattern characteristics such
bution, and a correspondingly narrow aperture-size distribution
as power-law size distributions (Clark et al., 1995; Hooker et al., 2014;
(Gillespie et al., 2001). In contrast, veins formed at depth, where frac­
Späth et al., 2022) and spatial clustering that is statistically distin­
tures were interpreted to have grown subcritically and barriers to
guishable from both periodic and random arrangements (Gillespie,
propagation were not active; the result was a power-law length distri­
2003; Hooker et al., 2018, 2023; Marrett et al., 2018; Bistacchi et al.,
bution and a wide aperture-size distribution (Gillespie et al., 2001). Here
2020; Corrêa et al., 2022). However, to date the two distinct chemical
we describe an example from the Scottish Highlands, near Dundonnell
effects mentioned above—chemically-assisted crack growth and synki­
(Fig. 1), which shows qualitatively similar populations of cemented and
nematic sealing—have not been incorporated into a model that can
uncemented fractures hosted in sandstone. Crack-seal microstructures
examine their effects independently and in combination.
and crosscutting relationships, revealed in the field and using scanning
This study aims to establish that the depth of fracture formation af­
electron microscope-based cathodoluminescence (SEM-CL), show that it
fects the resulting patterns, not only through the well-established
is cementation that permits fracture crosscutting. Size distributions and
physical mechanisms of overburden and confining stress, but also
structural styles are indeed strongly depth-dependent, but in this case,
through thermally controlled chemical mechanisms of cementation and
they owe as much to the chemical environment as to the physical.
chemically-assisted fracture growth. A fundamental assumption is that
Field exposures of the Cambrian Eriboll Group sandstones facilitate
both precipitation and dissolution reactions are promoted through
enhanced kinetic rates at higher temperatures (e.g., Meredith and
Atkinson, 1985; Lander et al., 2008; Ankit et al., 2015). Here we support
this view using new data from natural fractures and numerical
modeling. In the following section, we present new fracture data from
three separate geologic settings that illustrate the importance of ther­
mally accelerated synkinematic cementation and chemically assisted,
subcritical fracture growth on fracture pattern evolution. Next, to
illustrate a simplified example of fracture pattern growth wherein these
processes dominate, we extend a spring-lattice numerical model of
fracture opening in response to layer-parallel extension (Hooker and
Katz, 2015) to combine synkinematic cementation with a
time-dependent, subcritical failure criterion. Finally, we identify paral­
lels between model output and natural patterns, schematizing several
categories of model results as fracture facies, which in part provides a
mechanistic basis to the oft-observed link between fracture patterns and
sedimentary facies (e.g., Bruna et al., 2015). Fig. 1. Geologic setting of Dundonnell fracture arrays, NW Scottish Highlands.
The town of Dundonnell lies approximately 4 km to the north. Sample location
indicated by circle (see Fig. 2). M: Moine Supergroup (Neoproterozoic); Ax:
Applecross Formation (Neoproterozoic); Eb: Eriboll Formation, Basal Quartzite
member (Cambrian); Ep: Eriboll Formation, Piperock member (Cambrian); An:
An-T-Sron Formation (Cambrian). Basemap photograph from Google Earth.
Contacts and faults based on Peach et al. (1907), Laubach et al. (2014), and
field observations from the present study.

2
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

fracture mapping upon rock faces polished and striated during Pleisto­ readily observable, but regional surveys (Laubach and Diaz-Tushman,
cene glaciation (Krabbendam and Glasser, 2011). The outcrops we focus 2009; Hooker et al., 2011) indicate that fractures range from hierar­
on are present meters away from the Moine Thrust Belt (Fig. 1), a chical to unbounded, in the sense of Hooker et al. (2013). Fractures of
regional suture zone reflecting Caledonian convergent tectonics (Trewin various orientations tend to abut or branch rather than crosscut one
and Rollin, 2002). Regional offshore and onshore surveys demonstrate another (Fig. 2). Where macrofractures are closely spaced they may
multiple distinct episodes of widespread faulting, including Devonian coalesce into faults (Figs. 2 and 3), some of which are visible on aerial or
ENE-WSW extension and Permo-Triassic NW-SE extension (Wilson et al., satellite images (Fig. 1) and have meters to tens of meters of offset.
2010). Fission-track analysis has revealed a complex post-Caledonian Microfractures were observed using SEM-CL, which allows precise
structural history, with multiple burial-exhumation episodes (Holford delineation of microfracture quartz cements in optical continuity with
et al., 2010). Consistent with these regional observations is the presence host-grain quartz (e.g., Laubach et al., 2004a, b). SEM-CL maps of
of multiple sets of cemented, opening-mode fractures having consistent microfractures illustrate systematic sets of quartz-filled microfractures
crosscutting relationships (Laubach and Diaz-Tushman, 2009; Elmore based on subparallel strikes, crosscutting relationships, and CL color
et al., 2010). (Fig. 3c). Microfractures locally contain crack-seal texture and
Macrofractures, here defined as fractures sufficiently large to be seen commonly crosscut one another, with little change in orientation near
without use of a microscope (> approximately 0.1 mm in aperture), are their intersections (Fig. 3c). We observed microfractures within three
present as centimeters to tens of meters-long, planar, near vertical samples—M1, M2, and M3 (Fig. 2)—taken from within the macroscopic
structures, containing little or no quartz cement (Fig. 2). We focus on a fault zone, at the margin of the fault zone, and outside the fault zone,
plan-view exposure of fractures (Fig. 2) so layer boundedness is not respectively. We quantified microfracture frequency using 1D scanlines
of observations, drawn upon layer-parallel thin sections. Along each
scanline we recorded fracture kinematic aperture and position where
each fracture intersects the scanline (e.g., Gillespie et al., 1993). We
trigonometrically corrected fracture apertures and spacings. To each
aperture size we assigned a cumulative number (Equation (1)), then
divided that number by scanline length, giving the cumulative fre­
quency for each aperture size. Plotting cumulative frequency versus
aperture size (Fig. 2b) gives all the information of a size histogram
without the need for selecting a bin size.
We also quantify fracture network properties from field and SEM-
scale photomosaics (Fig. 3b, d) according to the I-X-Y method of Man­
zocchi (2002) and Sanderson et al. (2018). In this method, I nodes are
defined as isolated fracture terminations. Y nodes form where three
fracture segments meet, for example at a branching point or an abut­
ment of one fracture against another. X nodes form where four segments
meet, usually marking a crosscutting intersection. An I-X-Y ternary plot
(Fig. 3e) accounts for a network’s relative abundance of the three types
of node. Differences in I-X-Y proportions are apparent between the
macro scale, which is dominated by Y nodes, and the micro scale, which
is dominated by I nodes, with subsidiary X nodes outnumbering Y nodes
(Fig. 3e, Table 1).
Interpretation. The microfractures and macrofractures we observed in
the Eriboll sandstone are likely not genetically related, based on a lack of
correspondence in strike (Fig. 2) and degree of cement infill. Further­
more, if microfractures and macrofractures formed together, then we
might expect microfracture frequency to systematically increase near
the macroscopic fault zone—or even to systematically decrease, if fault
slip resulted in stress shadowing. Instead, we observe no consistent
relationship between microfracture frequency and distance (Fig. 2). The
salient difference between microfracture and macrofracture networks is
that the latter are dominated by Y nodes, manifest as abutting in­
tersections. We interpret that the macrofractures are younger, having
formed after the microfractures were filled with cement. Macrofracture
propagation was halted where fractures intersected pre-existing, unce­
mented fractures. Halting was enabled by sliding along the pre-existing
fracture. Previous analysis suggests that during evolution of wing-crack
arrays into faults, an opening-mode “parent” crack orientation is critical
to promoting sliding, localization, and incipient fault slip (Myers and
Aydin, 2004).
In contrast, microfractures commonly form X nodes where they
Fig. 2. (A) Macrofractures at Dundonnell outcrop. Highlighted in yellow is a intersect, meaning they crosscut rather than abut. Distinct CL response
segment of a macrofracture swarm, forming an incipient fault via linkage of of microfracture cement implies distinct generations, where previous
uncemented opening-mode fractures. Locations of microfracture samples
fractures opened and sealed before later generations of fractures form.
shown, with corresponding rose diagrams of microfracture strike. (B) Aperture-
This interpretation is consistent with multiple fracturing episodes
size cumulative frequency distribution of microfracture populations measured
using SEM-CL in samples M1, M2, and M3. Note M3 has the greatest micro­ throughout a protracted tectonic history in a rock prone to brittle
fracture frequency and lies at greatest distance from macrofracture swarm. (For deformation (Laubach and Diaz-Tushman, 2009; Holford et al., 2010;
interpretation of the references to color in this figure legend, the reader is Hooker et al., 2011). Cementation precluded sliding on pre-existing
referred to the Web version of this article.) microfractures, enabling continued propagation and the formation of

3
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 3. (A) Detail of macrofracture array shown in Fig. 2, with fracture network nodes interpreted—see text. G.S.: glacial striations. (B) Interpretation of fractures and
nodes in (A). (C) SEM-CL image of microfractures, sample M3 (Fig. 2). (D) Interpretation of fractures and nodes in (C). (E) Ternary plot showing frequencies of I, X,
and Y nodes, from micro- and macrofracture populations in Fig. 2.

2.2. Contrasting fracture-bedding relations: an example from the

Table 1 Appalachian basin
I-X-Y data, Dundonnell outcrop (Scottish Highlands).
Sample I nodes X nodes Y nodes The previous example shows how the prevalence of abutting, rather
Outcrop macrofractures 73 88 262 than crosscutting, can drastically impact resulting fracture patterns.
M1 32 8 5 Abutting is favored where traction-free (barren), pre-existing fractures
M2 18 7 0 can accommodate the opening of a new intersecting fracture via sliding
M3 68 19 13
and formation of a Y node. Crosscutting, X nodes are favored where
fractures are sealed and cannot slide. For the same reason­
X nodes. Similar resistance to reactivation stemming from cementation —sliding—bedding planes may be equally important as pre-existing
was inferred from fracture patterns in the Flathead sandstone by For­ fractures in sedimentary rocks. Shales, in particular, are known for
stner and Laubach (2022). their mechanical anisotropy, imparted by weak bedding planes, which
has been shown to affect fracture propagation in previous studies (e.g.,
Peacock and Sanderson, 1992; Gomez-Rivas and Griera, 2012; Lee et al.,
2014; Hooker et al., 2020; Haluch et al., 2023). An example derived

