T42C-01
Power-law Scaling of Fracture Aperture Sizes in Otherwise-Undeformed Foreland Basin Sandstone: An Example From the Cozzette Sandstone, Piceance Basin, Colorado
Power-law variation of aperture size with cumulative frequency has been documented in vein arrays, but such patterns have not been conclusively demonstrated from open or incompletely mineralized opening-mode fractures (joints) in otherwise-undeformed sedimentary rocks. We used subhorizontal core from the nearly flat- lying Cretaceous Cozzette Sandstone, Piceance Basin, Colorado, to document fracture aperture sizes over five orders of magnitude. We measured microfractures (0.0004-0.1164 mm in aperture) along a 276-mm-long scanline using scanning electron microscope-based cathodoluminescence; we measured macrofractures (0.5- 2.15 mm in aperture) in 35 m of approximately horizontal core cut normal to fracture strike. Microfractures are typically filled with quartz. Macrofractures are mostly open and resemble non-mineralized joints, except for thin veneers of quartz cement lining their walls. Micro- and macrofractures share both a common orientation and the same timing with respect to diagenetic sequence, only differing in size and the degree to which they are filled with quartz cement. Power-law scaling equations were derived by fitting trendlines to aperture vs. cumulative frequency data for the microfractures. These equations successfully predicted the cumulative frequencies of the macrofractures, accurate to within a factor of four in each test and within a factor of two in 75 percent of tests. Our results show that tectonic deformation is not prerequisite for power-law scaling of fractures, but instead suggest that scaling emerges from fracture interaction during propagation.
T42C-02
Finite Element Modeling of Fracture Reactivation and Bedding Slip During Folding
We use mechanical models to study the reactivation (opening and shearing) of pre-existing fractures and the evolution of slip on bedding surfaces during the formation of an asymmetric anticline using finite elements and large deformation contact mechanics. The models consist of three layers in frictional contact and the middle layer contains layer-perpendicular fractures. A lower boundary layer is in frictionless contact and serves to transmit displacements that mimic the fold profile at Sheep Mountain anticline, Wyoming, a thrust fault related fold. The upper boundary is subject to normal tractions appropriate for the depth of burial. Lateral boundaries are displaced horizontally to represent the Laramide tectonic shortening. The numerical simulations show the effects of material properties, slip on bedding surfaces, and the ratio of tectonic shortening to vertical uplift on the response of the multilayer system. The model uses frictional contact elements to capture opening and slip of pre- existing fractures and slip on bedding surfaces. The computational results demonstrate the sequence, kinematics, and spatial variations of reactivation, and mode of deformation (opening versus shearing) of bed- perpendicular fractures. We show that fractures located on the hinge are reactivated as joints and that those on the forelimb are predominantly reactivated as thrust faults. We compare the numerical results with fracture data observed at Sheep Mountain Anticline and discuss the similarities and differences between the field observation and the model results. Supported by U.S. Department of Energy, Grant No. DE-FG02-03ER15454, and U.S. National Science Foundation, Grant No. CMG-0417521.
T42C-03
The Effects of Carbon Films Deposited on New Fracture Surfaces on Rock Strength and Electrical Conductivity
Hollow cylinders of Sioux quartzite, jacketed by silver, were hydrostatically loaded to failure at temperatures up to 400 °C by applying pressurized Ar gas at the outer diameter (reaching ~290 MPa at a rate of 0.1 MPa/s) while maintaining a constant pore pressure at the inner diameter. Pore fluids consisted of CO, CO2, CH4, a 1:1 mixture of CO2 and CH4 (each with pore pressures of 2.0 to 4.1 MPa), and air (at atmospheric pressure). Biaxial-stress states were calculated using elastic-stress solutions that account for the applied pressures and hollow-cylinder dimensions. For the inner wall of the cylinders, effective radial stress was zero and calculated effective differential stress reached 1225 MPa. Failure occurred by the formation of mode II shear fractures that transected the hollow cylinder walls. The distribution of carbon in the run products was mapped by scanning electron microscopy and electron probe. Samples deformed in CO2 and air contained little or no carbon above the small amount that exists in the undeformed rock. Samples deformed in CO contain ubiquitous carbon films on the fracture surfaces that formed during deformation. Because carbon is absent on other free quartz surfaces present in the experiments, we conclude that the carbon films formed preferentially on the fractures as they formed. The radial resistivity of dry, undeformed Sioux quartzite cylinders is extremely large in the ambient laboratory atmosphere (>23 MØmega- m). The radial resistivity of Sioux quartzite cylinders that failed in pore fluids that promote carbon deposition is relatively low (2.9 to 3.1 MØmega-m for CO tests; 15.2 to 16.5 MØmega-m for CO2:CH4 tests). The results of this study help to isolate the role of carbon deposition on fresh fracture surfaces in altering the electrical properties of rocks with little initial porosity from that of carbon deposition on fractures and preexisting equant voids of porous rocks. Taken together, our results and those of Roberts et al. (1999) indicate that electrical conductivity in rocks may be enhanced due to carbon deposition on grain and/or fracture surfaces. Our results are important for studies linking variations of crustal electrical properties to seismogenesis, as well as to illuminate physico-chemical mechanisms that may be exploited to monitor injection sites for carbon sequestration. No decrease in rock strength was observed owing to weakening caused by the presence of these carbonaceous pore fluids. This work was supported by the National Science Foundation and the Department of Energy. Work performed under the auspices of the U.S. Department of Energy by the University of California Lawrence Livermore National Laboratory was carried out under contract W-7405-ENG-48 and supported specifically by funding from the Office of Basic Energy Sciences.