4
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

from cores through the Marcellus Formation in the Appalachian basin interpreted to stem from organic maturation, low permeability, and
shows how fracture orientations in folded, laminated rocks vary signif­ mechanical properties of shales (Hooker et al., 2017a; Engelder and
icantly by degree of cementation. Here again, fracture style reflects the Gross, 2018).
combined effects of the physical and chemical environment of fracture Core for this study sampled the Marcellus shale (Middle Devonian).
formation. Most fractures are parallel or perpendicular to bedding, which dips
Fractures were recovered from shallow (maximum true vertical about 30◦ to NNW. The database includes 496 total fractures, of which
depth 400 ft (120 m)) vertical core drilled through the Marcellus shale. 54 show evidence of shear offset, in the form of slickensides, slick­
Core was collected from a moderately (~30◦ ) dipping forelimb in the encrysts, or pressure-solution cleavage oblique to fracture walls.
Valley and Ridge province of the Appalachian basin in central Penn­ Opening-mode fractures have planar geometries and discrete, smooth
sylvania, USA (Fig. 4). The Appalachian basin includes a Paleozoic walls (Fig. 5). Because of the limited fracture height preserved in core,
foreland basin sedimentary sequence that was folded into kilometer- fracture tips are not generally preserved; however, tips that are pre­
scale anticlines with associated thrust faults during the Alleghanian served indicate that fractures commonly abut against bedding planes.
orogeny (Faill, 1998). Structural style reflects both stratigraphic archi­ Owing to extensive core breakages, along fractures and bedding planes,
tecture and burial depth. Kilometer-scale folds were detached above fracture strike and dip were identified within ten-degree windows, and
Cambrian shales in the Valley and Ridge province, producing relatively not more precisely.
tight fold geometries, compared to the broad anticlines within the Ap­ Fractures are dominantly calcite cemented, with subsidiary quartz
palachian plateau to the northwest, where Silurian evaporites serve as and pyrite. Fractures were grouped by degree of cementation according
the detachment (Mount, 2014). Areas of greater burial depth, such as in to the scheme shown in Table 2.
eastern Pennsylvania and southwestern Pennsylvania, western Mary­ Group 0 fractures are dominantly bedding-perpendicular and strike
land, and West Virginia, have relatively widely spaced thrust faults, NE, with subsidiary NW-striking counterparts (Fig. 6). Group 1 fractures
compared to the study area in central Pennsylvania, where the show strike patterns similar to those of Group 0, but with more variation
maximum burial depth was shallower and the thrust faults are more in dip. Although most Group 1, NE-striking fractures are bed-
closely spaced (Evans, 2023). perpendicular, a considerable subset dips more shallowly to SE. These
Regional fractures strike systematically parallel and perpendicular to latter fractures are therefore oblique to bedding and to the horizon.
fold axes (Srivastava and Engelder, 1990), with steeply dipping, Group 2 and 3 fractures are almost uniformly parallel or perpendicular
fold-axis-oblique fractures present that pre- and post-date folds (Evans, to bedding, omitting fractures with evidence of shear displacement.
2010; Evans et al., 2014). Wilkins et al. (2014) noted that curvature of Group 3 includes a population of bedding-parallel fractures, which are
the thrust belt into the Pennsylvania salient could explain scatter in entirely filled with cement.
fracture orientation, and so favored a single, though protracted, brittle Interpretation. Group 3 fractures are thoroughly cemented, in some
deformation event. Layer-parallel fractures are particularly abundant in cases with fibrous cement, interpreted to form during fracture opening,
organic-rich shales, including within the sampled Marcellus Formation, based on kinematic models (Urai et al., 1991). The thorough cementa­
as detailed below. The broad distribution of layer-parallel fractures in tion of Group 3 fractures is unlikely to have coincided closely in time
the overlying shales of the Catskill Delta Complex, and in organic-rich with opening of Group 0 or 1 fractures, which contain no or little
shale layers throughout the Appalachian plateau in northwestern cement. A parsimonious interpretation is that Group 3 formed earliest,
Pennsylvania and in New York (Engelder and Gross, 2018) has been followed by Groups 2, 1, and 0, accounting for the paucity of cement as a

Fig. 4. Geologic setting for Appalachian basin core sample. Approximate core location shown (circle). Dha: Devonian Hamilton Group, the base of which is the
Marcellus shale. Dol: Devonian Onondaga limestone. Silurian map area contains lower-most Devonian, the contact between which lies within the undivided Keyser
and Tonoloway formations. Background photograph taken from Google Maps; geologic contacts modified after Berg et al. (1980).

5
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

the most heavily cemented fractures (Groups 2 and 3) are also arranged
parallel and perpendicular to bedding, in contrast to lightly cemented
fractures (Group 1), which lie oblique to bedding. Fractures can form at
any time during the burial history of a sedimentary basin, even at very
early, pre-lithification stages (Hooker et al., 2017b; Petit et al., 2022);
nevertheless, fractures that strike parallel and perpendicular to
map-scale folds, as well as layer-parallel fractures, are all anticipated to
form during various phases of the evolution of fold-thrust belts (Ferrill
et al., 2021). This evolution proceeds from a normal faulting stress
regime during basin extension, to strike-slip and thrust-faulting with
increasing horizontal tectonic load. Therefore the present fracture
groups can all be interpreted in the context of Alleghanian tectonics.
Furthermore, fluid inclusion studies have linked regional fractures to
Alleghanian fold tightening and concomitant fluid-flow (Evans, 2010).
The opening-mode fractures that lie at high angle to bedding (Fig. 6)
are kinematically coherent as pre- or early-syn-folding structures. For
NE-striking fractures, a bed-perpendicular fracture will dip about 60◦ to
SE. However, many NE-striking Group 1 fractures dip more shallowly to
SE. If the dip were steeper, then a potential interpretation would be that
these fractures formed subvertically, post-folding, and have not been
rotated. Instead, their shallow dip would still restore to an oblique dip,
even in restored bedding. This orientation is consistent with formation
amid a thrust-faulting regime, such as the one that was present during
Alleghanian folding (Engelder and Whitaker, 2006) or the one that
persists today (Heidbach et al., 2018).
Interestingly, the attitude of uncemented Group 0 fractures is
roughly perpendicular to bedding (Fig. 6). A potential explanation for
this change in fracture orientation is that the Group 0 fractures formed at
the shallowest levels, wherein the low burial stresses facilitated decou­
pling between strata, polarizing the stress field and driving the fractures
in a more layer-perpendicular orientation, as is commonly observed for
cross-joints linking pre-existing parent joints (e.g., Ji et al., 2021). If that
explanation is true, then the oblique orientation, with respect to
bedding, of Group 1 fractures is consistent with their formation under
greater burial stress, which locked bedding planes and inhibited any
such control of bedding on fracture orientation. This interpretation,
Fig. 5. Natural fractures recovered from State Game Lands 252 core. (A) Group though speculative, is also consistent with thicker cements accumulating
0 fracture, entirely lacking cement, depth 141 ft. Core has split along shale on older fractures that formed deeper and under greater thermal
bedding plane. (B) Group 1 fracture, with thin, inconspicuous cement lining, exposure.
cutting two Group 3 fractures, filled with white calcite cement. Depth 188 ft.
(C) Group 2 fracture, with white but thin (<1 mm) cement lining, which is
2.3. Variation in fracture intensity with depth: an example from the
patchy rather than fully blanketing the fracture surface. Depth 191 ft.
Piceance basin

Table 2 The previous two examples show contrasting fracture patterns that
Categorization of opening-mode fractures based on fill, Appalachian basin. also have varying spanning potentials. Those spanning potentials appear
Cement Total Description
to have materially affected the resulting pattern, but in each case the
Group N protracted geologic history would have introduced other confounding
variables, such as differential stress, strain rate, or fluid fluxes. To con­
0 79 No evidence of cement
1 203 Walls continually or patchily covered by a patina of trol for such variables, our final example focuses on fractures recovered
mineral cement, <0.1 mm thick, which has a translucent from core at relatively great depth (2000–13,000 feet; 600–4000 m),
or sugary appearance from a basin with a comparatively simple burial history, and having a
2 72 Walls continually or patchily covered by mineral cements single predominant, regional fracture set. Therefore, we argue, varia­
that are locally thick enough to be opaque and white, but
throughout <1 mm thick
tions in fracture pattern attributes—particularly fracture intensity—­
3 92 Walls continually covered by opaque, white mineral with depth can be attributed to depth-dependent variables like
cements; thicknesses commonly 1 mm or more temperature, pressure, and perhaps fluid chemistry.
The Piceance basin (Fig. 7) is an intermontane sedimentary basin,
whose sediments were deposited within the Cretaceous Western Interior
consequence of less thermal exposure as fracturing proceeded during
Seaway that covered much of North America (Kauffman, 1984). The
exhumation. However, we have little direct evidence of relative timing
stratigraphic sequence contains gently folded, fluvial to marine sand­
in the form of crosscutting relationships. It may rather be that some
stones and shales, and hosts a regional natural fracture set striking ENE,
fractures escaped mineralization as a consequence of gas generation,
with subsidiary sets striking at high angle to this regional set (Lorenz and
which may have driven off water, or by influx of undersaturated surface
Finley, 1991). These fractures have garnered considerable attention for
water, which may have diluted pore waters. We can conclude conser­
their effects on storage and flow of natural gas (Cumella and Scheevel,
vatively that the four fracture groups formed at distinct times.
2008). Combined burial history models and fluid inclusion studies
The four fracture groups vary systematically in orientation, also
support a basin-centered gas accumulation model for natural fracturing
suggesting a distinct time of formation for each group. We observe that
here (Fall et al., 2012), whereby maturation of coal beds at depth

6
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 6. Orientation of natural fractures recovered from Appalachian basin core. Orientations shown with bedding restored to horizontal. Azimuths (degrees) listed
down either side represent the center of the strike window for the corresponding histogram. X-axis indicates maximum of dip window (degrees), with 0 indicating
layer-parallel fractures. Symbol fill legend indicates kinematic mode (opening or shear) and cement group (Table 2). For example, “I0” means mode-1 (opening-
mode) and Group 0; “II3” means mode-2 (shear-mode) and Group 3. Note moderately dipping fractures are dominated by opening-mode, Group 1 fractures—see
especially fractures striking 045–135◦ .

resulted in fluid overpressures throughout the low-permeability sand­ Here we focus on 47 core samples retrieved from fractured sand­
stones above, eventually leading to fracture opening—as opposed to a stones of the Mesaverde Group, from two distinct localities within the
conventional petroleum system in which gas accumulates in geometric basin; namely, two clusters of oil fields in the north and south parts of
traps, created by fractures or otherwise. the basin (Fig. 7). These two geographically distinct groups of samples

7
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

have undergone similar maximum thermal exposure, based on vitrinite

reflectance (Nuccio and Roberts, 2003, Fig. 7). The southern sample
suite (20 samples) currently lies at a shallower level (2000–8000 feet
(600–2500 m) true vertical depth versus 8000–13,000 feet (3500–4000
m) in the northern suite of 27 samples), owing to a steeper uplift history
in the south (Fall et al., 2015).
Consistent with previous SEM-CL surveys from the Piceance basin
(Hooker et al., 2009), macroscopic fractures (Fig. 8a) are present amid
parallel microfractures (Fig. 8d). Both size ranges contain quartz cement
bearing assemblages of fluid inclusions, aligned generally parallel to
fracture walls, and marked by parallel bands of crack-seal texture,
apparent in SEM-CL (Fig. 8b). Dimly luminescent calcite cement over­
laps quartz and generally does not show crack-seal banding (Fig. 8c).
Larger microfractures (i.e., microscopic fractures wider than about 10
μm) and macrofractures are less likely to be entirely sealed by quartz,
instead preserving considerable pore space or postkinematic calcite
cement (Fig. 8).
Microfracture frequency was quantified by drawing scanlines on
SEM-CL photomosaics constructed to form uninterrupted maps across
core samples, perpendicular to fracture strike. Fig. 8d is a detail of one
such image mosaic. Fracture intensity was then quantified by generating
a cumulative frequency-aperture size plot (as in Fig. 2b) and best-fitting
the y-intercept of a power-law equation having a slope (exponent) of
− 0.8, which is preferred when data are sparse (Hooker et al., 2014). This
y-intercept is therefore equivalent to the predicted frequency of
1-mm-wide fractures, per mm of rock (Fig. 9).
The samples from the north and south fields of the Piceance basin
derive from equivalent strata, but their current true vertical depths do
not overlap (Fig. 9), owing to the aforementioned differential uplift (Fall
et al., 2015). In both fields, we distinguished samples taken from below
Fig. 7. Map of the Piceance basin. Solid contours show depths (in feet) to the
top of the Rollins sandstone (after Cumella and Scheevel, 2008). Vitrinite and above the top of continuous gas saturation, identified by Cumella
reflectance maturity data from Nuccio and Roberts (2003). Dashed lines indi­ and Scheevel (2008). The shallowest samples from either sample site are
cate fields from which north- and south- Piceance basin core samples all among the lowest fracture frequencies observed: less than 0.002
were collected. fractures/mm. In contrast, deeper samples—those from below the top of
continuous gas—show a wide range of fracture frequencies, from as low
as that of shallow samples, to as much as 0.008 fractures/mm near the

Fig. 8. Natural fractures, Piceance basin. (A) Vertical macrofracture in core, south Piceance basin. Macrofracture is partially filled by quartz crystals, not apparent at
scale of photo. (B) Diagram of scanline geometry. Scanlines are made from SEM-CL images taken contiguously, parallel to layering, and across macroscopic fractures.
(C) Color SEM-CL photomosaic of blue-luminescing quartz deposit (Q) bridging across vertical fracture, north Piceance basin. Fracture walls indicated by dashed
lines. Pore space luminesces a yellowish hue. Crack-seal texture is present as fracture wall-parallel bands of blue quartz cement. (D) Grayscale SEM-CL image of a
diffuse cluster of microfractures, north Piceance basin. Quartz fracture cement (Q) luminesces a medium gray. Relatively wide fracture at lower-left is lined with
quartz and filled by non-luminescent, postkinematic calcite cement (C). Thinner microfractures are entirely filled by synkinematic quartz cement. (For interpretation
of the references to color in this figure legend, the reader is referred to the Web version of this article.)

8
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

1993). These considerations motivate a new modeling approach to

combined subcritical fracturing and cementation.

2.4. Summary of geologic examples

The evidence from the data collected points to a range of depth-

related influences on the resulting fracture pattern. In the Scottish
Highlands example, closely spaced, synkinematically cemented frac­
tures have microscopic apertures and a tendency to open in crack-seal
fashion, in which fractures reactivate without changing orientation, or
causing slip on previous fractures. In contrast, macroscopically visible,
partially open fractures grow to form complex networks in which frac­
tures abut and reactivate one another, eventually coalescing to form
faults.
In the Appalachian basin example, likely deep-seated, Group 3 and 2
cemented fractures are parallel and perpendicular to bedding. Lightly
cemented, Group 1 fractures formed oblique to bedding in a thrust-
faulting stress regime, likely at a late stage with respect to regional
folding. Barren, Group 0 fractures formed parallel and perpendicular to
bedding, despite bedding’s oblique orientation with respect to the hor­
izontal. Such a pattern could arise from shallow overburden and
decoupling of bedding planes, which, in the absence of fast cementation
or creep processes, created surfaces free of shear tractions that reor­
iented the stress field parallel to those surfaces.
In the Piceance basin, natural fractures are present in parallel arrays
of widely spaced macrofractures and closely spaced microfractures. All
fracture sizes include fracture-spanning cements, but the microfractures
are more thoroughly filled by cements, pointing to a control on fracture
size and spacing by those cements, and the degree to which fractures are
sealed as they grow.
Fig. 9. Fracture intensity versus depth, Piceance basin. Intensity is calculated In all three cases, the depth of burial during fracture formation ap­
according to the method of Hooker et al. (2014)—see text. Circles, south pears to have influenced fracture pattern development. Greater burial
Piceance basin (Fig. 7); squares, north Piceance basin; open symbols, samples depth is expected to increase vertical compression and confining stress,
above top of continuous gas saturation (Cumella and Scheevel, 2008); filled one result of which is to inhibit slip along bedding planes and pre-
symbols, samples from within the gas-saturated zone. existing fractures. Other possibilities exist as well, but we suggest that
an underappreciated control on fracture patterns stems from
greatest depths sampled (Fig. 9). temperature-dependent processes of cement precipitation and crack-tip
Interpretation. Crack-seal texture in quartz (Fig. 8c) indicates quartz corrosion reactions. Because all of these processes are expected to vary
precipitation during fracture opening that was rapid enough to span the systematically with burial depth, their resulting effects on fracture
gap from one fracture wall to the opposite wall, often at spatially iso­ pattern growth are difficult to understand in isolation, without
lated quartz deposits or bridges surrounded by pore space or post­ modeling.
kinematic calcite. Where crack-seal texture forms, continued opening of
fractures apparently necessitates re-breaking of cements, based on crack 3. Spring lattice model
opening increments within fracture-spanning cement deposits in Fig. 8c.
In some cases, re-breaking is well localized within a single fracture, with To explore the ramifications of this interpretation, we employ a
little or no host-rock preserved amid successive fracture opening in­ numerical model that simulates both of the temperature-dependent, and
crements (Fig. 8c). In other cases, microfractures are present as clusters therefore depth-dependent, chemical processes that are hypothetically
of distinct fractures, preserving host-rock in between (Fig. 8d). It was at work during fracture pattern evolution: cementation of fractures and
proposed that the degree of cementation itself plays some role in corrosive rock failure. Here we examine these effects in a hypothetical,
localization (Hooker et al., 2014), based on fracture traces that are structurally simple setting with steady lateral extension as a boundary
locally diffuse, where spanning potential was high and crack-seal in­ condition. To fully recreate the complex deformations that occur in
crements abundant, and locally discrete, where spanning potential was nature, particularly in compressional belts like our Caledonian and
low and crack-seal texture rare. The interpretation was that high span­ Alleghanian examples above, would require more robust modeling that
ning potential promoted delocalization of progressive opening in­ is beyond our present scope. The model we employ extends the nu­
crements, and so affected both the size and spatial arrangement of merical model of Hooker and Katz (2015), which simulated a hexagonal
fracture arrays (Hooker et al., 2014). spring lattice (Fig. 10) whose interconnected springs display
However, it is likely that such cements are only part of the story, and linear-elastic behavior. Specifically:
that inelastic processes such as chemically assisted cracking also play a
Fij = kij δΔij (2)
role. For example, bridging cement deposits alone should not change the
stress distribution around fractures, although they can modify the
where F is the force exerted by the spring, k is the spring constant, δΔ is
evolving stress field around fractures opening within a progressively
the change in spring length from its equilibrium length, and subscripts i
extending host rock (Hooker and Katz, 2015). In contrast, subcritical
and j refer to spring positions within the lattice (Fig. 10). The model
cracking has been shown to enable a close spacing among fracture
simulates extension fracturing by subjecting the lattice to a constant
populations, as multiple nearby fractures can propagate simultaneously,
along-layer extension rate, V, as a boundary condition. The springs
before resulting stress shadows inhibit nearby fracture growth (Olson,
within the fracturing layer have a finite breaking strength,

9
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

nodes it previously connected, pre-failure:

Δ′ij = Δij + ftij∗ (4)

where t* is the time since failure. At that moment of re-connection, a

new neutral length is set to the current length of spring. The model is
non-dimensionalized by normalizing all lengths to the initial spring
length Δ0; model time is scaled to t = Δ0/V; spring constants are rescaled
by a reference value k0; forces are then rescaled by k0Δ0. Full model
details, including implementation of the boundary conditions, are given
in Hooker and Katz (2015).
The numerical experiments of Hooker and Katz (2015) suggested
that, holding all else constant, both fracture aperture and spacing can be
expected to decrease with increasing cementation rate.
This model is a simplification of nature in many respects, most
especially in its assumption that deformation is purely elastic in settings
in which fractures grow and fill with mineral cements. Here we extend
this model by including a term that is intended to simulate time-
dependent dissolution reactions that are thought to be particularly
active at the highly tensile regions near fracture tips, and especially
where pore fluids are acidic or high-ionic-strength brines at elevated
temperature (e.g., Rinehart et al., 2016). Such dissolution reactions have
been invoked to promote subcritical crack propagation (Atkinson,
1984). Subcritical crack growth, by corrosive reactions concentrated at
crack tips, is likely a key process in natural fracture formation, based on
the very slow growth of natural fractures inferred from fluid inclusion
studies (e.g., Becker et al., 2010), including those of crack-seal fractures
in the Piceance basin strata studied here (Fall et al., 2015). Corrosion
reactions would also provide a potential source of silica for the quartz
cement observed in the sandstone fractures of this study (Figs. 2 and 8).
We enable subcritical spring failure in the model by adding a second,
time-dependent spring-failure criterion. Specifically, we integrate the
force that accumulates on each fracturing-layer spring with respect to
time, and require that springs fail once this integral exceeds a predefined
value, Z:
∫
Fij dtij∗ > Zij (5)
Fig. 10. Spring lattice model. (A) Layer setup. Three layers are composed of
springs. The layers are stretched in the +x direction at constant rate. Springs in This integral increases in value so long as the spring is longer than its
the matrix layers are indestructible; springs in the fracturing layer break once neutral length. This failure criterion is therefore an accumulated time-
their critical breaking length (b) or subcritical failure integral (Z) is exceeded.
stress integral, rather than simply a critical value of stress alone. Both
Inset: Detail of spring lattice at initiation. (B) Breaking and cementation pro­
failure criteria lead to fracture propagation, because tensile stresses tend
cedure. At Time 1, five nodes are connected by springs at their default spring
constants (k) and neutral lengths, Δe. Reference position of springs and nodes
to accumulate at fracture tips. The Z criterion can simulate time-
shown in pale gray at later times. At Time 2, one of the springs is stretched to its dependent reactions that assist fracture propagation. Springs that fail
elastic limit (length = b). The spring fails and the nodes move in response to according to Eq (4) are said to have failed subcritically, by analogy to
force balance. The broken spring snaps back to its neutral length from Time 1. rock fractures that propagate at lower stresses than those required to
The spring grows via cementation at a constant rate, f. By Time 3, the spring has rupture bonds at fracture tips.
grown far enough to reach the separated node. At this time, the spring is re- Of critical interest is how the size and spacing of fractures vary. We
attached with its original spring constant k and a new neutral length is quantify resulting failure patterns based on the x-coordinates of nodes
assigned. (C) Quantification of strain heterogeneity, Q. Dashed lines represent along the middle row of our fracturing layer.
the difference between the actual x coordinate and the homogeneous-strain x ∑
coordinate of each node along the middle row (j = Nj/2, rounded up). Q is the |x − xhom |
Q= (6)
sum of each such difference, divided by (Ni × Nj). Ni × Nj