T42C-04
Formation of Network Fractures During Hydraulic Fracturing of the Barnett Shale, a Tight Gas Shale with Preexisting Fractures
Hydraulic fracturing operations generate new fractures as well as dilate preexisting fractures, creating networks of fractures. Here we model the complexity of the created network fractures and apply the results to wellbore log data and hydrofracture operations in the Barnett Shale, a tight gas-shale that requires artificial fracture stimulation to produce. It is shown that the resulting fracture geometry is related to the state of local stress, orientations of the existing fractures, and qualities of the hydraulic fracturing operation. The model assumes that preexisting fractures dilate when the hydrofracture pressure is larger than the normal stress across them. The orientations of the dilating fractures can be presented as 2θw and 2θL (width and length of the dilating fracture population on a stereographic projection). The model indicates that sin2θw /sin2θL = (σ2 - σ3)/(σ1 - σ3) and sin2θw = (Pm - σ3)/(σ1 - σ3), where σ1, σ2, and σ3, are the principal stresses of the local stress field and Pm is the hydrofracture fluid pressure. We expect the shape of the stimulated rock volume to vary (e.g., spheroidal, elliptical, penny- shaped) under different local conditions corresponding to 2θw and 2θL. We apply the above relations to hydrofracturing data from the Barnett Shale, Fort Worth Basin, Texas. We combined the pressure data of the hydrofracture operations with pre- and post-treatment fracture orientations determined on down-hole image logs to characterize the local stress magnitude and orientation. It was found that in the studied wells: (1) σ1 =Sv, calculated from overburden at the depth of fluid penetration; (2) σ2 =SH from the inversion of borehole sonic data (run as a post-drilling logging tool) or from solving the stress state around a borehole; and (3) σ3 =Sh, calculated as the instantaneous shut-in pressure from pressure-time curves. The direction of SH corresponds to the orientation of drilling-induced fractures, interpreted from formation multi-imaging logs (FMI). We use the spatial distribution of micro-seismic events recorded during the hydrofracture operations as a proxy for the shape of the stimulated rock volume. We use these observed volumes to test our model predictions based on monitored hydrofracture fluid pressure and the calculated local stresses. It is shown that at sites where near-field stresses are highly anisotropic, SH ≫ Sh, stimulation is restricted to preferentially oriented fractures and the volumetric zone of stimulation is a prolate spheroid, elongated in the direction of the far-field maximum horizontal stress. Under conditions of low near-field stress anisotropy, SH=Sh, dilation of multiple fracture orientations is possible and the volumetric zone of stimulation is an oblate spheroid. Additionally, we show that when detailed fracture orientation data is available, the stimulation model can be used to back-calculate the stress state.
T42C-05 INVITED
Deformation of Sedimentary Rock Across the San Andreas Fault Zone: Mesoscale and Microscale Structures Displayed in Core From SAFOD
Sedimentary rocks captured in cores taken at the San Andreas Fault Observatory at Depth (SAFOD) provide an unparalleled sampling of deformation in the transition zone between creeping and locked segments of a major transform fault at 2.5-3.1 km vertical depth. These samples provide the unique opportunity to study deformation processes and the development of brittle structures within porous and granular rocks that have been subjected to variable loading rates and chemically reactive fluids while residing at the top of the seismogenic zone. The samples provide a transect from relatively undeformed host rock through highly fractured and sheared rock, and capture the two prominent zones of active, aseismic slip. Core recovery was almost complete. Wrap-around 1:1 map tracings of the outer surfaces of all cores characterize the lithology and mesoscale deformation. Cores from 3056-3067 m and 3141-3153 m measured depth (MD) sample moderately deformed rock at the western boundary of the fault zone. The cores display massive to finely laminated, pebbly arkosic sandstones with lesser amounts of fine-grained sandstone and siltstone. Numerous shear fractures and cm-thick cataclastic shear zones form a conjugate geometry indicating contraction at high angles to the San Andreas fault. Both intervals display minor faults that juxtapose different lithologies consistent with meters or greater of slip. Fracture density is variable but tends to increase with proximity to the minor faults. Cross-cutting relationships between shear fractures and cataclastic zones indicate a general progression from early faulting along thicker shear zones to later, more localized slip within shear zones and along fractures. Microstructures provide ample evidence for densification of the sandstones through grain-scale fracture and crushing, as well as fluid assisted processes of crack-sealing, dissolution-precipitation, and alteration-neocrystallization. Grain-scale features are consistent with these processes having operated concurrently and cyclically. The microstructure of wear product along some slip surfaces, and associated mobilization and injections of wear microbreccia, are consistent with dynamic slip and local pore fluid pressurization. Cores from the actively creeping segments at 3194 m and 3301 m MD display a variety of moderately to highly sheared rocks. Disrupted bedding and cataclastic foliations oriented subparallel to the macroscopic orientation of the San Andreas fault are common. Protoliths of the fault rocks include interbedded sandstone, siltstone, shale, and serpentinite. Mesoscale calcite veins are abundant locally, yet are rare in the other cores. Zones of foliated cataclasites record a distributed shear within meters-thick zones, but sharp contacts between different types of cataclasite are also consistent with extremely localized slip having occurred at various stages of deformation. Zones of incohesive cataclasite and scaly fabrics could correlate with current locations of active, aseismic creep.