corresponding to a maximum length b the springs can attain before where x-xhom is the difference between a node’s x coordinate and the x
breaking: coordinate it would occupy if deformation were homogeneous
(Fig. 10c). We design our statistic to reflect the geometry of the fractures
Δ0ij + δΔij = bij (3) within the bed, rather than the resolution of the nodes. Dividing by Ni
averages the excess displacement among the nodes, and so controls for
where Δ0 is the initial spring length, and also the starting equilibrium the horizontal resolution of the model space. Dividing by Nj normalizes
length. Springs within the matrix layers are indestructible. the excess displacement to the layer thickness. This step is important
Fractures form when sets of adjacent springs fail; heterogeneity in because we anticipate that fracture width scales with length and
strength is introduced either through randomly distributed pre-broken therefore layer thickness.
springs, variation in critical breaking length b, or variation in spring We track a quantity analogous to porosity in the model as the sum of
constant k. Hooker and Katz (2015) introduced cementation by all broken spring lengths divided by the sum of all spring lengths, within
re-growing broken springs at a constant rate f, restoring the strength of the fracturing layer. These broken springs mechanically represent the
the spring once its new length Δ’ reconnects across the gap between the uncemented parts of fractures and, at the initiation of model runs, the

10
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

flaws dispersed throughout the host rock. fracturing and matrix layers, and healing is virtually absent (Fig. 13a
We ran a suite of 20 simulations (Table 3) aimed at investigating and b). Fractures centered at these layer-boundary breakages are nearly
fracture opening at varying values of depth and thus temperature. We box shaped, in that there is very little tapering of the fracture walls from
hypothesize that increasing temperature increases kinetic reaction rates, the center of the fracture toward the tips. Holding Z constant but
thereby increasing cement precipitation rates and subcritical failure increasing cementation rate f to 10, fractures are more tapered, and
rates. We implement the former by increasing f, which seals fractures failures along the layer boundaries are readily apparent (Fig. 13c), but
more quickly, and we implement the latter by decreasing Z, which they heal repeatedly throughout the simulation (Fig. 13d). Indeed, more
promotes subcritical failure. Parameters used in the simulations for this healing-rebreaking episodes are apparent near fracture tips and layer
study are listed in Table 3. boundaries, compared to fracture centers.
The effect of decreasing Z, thus accelerating subcritical failure, on
4. Results fracture propagation, can be appreciated by comparing Fig. 13a and e. In
the latter figure, in which Z = 0.1 and f = 0.1, healed springs are
Fig. 11 shows the state of the spring lattice at a final layer-parallel common only in the host rock, where displacement between neigh­
stretch (final length divided by initial length) of 1.6 for various values boring nodes, beyond their initial separation, is negligible. These springs
of subcritical failure criterion Z and cementation rate f. This large stretch broke because the subcritical failure criterion will be reached even with
was used so that the locations of fractures would be clear while using a tiny amounts of elastic lattice-extension, given sufficient time. Upon
relatively low-resolution, computationally efficient lattice. The model breakage, the extension (and hence tensile stress) between neighboring
contains no specific definition of a fracture, although arrays of broken or nodes was so small that breaking of the spring resulted in only a small
re-healed springs can be recognized as extending vertically, or approx­ change in node position, and the gap between the nodes was quickly
imately so, across layering (Fig. 11). healed, even under a slow cementation rate. Amid this relaxed stress
Cementation rate f increases toward the right in Fig. 11; Z decreases state in the host rock, fractures tended to grow gradually, via a combi­
downward. Therefore the two hypothetical effects of increasing nation of subcritical and critical spring breakages. As a result, fractures
depth—faster cementation and faster breakage by subcritical fail­ are numerous and narrow, with a tendency to branch and cluster
ure—increase downward and to the right in the figure. At the top left, (Fig. 13e), compared to the high-Z default (shallow) scenario (Fig. 13a).
the spring failures manifest primarily as isolated fractures that are wide As f approaches infinity and Z approaches zero, the lattice extends
and widely spaced, and that span the fracturing layer from top to bot­ homogeneously (Fig. 11, bottom-right). Here all nodes approach the
tom. Cementation and subcritical breakages are absent. At the bottom subcritical failure criterion at low lattice extension, and the propagation
right, deformation is close to homogeneous. Most springs are currently of failures that produces fracturing in higher-Z simulations is minimized
cemented; a subset is currently broken via subcritical failure, and these because abundant subcritical failures happen at such small lattice-
latter are homogeneously distributed throughout the lattice. Discrete extension values that little tensile stress is imparted to neighboring
fractures are not present. springs upon breakage. As we discuss below, this model space is anal­
At intermediate positions, fractures are present, but tend to be ogous to cleavage formation at great depth. But, at intermediate f and Z
smaller, compared to the shallow end-member. Fractures also have a values (e.g., f = 1; Z = 3), simulating depths where both cementation
wider range of widths, lengths, and spacings—particularly for f = 10, Z and subcritical fracturing exert a modest effect with neither over­
= 6. Where cementation is slow and subcritical cracking dominates (e.g., powering the other, patterns emerge wherein fractures have a wide
f = 0.1, Z = 1), fractures are commonly shorter than the fracturing-layer range of sizes and spacings (Fig. 13g). Here large fractures span the
height, showing a tendency to branch. entire fracturing layer and preserve porosity, especially near the center,
Because spring strength is re-established upon cementation, cemen­ where fractures are widest. These large fractures are surrounded by
ted springs can re-fracture (Fig. 12). This behavior is analogous to crack- small, mostly sealed, microfractures, which manifest as linear arrays of
seal deformation, visible as myriad fracture opening increments having broken and re-healed springs, which generally do not extend across the
microscopic widths, but that form side-by-side, thereby comprising entire fracturing layer (Fig. 13g).
larger composite fractures (Fig. 8). We track the number of times each Fig. 14 graphs the strain heterogeneity index Q and porosity as a
individual spring breaks throughout the simulation, sampling for new function of stretch (final layer length divided by initial layer length) for
breaks with each recorded model output, which we make at regular each simulation. Each simulation begins with a porosity of approxi­
intervals during extension. Re-breaking is common, particularly along mately 0.1, a consequence of setting B* to 0.1; i.e., beginning with 10%
bed-boundaries at moderate f (Fig. 12; f = 1, Z = 6, 10); throughout large of springs broken. This step helps introduce heterogeneity at an early
veins at high f (Fig. 12; f = 10, Z = 6, 10); and throughout the entire stage and tends to stabilize convergence of numerical solutions. With
fracturing layer at low Z (Fig. 12; f > 0.1, Z = 0.1). progressive cementation, this initial porosity declines to near 0; with
High-resolution images from Figs. 11 and 12 illustrate some salient increasing cementation rate, the initial porosity does not appear in
outcomes of the simulation (Fig. 13). In our shallowest simulation (f = Fig. 14, having already been filled by cement by the model time at which
0.1, Z = 10), spring failures are abundant at the boundary between the results are extracted for the plot.
At low cementation rate f and high subcritical failure criterion Z (i.e.,
shallow-level simulations), porosity and strain heterogeneity both tend
Table 3 to increase over time, after the initial porosity decrease (Fig. 14; f = 0.1,
Model parameters. Z = 10). The effect of decreasing Z is to moderate porosity, such that it
Parameter Explanation Number/Range initially spikes and then declines to roughly a steady-state value (Fig. 14,
used f = 0.1; Z = 0.1, 1). At very low Z, strain heterogeneity likewise reaches a
Ni Number of nodes in x direction 60 plateau early in the simulation (Fig. 14; f > 0.1, Z = 0.1). The effect of
Nj Number of nodes in y direction 90 high f, relative to the shallow case, is mostly to decrease porosity. Pore
Njb Number of nodes in matrix layers 15 space opens rapidly as fractures repeatedly open, then the fractures
Wk Spring constant distribution parameter 5 quickly seal, producing a spiky signal with troughs near 0 (Fig. 14; f =
kpa k perturbation amplitude 0.5
B* Fraction of fracturing-layer springs initially 0.1
10, Z > 3). In these simulations, strain heterogeneity produces a stair-
broken step pattern, with discrete jumps in heterogeneity coinciding with pos­
bm Breaking length minimum 0.2 itive spikes in porosity, which mark the opening of fractures. At deep
Z Subcritical breaking integral minimum –[0.1,1,3,6,10] simulations with high f and low Z, porosity tends to oscillate with
f Cement accumulation rate –[0.1,0.5,1,10]
amplitude that decays toward a steady state, amid generally

11
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 11. Springs simulation results. Unbroken springs are black; healed springs are green; broken springs are red if broken by overextension, magenta if broken
subcritically. Unbreakable matrix-layer springs, in gray, lie above and below the fracturing layer. Locations of detailed lattice areas in Fig. 13 shown. (For inter­
pretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

synchronous fluctuations in strain heterogeneity (Fig. 14; f > 0.1, Z = affects mineral spanning potential through the inherent growth anisot­
0.1). ropies of various minerals as well as the tendency of some minerals to
preferentially accumulate on certain substrates (Lander and Laubach,
5. Discussion 2015). Our modeling work addresses the potential for temperature and
pressure to affect fracture patterns through their effects on chemical
Our model produces different types of patterns (Figs. 11 and 12) that reactivity.
would correspond to different temperature-pressure-composition con­ Moving from shallow (low cementation rate f, high subcritical failure
ditions. The results therefore delineate what we call fracture facies, based criterion Z) to deep (high f, low Z) end-members of our simulations
on the fracture patterns’ relationships to the environmental formation produces a spectrum of fracture patterns (Fig. 15). At the shallow
conditions. “Fracture facies” is also used where fracture patterns are extreme, fractures are discrete arrays of broken springs having wide
particular to the hosting sedimentary facies (e.g., Bruna et al., 2015; Li spacings, simple geometries, and large amounts of porosity (Fig. 14). We
and Robinson, 2018), and so this term could also be applied to meta­ refer to this end-member pattern development as “exclusionary macro­
morphic facies. The physical constituents of any rock will control its fracturing.” Here fractures tend to propagate the entire vertical distance
mechanical properties and therefore influence its fracturing behavior, at across the fracturing layer, and the resulting relief of tension creates an
a given temperature and pressure. Moreover, rock composition also exclusionary zone or stress shadow to either side, in which the opening

12
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 12. Spring lattice simulation results. Circle diameter proportional to the number of times a spring has broken.

of new fractures is suppressed. In such cases of low f and high Z, fracture produce isolated fractures, but instead the neighboring node-separation
size, spacing, and porosity distribution-evolution are unaffected by is close to uniform. Cementation is rapid and springs readily fail sub­
chemical processes. critically, such that springs break and re-seal many times throughout the
At the deep extreme (Fig. 15, bottom-right), the model does not simulation, without the buildup of large stress concentrations. Strain

13
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 13. Details of spring lattices from model output. (A, C, E, G): springs, colors as in Fig. 11). (B, D, F, H): circle size proportional to number of times broken, as in
Fig. 12. See Fig. 11 for locations and text for discussion. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of
this article.)

heterogeneity and porosity are low and reach their maxima early in the tectonic “stretching veins” (Bons et al., 2012; their Figure 17b). The rock
simulation (Fig. 14 Z = 0.1, f = 10). This behavior mimics the devel­ deformation and evolving stress distribution in such cases is akin to that
opment of myriad very closely spaced, sealed fractures, which have been which develops in compressive environments as a disjunctive cleavage
noted in sandstones (Laubach, 1989; Onasch, 1990; Hooker et al., 2014; (Engelder, 1979). There, dissolution of soluble material from closely
their Fig. 11b) and limestones (Davis, 2014; Hoyt and Hooker, 2021; spaced, microscopic surfaces achieves a volume loss. Based on the
their Figure 16). The texture sometimes forms at the slow-opening tips of similar distribution of structures, only here resulting from extension and

14
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 14. Spring lattice simulation results. Porosity and strain heterogeneity (Q) plotted versus stretch, which is equivalent to model time.

rapid cementation of ephemeral fracture pore space, we term this cementation with higher f (Figs. 11, Figure 12, Fig. 13). Our choice to
behavior “penetrative microfracturing.” make matrix layers indestructible forces fracturing to continue, even to
For exclusionary macrofracturing, note that we anticipate layer great width/height ratios, in all simulations, but in reality, those matrix
delamination to be an important process in halting fracture propagation layers would eventually fail, and likely at lower extension values in cases
(Fig. 15), consistent with model results (Figs. 11, Figure 12, Fig. 13). We where cementation precludes delamination and layer-parallel sliding.
also note open fractures in the Dundonnell outcrop are more likely to At intermediate positions on the spectrum, rock failure and cemen­
halt, or deflect, the propagation of later intersecting fractures, compared tation are relatively balanced, in the sense that some growing fractures
to the small halting power of cemented fractures (Fig. 3). Likewise, we become sealed while others remain open. Resulting fracture sizes and
hypothesize that delamination may be relatively important for the spacings are more variable (Fig. 15 c-h). Large fractures form amid
orientation of uncemented fractures in the Marcellus core samples we clusters of small fractures, and many fractures do not extend all the way
documented (Fig. 6). Delamination tends to be suppressed via across the fracturing layer (Fig. 11). Although our model resolution is

15
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 14. (continued).

insufficient to illustrate fracturing opening over multiple orders of hot, mineral-laden fluids invade an otherwise cool rock body. Here
magnitude, the relatively wide range of widths produced is analogous to cement precipitation can strengthen the rock, including within previ­
the power-law size distributions of kinematic apertures commonly ously open fractures (Laubach, 1988; Laubach et al., 2009; Major et al.,
observed among cemented fractures (Fig. 2b). Such fractures also 2018; Callahan et al., 2019a), at shallow levels in which host-rock
commonly contain crack-seal cements, implying multiple episodes of re- deformation is close to elastic (Callahan et al., 2019b).
cracking, as observed at high f (Fig. 12). We term this fracture facies A useful metric for judging ‘rapid cementation’ and its contrasts with
“multi-scale” fracturing. depth and rock type is spanning potential (Lander and Laubach, 2015).
The extreme off-axis positions, namely, positions C and G in Fig. 15, Although the example in that paper is specifically quartz, the principle is
reflect conditions for a combination of shallow and deep settings, but the same for all mineral systems. For a constant, temperature based
may nonetheless correspond to natural fracturing environments. In mineral accumulation rate and fractures opening at the same rate, rock
Fig. 15c, cementation is rapid yet subcritical failure is minimal. Such a composition can lead to differences in mineral spanning. For example, in
situation could correspond to shallow geothermal reservoirs, in which sandstones, quartz has diminished spanning (lower overall

16
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Fig. 15. Conceptual diagram of the effects of

subcritical failure and cementation on extensional
fracture patterns. Note correspondence in axes with
Figs. 11, 12 and 14. With increasing depth and tem­
perature, we move toward the lower-right on the di­
agram. Nine patterns or fracture facies (A through I)
illustrate fractures forming in a brittle layer encased
within unfractured matrix layers. White fractures are
entirely cement-filled; blue fractures are at least
partially open. Hypothetical shallowest-forming
fracture pattern is (A), here the “exclusionary mac­
rofracturing” facies, and both cementation and
subcritical failure increase with depth. We therefore
expect increasing burial depth to correspond to a shift
toward the lower-right on the diagram, through the
“multi-scale” facies (E) and beyond to the “penetra­
tive microfracturing” facies (I). See text for descrip­
tion and interpretation of the remaining facies. (For
interpretation of the references to color in this figure
legend, the reader is referred to the Web version of
this article.)

accumulation) in feldspathic sandstones compared to quartzose sand­ occurrence of “shallow” fracture facies.
stones owing to slower accumulation of quartz on feldspar substrates. Fig. 15g represents a case of rock readily failing subcritically where
Laubach et al. (2014) describe adjacent sandstones cut and fractured by cementation is slow. Such a situation might be present in the deep
the same faults that have different spanning owing to contrasts in feld­ subsurface where cementation is suppressed, whether by the presence of
spar content, and associated differences in aperture size distributions. hydrocarbons or other nonaqueous pore fluids, or where clay minerals
Likewise, in shales and other fine-grained rocks, spanning is suppressed coat potential templating surfaces such as quartz grains. High hydro­
by crystals nucleated on small substrates achieving slow euhedral carbon saturation was invoked as a mechanism to preserve uncemented
growth rates sooner (Lander and Laubach, 2015). As with metamorphic joints through the development of a penetrative cleavage in black shales
facies, rock composition as well as thermal history can influence what and siltstones in the Appalachian basin (Engelder et al., 2001). Many of
constitutes a facies boundary. Thermal exposure often exerts the greatest these fractures are present in variably layerbound, clustered arrange­
control; therefore, where geothermal gradients are low, such as the ments (Tan et al., 2014), a pattern readily explained by subcritical crack
Tarim basin of China (Laubach et al., 2023), we expect deep-seated growth (Olson, 1993). As well, Savalli and Engelder (2005) interpreted

17
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

smooth plumose texture and regular arrest lines as evidence that interactions between fluid overpressure and fracture opening or slip
regional joints propagated subcritically. (Lacazette and Engelder, 1992; Fischer et al., 1995; Olson, 2003; de
Clay coatings were found to have preserved reservoir quality by Riese et al., 2020; Hooker and Fisher, 2021) could dominate
inhibiting cement precipitation on sandstone grains in the Bonaparte fracture-opening sequences in ways that the present work does not
basin of Australia (Saïag et al., 2016). The effect was demonstrated illustrate.
experimentally through growth of chlorite grain coatings by Charlaftis
et al. (2021). Although fracturing generally exposes new surface area 6. Conclusion
according to the composition of the grain that is cut, a pore fluid rich in
suspended clays, or a fracture that preferentially propagates along clay Natural fracture arrays illustrate some of the many effects that depth
minerals, could result in fracture opening amid limited cementation. of formation exerts on natural fracture patterns. Previous studies have
Another potential natural example of abundant subcritical crack documented physical, depth-related effects, including suppressing slip
growth paired with minimal cementation would be crevasses in glacier along bedding planes and pre-existing fractures via high confining
ice (Koehn and Sachau, 2014; Hudleston, 2015; Colgan et al., 2016). stress. We presented a spring lattice model that incorporates two key
Here deformation is dominated by creep, such that ice bonds break and thermochemical effects: synkinematic cementation, which provides
reform via inelastic processes. Although the micromechanics of creep adhesion across fractures, and subcritical failure. Although these effects
are outside the scope of our modeling effort, at maximum susceptibility are sensitive to the chemical environment dictated by host rock
to subcritical failure (minimum Z) our lattice responds to extension in a composition and fluid flow history, they are also promoted by high
similar manner, whereby elastic stresses are dissipated over time by temperatures and high ionic-strength pore waters. These effects are
myriad subcritical breakages and re-sealings (Fig. 11 bottom-left). Cre­ therefore dependent upon thermal history, and thus in many cases burial
vasses represent open fractures and so are “cemented” (filled by ice) history, suggesting that for opening-mode fractures at least, systematic
only slowly, compared to their opening rates, if at all (Colgan et al., size, spatial arrangement, and porosity patterns develop with depth
2016). The topographic gradients involved in glacial creep exert a strong (thermal exposure). We can categorize depth-dependent fracture pat­
control on fracture propagation and morphology which we likewise do terns, for a given rock type, as fracture facies. Shallow (cool) environ­
not address here. But with continued climatic warming, in which ments with minimal effects of synkinematic cementation or chemically
deformation processes may accelerate and ice growth within fractures is assisted cracking produce widely spaced, barren or minimally cemented
expected to diminish, the process of ice crevasse formation may be ex­ joints (“exclusionary macrofracturing” facies). At intermediate depths,
pected to move progressively toward the lower left of Fig. 15. which, for a given rock type, geothermal gradient, and loading path,
The limitations of our modeling approach are many, and worth could range from 2 to 10 km, fractures that are moderately to exten­
reviewing, both for perspective about our interpretations and for po­ sively cemented during formation tend to have wide aperture size
tential future avenues of research. The model is 2-D, and so ignores ranges, wide variability in spatial arrangement tending toward clustered
pattern formation along fracture strike. The propagation of fractures patterns, and a considerable population of microfractures (“multiscale
along bedding likely produces considerably greater fracture interaction, fracturing” facies). Model results and limited observations suggest that
which has been shown to materially affect pattern outcomes (Olson and at high temperature patterns may develop having narrow aperture size
Pollard, 1989; Cladouhos and Marrett, 1996). ranges, decreased variability in spatial arrangement, and a preponder­
Cementation of springs proceeds at a constant rate within each ance of microfractures (“penetrative microfracturing” facies). Other
simulation. The assumption is that transport of mineralizing solutes is regions of model space have the potential for representing shallow
not rate-limiting, throughout the model. Instead, the precipitation step geothermal reservoir fracturing (fast cementation, resistant to subcrit­
is assumed to be rate-limiting (Walderhaug, 1996; Lander et al., 2008). ical failure) and cement-suppressed cases such as fracturing amid satu­
This case was called reaction-limited by Romano and Williams (2022). rated hydrocarbons and ice crevasse formation (slow cementation,
We base this assumption on textural evidence from synkinematically susceptible to subcritical failure).
cemented sandstones, in which cement accumulation rates vary
considerably by several aspects of the precipitation substrate. These Author statement
aspects include grain size, mineralogy, and whether the growth surface
is euhedrally or anhedrally terminated (Lander and Laubach, 2015). John Hooker: Conceptualization, Methodology, Software, Investi­
Including these various effects, as well as running transport-limited gation, Writing, Visualization, Funding acquisition Rich Katz: Method­
simulations, would allow us to simulate a wider range of possibilities, ology, Software, Formal analysis, Writing Steve Laubach:
and possibly yield insights into important phenomena such as Conceptualization, Investigation, Writing, Funding acquisition Joe
mineral-grain-scale effects on strain localization and porosity evolution Cartwright: Conceptualization, Resources Peter Eichhubl: Investigation,
(e.g., Vass et al., 2014; Liu et al., 2020; Monsees et al., 2021; Qin et al., Writing Esti Ukar: Investigation, Writing Dana Bloomfield: Investiga­
2022). tion, Writing Terry Engelder: Conceptualization, Resources.
Our treatment of subcritical failure is intended to address a prob­
lematic simplification presented by previous similar models (e.g., Declaration of competing interest
Hooker and Katz, 2015). Nevertheless, our subcritical failure term, like
our cementation rate term, remains a simplified representation of the The authors declare that they have no known competing financial
natural process. Laboratory tests have shown that subcritical crack interests or personal relationships that could have appeared to influence
propagation is sensitive to both the host rock (Holder et al., 2001) and the work reported in this paper.
the fluid chemistry (Callahan et al., 2019b; Chen et al., 2020). Our
model does not directly account for pore fluids, and so ignores any po­ Data availability
tential variability in space and time for susceptibility to corrosive re­
actions that lead to subcritical crack propagation (Atkinson, 1984). Data will be made available on request.
Furthermore, by ignoring fluids, we ignore any dynamic effects that
pore fluids might exert on the fracturing dynamics. Fluid pressures are Acknowledgments
thought to enable fracture opening at depth in general (Secor, 1965;
Bons et al., 2022). In this light, our simulations can be thought of as an JNH is supported by U.S. National Science Foundation Tectonics
end-member scenario in which high host-rock permeability suppresses program grant EAR-2214325. Aspects of this study were supported by
spatial variability in fluid pressure. If this were not the case, then cyclic grant DE-SC0022968 from Chemical Sciences, Geosciences and

18
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Biosciences Division, Office of Basic Energy Sciences, Office of Science, Engelder, T., Gross, M., 2018. Pancake joints in Utica gas shale: mechanisms for lifting
overburden during exhumation. J. Struct. Geol. 117, 241–250.
U.S. Department of Energy and by the Fracture Research and Applica­
Evans, M.A., 2010. Temporal and spatial changes in deformation conditions during the
tion Consortium at The University of Texas at Austin. formation of the Central Appalachian fold-and-thrust belt: evidence from joints, vein
mineral paragenesis, and fluid inclusions. In: Tollo, R.P., Bartholomew, M.J.,
References Hibbard, J.P., Karabinos, P.M. (Eds.), From Rodinia to Pangea: the Lithotectonic
Record of the Appalachian Region. Geological Society of America Memoir 206,
pp. 477–552.
Ankit, K., Urai, J.L., Nestler, B., 2015. Microstructural evolution in bitaxial crack-seal Evans, M.A., 2023. Syntectonic sediment loading and fold-thrust belt structural
veins: a phase-field study. J. Geophys. Res. Solid Earth. https://doi.org/10.1002/ architecture: An example from the central Appalachians (USA). Geosphere. https://
2015JB011934. doi.org/10.1130/GES02573.1.
Atkinson, B.K., 1984. Subcritical crack growth in geological materials. J. Geophys. Res. Evans, M.A., DeLisle, A., Leo, J., Lafonte, C.J., 2014. Deformation conditions for
89, 4077–4114. fracturing in the middle devonian sequence of the central appalachians during the
Bai, T., Pollard, D.D., 2001. Getting more for less: the unusual efficiency of fluid flow in late paleozoic alleghanian orogeny. AAPG (Am. Assoc. Pet. Geol.) Bull. 98 (11),
fractures. Geophys. Res. Lett. 28, 65–68. 2263–2299.
Becker, S.P., Eichhubl, P., Laubach, S.E., Reed, R.M., Lander, R.H., Bodnar, R.J., 2010. Faill, R.T., 1998. A geologic history of the north-central Appalachians, Part 3. The
A 48 m.y. history of fracture opening, temperature, and fluid pressure: Cretaceous Alleghanian orogeny. Am. J. Sci. 298, 131–179.
Travis Peak Formation, East Texas basin. GSA Bulletin 122 (7–8), 1081–1093. Fall, A., Eichhubl, P., Cumella, S.P., Bodnar, R.J., Laubach, S.E., Becker, S.P., 2012.
Berg, T.M., Edmunds, W.E., Geyer, A.R., Glover, A.D., Hoskins, D.M., MacLachlan, D.B., Testing the basin-centered gas accumulation model using fluid inclusion
Root, S.I., Sevon, W.D., Socolow, A.A., 1980. Geologic Map of Pennsylvania (1: observations: southern Piceance Basin, Colorado. AAPG (Am. Assoc. Pet. Geol.) Bull.
250,000 Scale). Pennsylvania Geological Survey. 96 (12), 2297–2318.
Bistacchi, A., Mittempergher, S., Martinelli, M., Storti, F., 2020. On a new robust Fall, A., Eichhubl, P., Bodnar, R.J., Laubach, S.E., Davis, J.S., 2015. Natural hydraulic
workflow for the statistical and spatial analysis of fracture data collected with fracturing of tight-gas sandstone reservoirs, Piceance Basin, Colorado. GSA Bulletin
scanlines (or the importance of stationarity). Solid Earth 11, 2535–2547. 127 (1/2), 61–75.
Bons, P.D., Elburg, M.A., Gomes-Rivas, E., 2012. A review of the formation of tectonic Ferrill, D.A., Smart, K.J., Cawood, A.J., Morris, A.P., 2021. The fold-thrust belt stress
veins and their microstructures. J. Struct. Geol. 43, 33–62. cycle: superposition of normal, strike-slip, and thrust faulting deformation regimes.
Bons, P.D., Cao, D., de Riese, T., González-Esvertit, E., Koehn, D., Naaman, I., Sachau, T., J. Struct. Geol. 148, 104362.
Tian, H., Gomez-Rivas, E., 2022. A review of natural hydrofractures in rocks. Geol. Fischer, M.P., Gross, M.R., Engelder, T., Greenfield, R.J., 1995. Finite-element analysis of
Mag. https://doi.org/10.1017/S0016756822001042. the stress distribution around a pressurized crack in a layered elastic medium:
Bruna, P.-O., Guglielmi, Y., Viseur, S., Lamarche, J., Bildstein, O., 2015. Coupling implications for the spacing of fluid-driven joints in bedded sedimentary rock.
fracture facies with in-situ permeability measurements to generate stochastic Tectonophysics 247, 49–64.
simulations of tight carbonate aquifer properties: example from the Lower Forstner, S.R., Laubach, S.E., 2022. Scale-dependent fracture networks. J. Struct. Geol.
Cretaceous aquifer, Northern Provence, SE France. J. Hydrol. 529, 737–753. 165, 104748.
Callahan, O.A., Eichhubl, P., Olson, J.E., Davatzes, N.C., 2019a. Fracture mechanical Gillespie, P.A., Howard, C.B., Walsh, J.J., Watterson, J., 1993. Measurement and
properties of damaged and hydrothermally altered rocks, Dixie Valley-Stillwater characterisation of spatial distributions of fractures. Tectonophysics 226, 113–141.
fault zone, Nevada, USA. J. Geophys. Res. Solid Earth 124, 4069–4090. Gillespie, P.A., Walsh, J.J., Watterson, J., Bonson, C.G., Manzocchi, T., 2001. Scaling
Callahan, O.A., Eichhubl, P., Olson, J.E., Davatzes, N.C., 2019b. Experimental relationships of joint and vein arrays from the Burren, Co. Clare, Ireland. J. Struct.
investigation of chemically aided fracture growth in silicified fault rocks. Geol. 23, 183–201.
Geothermics 83, 101724. Gillespie, P., 2003. Comment on “The geometric and statistical evolution of normal fault
Charlaftis, D., Jones, S.J., Dobson, K.J., Crouch, J., Acikalin, S., 2021. Experimental systems: an experimental study of the effects of mechanical layer thickness on
study of chlorite authigenesis and influence on porosity maintenance in sandstones. scaling laws” by R.V. Ackermann, R.W. Schlische and M.O. Withjack. J. Struct. Geol.
J. Sediment. Res. 91 (2), 197–212. 25, 819–822.
Chemenda, A.I., Lamarche, J., Matoniti, C., Bazalgette, L., Richard, P., 2021. Origin of Gomez-Rivas, E., Griera, A., 2012. Shear fractures in anisotropic ductile materials: an
strong nonlinear dependence of fracture (joint) spacing on bed thickness in layered experimental approach. J. Struct. Geol. 34, 61–76.
rocks: mechanical analysis and modeling. J. Geophys. Res. Solid Earth 126, Haluch, A., Rybak-Ostrowska, B., Wyglądała, M., Konon, A., Roszkowska-Remin, J.,
e2020JB020656. 2023. Interplay of organic matter, rock anisotropy, and horizontal shortening in bed-
Chen, X., Eichhubl, P., Olson, J.E., Dewers, T.A., 2020. Salinty, pH, and temperature parallel vein development within the lower Palaeozoic shale formations from the
controls on fracture mechanical properties of three shales and their implications for northern part of the Caledonian Foredeep basin (Poland). Mar. Petrol. Geol. https://
fracture growth in chemically active fluid environments. Geomechanics for Energy doi.org/10.1016/j.marpetgeo.2023.106387.
and the Environment 21, 100140. Heidbach, O., Rajabi, M., Cui, X., Fuchs, K., Müller, B., Reinecker, J., Reiter, K.,
Cladouhos, T.T., Marrett, R., 1996. Are fault growth and linkage models consistent with Tingay, M., Wenzel, F., Xie, F., Ziegler, M.O., Zoback, M.-L., Zoback, M.D., 2018. The
power-law distributions of fault lengths? J. Struct. Geol. 18 (2/3), 281–293. World Stress Map database release 2016: crustal stress pattern across scales.
Clark, M.B., Brantley, S.L., Fisher, D.M., 1995. Power-law vein-thickness distributions Tectonophysics 744, 484–498.
and positive feedback in vein growth. Geology 23 (11), 975–978. Hobbs, D.W., 1967. The formation of tension joints in sedimentary rocks: an explanation.
Colgan, W., Rajaram, H., Abdalati, W., McCutchan, C., Mottram, R., Moussavi, M.S., Geol. Mag. 104 (6), 550–556.
Grigsby, S., 2016. Glacier crevasses: observations, models, and mass balance Holder, J., Olson, J.E., Philip, Z., 2001. Experimental determination of subcritical crack
implications. Rev. Geophys. 54, 119–161. growth parameters in sedimentary rock. Geophys. Res. Lett. 28 (4), 599–602.
Corrêa, R.S.M., Marrett, R., Laubach, S.E., 2022. Analysis of spatial arrangement of Holford, S.P., Green, P.F., Hillis, R.R., Underhill, J.R., Stoker, M.S., Duddy, I.R., 2010.
fractures in two dimensions using point process statistics. J. Struct. Geol. 163, Multiple post-Caledonian exhumation episodes across northwest Scotland revealed
104726. by apatite fission track analysis. Journal of the Geological Society [London] 167,
Cumella, S.P., Scheevel, J., 2008. The influence of stratigraphy and rock mechanics on 675–694.
Mesaverde gas distribution, Piceance Basin, Colorado. In: Cumella, S.P., Shanley, K. Hooker, J.N., Gale, J.F.W., Gomez, L.A., Laubach, S.E., Marrett, R., Reed, R.M., 2009.
W., Camp, W.K. (Eds.), Understanding, Exploring, and Developing Tight-Gas Sands: Aperture-size scaling variations in a low-strain opening-mode fracture set, Cozzette
2005 Vail Hedberg Conference: American Association of Petroleum Geologists Sandstone, Colorado. J. Struct. Geol. 31, 707–718.
Hedberg Series, vol. 3, pp. 137–155. Hooker, J.N., Laubach, S.E., Gomez, L., Marrett, R., Eichhubl, P., Diaz-Tushman, K.,
Davis, G.H., 2014. Quasi-flexural folding of pseudo-bedding. GSA Bulletin 126 (5–6), Pinzon, E., 2011. Fracture size, frequency, and strain in the cambrian Eriboll
680–701. Formation sandstones, NW scotland. Scot. J. Geol. 47 (1), 45–56.
de Riese, T., Bons, P.D., Gomez-Rivas, E., Sachau, T., 2020. Interaction between crustal- Hooker, J.N., Gomez, L.A., Laubach, S.E., Gale, J.F.W., Marrett, R., 2012. Effects of
scale Darcy and hydrofracture fluid transport: a numerical study. Geofluids 2020. diagenesis (cement precipitation) during fracture opening on fracture aperture-size
https://doi.org/10.1155/2020/8891801. scaling in carbonate rocks. In: Garland, J., Neilson, J.E., Laubach, S.E., Whidden, K.J.
de Joussineau, G., Petit, J.-P., 2021. Mechanical insights into the development of fracture (Eds.), Advances in Carbonate Exploration and Reservoir Analysis, vol. 370.
corridors in layered rocks. J. Struct. Geol. 144, 104278. Geological Society [London] Special Publications, pp. 187–206.
Eichhubl, P., 2004. Growth of ductile opening-mode fractures in geomaterials. Hooker, J.N., Laubach, S.E., Marrett, R., 2013. Fracture-aperture size—frequency, spatial
Geological Society, London, Special Publications 231 (1), 11–24. distribution, and growth processes in strata-bounded and non-strata-bounded
Elmore, R.D., Burr, R., Engel, M., Parnell, J., 2010. Paleomagnetic dating of fracturing fractures, Cambrian Mesón Group, NW Argentina. J. Struct. Geol. 54, 54–71.
using breccia veins in Durness group carbonates, NW Scotland. J. Struct. Geol. 32, Hooker, J.N., Laubach, S.E., Marrett, R., 2014. A universal power-law scaling exponent
1933–1942. for fracture apertures in sandstones. Soc. Am. Archaeol. Bull. 126 (9/10),
Engelder, T., 1979. Mechanisms for strain within the upper devonian clastic sequence of 1340–1362.
the Appalachian plateau, western New York. Am. J. Sci. 279, 527–542. Hooker, J.N., Katz, R.F., 2015. Vein spacing in extending, layered rock: the effect of
Engelder, T., Haith, B.F., Younes, A., 2001. Horizontal slip along Alleghanian joints of synkinematic cementation. Am. J. Sci. 315, 557–588.
the Appalachian plateau: evidence that mild penetrative strain does little to change Hooker, J.N., Cartwright, J., Stephenson, B., Silver, C.R.P., Dickson, A.J., Hsieh, Y.-T.,
the pristine appearance of early joints. Tectonophysics 336, 31–41. 2017a. Fluid evolution in fracturing black shales, Appalachian Basin. AAPG (Am.
Engelder, T., Whitaker, A., 2006. Early jointing in coal and black shale: evidence for and Assoc. Pet. Geol.) Bull. 101 (8), 1203–1238.
Appalachian-wide stress field as a prelude to the Alleghanian orogeny. Geology 34, Hooker, J.N., Huggett, J.M., Cartwright, J., Ali Hussein, M., 2017b. Regional-scale
581–584. development of opening-mode calcite veins due to silica diagenesis. G-cubed.
https://doi.org/10.1002/2017GC006888.

19
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Hooker, J.N., Laubach, S.E., Marrett, R., 2018. Microfracture spacing distributions and Liu, Z., Zhang, L., Liu, G.R., Li, W., Li, S., Wang, F., Ma, Y., Li, H., Zhang, H., Han, L.,
the evolution of fracture patterns in sandstones. J. Struct. Geol. 108, 66–79. 2020. Material constituents and mechanical properties and macro-micro-failure
Hooker, J.N., Ruhl, M., Dickson, A.J., Hansen, L.N., Idiz, E., Hesselbo, S.P., modes of tight gas reservoirs. Energy Explor. Exploit. 38 (6), 2631–2648.
Cartwright, J., 2020. Shale anisotropy and natural hydraulic fracture propagation: Lorenz, J.C., Finley, S.J., 1991. Regional fractures II: fracturing of Mesaverde reservoirs
an example from the Jurassic (Toarcian) Posidonienschiefer, Germany. J. Geophys. in the Piceance basin, Colorado. AAPG (Am. Assoc. Pet. Geol.) Bull. 75 (11),
Res. Solid Earth. https://doi.org/10.1029/2019JB018442. 1738–1757.
Hooker, J.N., Fisher, D.M., 2021. How cementation and fluid flow influence slip behavior Major, J.R., Eichhubl, P., Dewers, T.A., Olson, J.E., 2018. Effect of CO2-brine-rock
at the subduction interface. Geology 49 (9), 1074–1078. interaction on fracture mechanical properties of CO2 reservoirs and seals. Earth
Hooker, J.N., Marrett, R., Wang, Q., 2023. Rigorizing the use of the coefficient of Planet Sci. Lett. 499, 37–47.
variation to diagnose fracture periodicity and clustering. J. Struct. Geol. 168, Manzocchi, T., 2002. The connectivity of two-dimensional networks of spatially
104830. correlated fractures. Water Resoures Research 38 (9). https://doi.org/10.1029/
Hoyt, E.M., Hooker, J.N., 2021. Silica diagenesis and natural facturing in limestone: an 2000WR000180.
example from the Ordovician of Central Pennsylvania. Mar. Petrol. Geol. 132, Marrett, R., Ortega, O.J., Kelsey, C.M., 1999. Extent of power-law scaling for natural
105240. fractures in rock. Geology 27 (9), 799–802.
Hudleston, P.J., 2015. Structures and fabrics in glacial ice: a review. J. Struct. Geol. 81, Marrett, R., Gale, J.F.W., Gómez, L.A., Laubach, S.E., 2018. Correlation analysis of
1–27. fracture arrangement in space. J. Struct. Geol. 108, 16–33.
Ji, S., Suwatari, K., 1998. A revised model for the relationship between joint spacing and Meredith, P.G., Atkinson, B.K., 1985. Fracture toughness and subcritical crack growth
layer thickness. J. Struct. Geol. 20 (11), 1495–1508. during high-temperature tensile deformation of Westerly granite and Black gabbro.
Ji, S., Li, L., Marcotte, D., 2021. Power-law relationship between joint spacing and bed Phys. Earth Planet. In. 39 (1), 33–51.
thickness in sedimentary rocks and implications for layered rock mechanics. Monsees, A.C., Blebricher, S.F., Busch, B., Feinendegen, M., Ziegler, M., Hilgers, C., 2021.
J. Struct. Geol. 150, 104413. Coupling diagenetic alterations and mechanical properties of Lower Permian
Kauffman, E.G., 1984. Paleobiogeography and evolutionary response dynamic in the siliciclastic sandstones: a pilot study. Environ. Earth Sci. 80, 141.
cretaceous western interior Seaway of North America. In: Westermann, G.E.G. (Ed.), Mount, V.S., 2014. Structural style of the Appalachian Plateau fold belt, north-central
Jurassic-Cretaceous Biochronology and Paleogeography of North America: Pennsylvania. J. Struct. Geol. 69, 284–303.
Geological Association of Canada Special Paper, vol. 27, pp. 273–306. Myers, R., Aydin, A., 2004. The evolution of faults formed by shearing across joint zones
Koehn, D., Sachau, T., 2014. Two-dimensional numerical modeling of fracturing and in sandstone. J. Struct. Geol. 26, 947–966.
shear band development in glacier fronts. J. Struct. Geol. 61, 133–142. Nuccio, V.F., Roberts, L.N.R., 2003. Thermal maturity and oil and gas generation history
Krabbendam, M., Glasser, N.F., 2011. Glacial erosion and bedrock properties in NW of petroleum systems in the Uinta-Piceance province, Utah and Colorado, Chapter 4.
Scotland: abrasion and plucking, hardness and joint spacing. Geomorphology 130, In: Petroleum Systems and Geologic Assessment of Oil and Gas in the Uinta-Piceance
274–383. Province, Utah and Colorado: U.S. Geological Survey Digital Data Series DDS-69-B
Lacazette, A., Engelder, T., 1992. Fluid-driven cyclic propagation of a joint in the Ithaca CD-ROM. USGS Uinta-Piceance Assessment Team, p. 39.
Siltstone, Appalachian basin, New York. In: Evans, B., Wong, T.-F. (Eds.), Fault Olson, J., Pollard, D.D., 1989. Inferring paleostresses from natural fracture patterns: a
Mechanics and Transport Properties of Rocks (U.S. Edition. Academic Press, San new method. Geology 17, 345–348.
Diego: California, pp. 297–323. Olson, J.E., 1993. Joint pattern development: effects of subcritical crack growth and
Ladeira, F.L., Price, N.J., 1981. Relationship between fracture spacing and bed thickness. mechanical crack interaction. J. Geophys. Res. 98 (B7), 12251–12265.
J. Struct. Geol. 3 (2), 179–183. Olson, J.E., 2003. Sublinear scaling of fracture aperture versus length: an exception or
Lander, R.H., Larese, R.E., Bonnell, L.M., 2008. Toward more accurate quartz cement the rule? J. Geophys. Res. 108 (B9), 2413. https://doi.org/10.1029/2001JB000419.
models: the importance of euhedral versus noneuhedral growth rates. AAPG (Am. Olson, J.E., 2004. Predicting fracture swarms – the influence of subcritical crack growth
Assoc. Pet. Geol.) Bull. 92 (11), 1537–1563. and the crack-tip process zone on joint spacing in rock. In: Cosgrove, J.W.,
Lander, R.H., Laubach, S.E., 2015. Insights into rates of fracture growth and sealing from Engelder, T. (Eds.), The Initiation, Propagation, and Arrest of Joints and Other
a model for quartz cementation in fractured sandstones. GSA Bulletin 127 (3–4), Fractures, vol. 231. Geological Society of London Special Publication, pp. 73–87.
516–538. Olson, J.E., Laubach, S.E., Lander, R.H., 2009. Natural fracture characterization in tight
Laubach, S.E., 1988. Subsurface fractures and their relationship to stress history in East gas sandstones: integrating mechanics and diagenesis. AAPG (Am. Assoc. Pet. Geol.)
Texas Basin sandstone. Tectonophysics 156 (1–2), 37–49. Bull. 93 (11), 1535–1549.
Laubach, S.E., 1989. Paleostress directions from the preferred orientation of closed Onasch, C.M., 1990. Microfractures and their role in deformation of a quartz arenite from
microfractures (fluid-inclusion planes) in sandstone, East Texas basin, USA. J. Struct. the central Appalachian foreland. J. Struct. Geol. 12 (7), 883–894.
Geol. 11 (5), 603–611. Ortega, O.J., Gale, J.F.W., Marrett, R.A., 2010. Quantifying diagenetic and stratigraphic
Laubach, S.E., Reed, R.M., Olson, J.E., Lander, R.H., Bonnell, L.M., 2004a. Coevolution of controls on fracture intensity in platform carbonates. J. Struct. Geol. 32, 1943–1959.
crack-seal texture and fracture porosity in sedimentary rocks: cathodoluminescence Peach, B.N., Horne, J., Gunn, W., Clough, C.T., Hinxman, L.W., Teall, J.J., 1907. The
observation of regional fractures. J. Struct. Geol. 26, 967–982. Geological Structure of the Northwest Highlands of Scotland. Memoirs of the
Laubach, S., Lander, R., Bonnell, L., Olson, J., Reed, R., 2004b. Opening histories of Geological Survey, U.K.
fractures in sandstone. In: Cosgrove, J.W., Engelder, T. (Eds.), The Initiation, Peacock, D.C.P., Sanderson, D.J., 1992. Effects of layering and anisotropy on fault
Propagation, and Arrest of Joints and Other Fractures, vol. 231. Geological Society of geometry. Journal of the Geological Society of London 149, 793–802.
London Special Publication, pp. 1–9. https://doi.org/10.1144/GSL. Petit, J.-P., Chemenda, A.I., Minisini, D., Richard, P., Bergman, S.C., Gross, M., 2022.
SP.2004.231.01.01. When do fractures initiate during the geological history of a sedimentary basin? Test
Laubach, S.E., Diaz-Tushman, K., 2009. Laurentian paleostress trajectories and case of a loading-fracturing path methodology. J. Struct. Geol. https://doi.org/
ephemeral fracture permeability, cambrian Eriboll Formation sandstones west of the 10.1016/j.jsg.2022.104683.
moine thrust zone, NW scotland. Journal of the Geological Society [London] 166, Pollard, D.D., Aydin, A., 1988. Progress in understanding jointing over the past century.
349–362. GSA Bulletin 100, 1181–1204.
Laubach, S.E., Olson, J.E., Gross, M.R., 2009. Mechanical and fracture stratigraphy. Price, N.J., 1966. Fault and Joint Development in Brittle and Semi-brittle Rock.
AAPG (Am. Assoc. Pet. Geol.) Bull. 93 (11), 1413–1426. Pergamon Press, Oxford, p. 176.
Laubach, S.E., Eichhubl, P., Hargrove, P., Ellis, M.A., Hooker, J.N., 2014. Fault core and Qin, S., Wang, R., Shi, W., Liu, K., Zhang, W., Xu, X., Qi, R., Yi, Z., 2022. Diverse effects
damage zone fracture attributes vary along strike owing to interaction of fracture of intragranular fractures on reservoir properties, diagenesis, and gas migration:
growth, quartz accumulation, and differing sandstone composition. J. Struct. Geol. insight from Permian tight sandstone in the Hangjinqi area, north Ordos Basin. Mar.
68, 207–226. Petrol. Geol. 137, 105526.
Laubach, S.E., Lander, R.H., Criscenti, L.J., Anovitz, L.M., Urai, J.L., Pollyea, R.M., Ramsay, J.G., 1980. The crack-seal mechanism of rock deformation. Nature 284,
Hooker, J.N., Narr, W., Evans, M.A., Kerisit, S.N., Olson, J.E., Dewers, T., Fisher, D. 135–139.
M., Bodnar, R.J., Evans, B., Dove, P., Bonnell, L.M., Marder, M.P., Pyrak-Nolte, L., Rinehart, A.J., Dewers, T., Broome, S.T., Eichhubl, P., 2016. Effects of CO2 on
2019. The role of chemistry in fracture pattern development and opportunities to mechanical variability and constitutive behavior of the lower tuscaloosa formation,
advance interpretations of geological materials. Rev. Geophys. 57 (3), 1065–1111. cranfield injection site, USA. Int. J. Greenh. Gas Control 53, 305–318.
Laubach, S.E., Zeng, L., Hooker, J.N., Wang, Q., Zhang, R., and Ren, B., submitted. Deep Romano, C.R., Williams, R.T., 2022. Evolution of fault-zone hydromechanical properties
and ultra-deep basin brittle deformation with focus on China. J. Struct. Geol.. in response to different cementation processes. Lithosphere 2022, 1069843.
Lee, H.P., Olson, J.E., Holder, J., Gale, J.F.W., Myers, R.D., 2014. The interaction of Saïag, J., Brigaud, B., Portier, É., Desaubliaux, G., Bucherie, A., Miska, S., Pagel, M.,
propagating opening mode fractures with preexisting discontinuities in shale. 2016. Sedimentological control on the diagenesis and reservoir quality of tidal
J. Geophys. Res. Solid Earth 120, 169–181. https://doi.org/10.1002/ sandstones of the upper cape hay formation (permian, Bonaparte basin, Australia).
2014JB011358. Mar. Petrol. Geol. 77, 597–624.
Li, B., Robinson, L., 2018. Natural Fractures, Fracture Facies, and Their Applications in Sanderson, D.J., Peacock, D.C.P., Nixon, C., Rotevatn, A., 2018. Graph theory and the
the Well Completion - Case Studies from the Permian Wolfcamp Formation, Midland analysis of fracture networks. J. Struct. Geol. https://doi.org/10.1016/j.
Basin. SPE/AAPG/SEG Unconventional Resources Technology Conference, West jsg.2018.04.011.
Texas, USA. https://doi.org/10.15530/URTEC-2018-2902102. Houston, Texas, Savalli, L., Engelder, T., 2005. Mechanisms controlling rupture shape during subcritical
USA, paper URTEC-2902102-MS. growth of joints in layered rocks. GSA Bulletin 117 (3/4), 436–449. https://doi.org/
Li, J.Z., Laubach, S.E., Gale, J.F.W., Marrett, R., 2018. Quantifying opening-mode 10.1130/B25368.1.
fracture spatial organization in horizontal wellbore image logs, core and outcrop: Schöpfer, M.P.J., Arslan, A., Walsh, J.J., Childs, C., 2011. Reconciliation of contrasting
application to Upper Cretaceous Frontier Formation tight gas sandstones, USA. theories for fracture spacing in layered rocks. J. Struct. Geol. 33, 551–565.
J. Struct. Geol. 108, 137–156. Secor Jr., D.T., 1965. Role of fluid pressure in jointing. Am. J. Sci. 263, 633–646.

20
J.N. Hooker et al. Journal of Structural Geology 173 (2023) 104915

Späth, M., Urai, J.L., Nestler, B., 2022. Incomplete crack sealing causes localization of Virgo, S., Abe, S., Urai, J.L., 2014. The evolution of crack seal vein and fracture networks
fracturing in hydrothermal quartz veins. Geophys. Res. Lett. 49, e2022GL098643. in an evolving stress field: insights from discrete element models of fracture sealing.
Srivastava, D.C., Engelder, T., 1990. Crack-propagation sequence and pore-fluid J. Geophys. Res. Solid Earth 119, 8708–8727.
conditions during fault-bend folding in the Appalachian Valley and Ridge, central Walderhaug, O., 1996. Kinetic modeling of quartz cementation and porosity loss in
Pennsylvania. GSA Bulletin 102, 116–128. deeply buried sandstone reservoirs. AAPG (Am. Assoc. Pet. Geol.) Bull. 80 (5),
Tan, T., Johnston, T., Engelder, T., 2014. The concept of joint saturation and its 731–745.
application. AAPG (Am. Assoc. Pet. Geol.) Bull. 98 (11), 2347–2364. Wilkins, S., Mount, V., Mahon, K., Perry, A., Koenig, J., 2014. Characterization and
Trewin, N.H., Rollin, K.E., 2002. Geologic history and structure of Scotland. In: development of subsurface factures observed in the Marcellus Formation,
Trewin, N.H. (Ed.), The Geology of Scotland. The Geological Society, London, Appalachian Plateau, north-central Pennsylvania. AAPG (Am. Assoc. Pet. Geol.) Bull.
pp. 1–25. https://doi.org/10.1144/GOS4P. 98 (11), 2301–2345.
Urai, J.L., Williams, P.F., van Roermund, H.L.M., 1991. Kinematics of crystal growth in Wilson, R.W., Holdsworth, R.E., Wild, L.E., McCaffrey, K.J.W., England, R.W., Imber, J.,
syntectonic fibrous veins. J. Struct. Geol. 13 (7), 823–836. Strachan, R.A., 2010. Basement-influenced rifting and basin development: a
Vass, A., Koehn, D., Toussaint, R., Ghani, I., Piazolo, S., 2014. The importance of reappraisal of post-Caledonian faulting patterns from the North Coast Transfer Zone,
fracture-healing on the deformation of fluid-filled layered systems. J. Struct. Geol. Scotland. In: Law, R.D., Butler, R.W.H., Holdsworth, R.E., Krabbendam, M.,
67, 94–106. Strachan, R.A. (Eds.), Continental Tectonics and Mountain Building: the Legacy of
Peach and Horne, vol. 335. Geological Society (London) Special Publications,
pp. 795–826.

21