T42C-06
Clay Neomineralization and its Role in a Weak San Andreas Fault: new Insights From the SAFOD Drillhole in Parkfield/California
Neomineralization in fault zones and its influence on deformation and shear strength has been suggested as a possible explanation for a weak fault behavior. However, studies often lack the localization and characterization of the minerals along the slip surfaces, which is crucial for understanding the role of mineralization in fault zone processes. In this study, a detailed examination of illite minerals has been conducted from three sedimentary core samples of the SAFOD drill hole (3066m, 3436m and 3992m), showing different degrees of shearing and deformation. Based on X-ray diffraction and electron microscopy (SEM, TEM), we measured the crystal structure and fabric, as well as the shape and size of illite, their chemical composition, water content and polytypes. All samples contain detrital quartz, feldspar, chlorite, muscovite and biotite, with authigenic illite, illite-smectite (I-S), chlorite-smectite and smectite. The I-S phase at 3066m contains ca 75% of illite with a long-range ordering (RˇÝ3), the I-S at 3436m consists of ca 70% illite with randomly ordered smectite layers (R0), and illite at 3992m reveal thicker crystals without distinct smectite interlayers. The chemical composition ranges from a muscovitic composition (~1.0pu) at 3992m to depleted interlayer values of 0.5pu at 3436m. However, all illite phases show mixtures of 2M1 and 1M polytypes, with the coarse-grained (~150nm thick) pseudohexagonal 2M1 phase dominating and the very fine grained (10-15nm thick) 1M illite occurring as fibrous or flaky pore-fillings. The fine grained mineral phases were likely formed during the circulation of aqueous fluids along permeable fractures and veins by dissolution-precipitation reactions, partly at the expense of the detrital packets. Based on CALCMIX modeling, the clays contain different amounts of interlayer water with the highest values recorded at 3436m. Also heat flow analyses indicate the highest volatiles content at 3436m (6.5% total volatiles with 2% interlayer water), and the lowest volatiles content (2.7% with 0% interlayer water) at 3992m. The timing of illite growth in the matrix is considered to be coeval or older than the growth reported along open fractures, which are notably smectite rich and are an indication of more extensive dissolution associated with enhanced fluid flow. The localization of clay indicates that these phyllosilicates may dominate the fault behavior at shallow depth, as it is the dominant phase in these fault rocks.
T42C-07
Hydro-Mechanical Processes in Faulting: an Experimental Analysis of the Variability of Dilatancy in Simulated Fault Gouge
Frictional hardening associated with the dilation of fault zones and subsequent depressurization of pore fluids levies an important control on the nucleation of earthquake slip in natural fault zones. The present work seeks to describe the dependency of dilatancy, described by Segall & Rice (1995), on key fault parameters. Velocity stepping tests were conducted in a double-direct shear configuration under true-triaxial loading conditions. Samples were jacketed and subjected to constant pore fluid pressure. Confining and pore pressures were maintained via high precision servo-controlled pressure intensifiers. Experiments described herein were run with constant pore pressure. Frictional contact area was 5 x 5 cm. Sliding velocity was systematically varied from a background loading rate of 1μm/s to speeds as high as 100 φm/s and concurrent dilation of the granular layer was measured to determine the dilatancy coefficient (ε = Δφ/Δln(v), where φ is porosity and v is shear velocity). The amount of dilation as a result of a change in shearing velocity was measured using both the physical dilation of the layer as measured by DCDT on the biaxial deformation apparatus, as well as by measuring the amount of fluid influx as delivered by the pore pressure intensifier. We report on the dependence of dilatancy on effective stress in the range 0.8 to 30 MPa, using pore pressures from 0.4 – 4 MPa. Each experiment, other than at 0.8 MPa, involved a high normal stress portion followed by a lower normal stress (e.g. 30 then 20 MPa, 15-10 MPa). Gouge layers were constructed using a precision leveling jig to an initial thickness of 4 mm prior to shear. Over this range of effective stresses ε increased from 5.8 x 10-5 at 0.8 MPa effective stress to 5.0 x 10-4 at 30 MPa. The effect of accumulated strain and gouge fabric development on ε was measured by repeated velocity steps between 1 μm/s and 10 μm/s over a displacement of ~18 mm at an effective stress of 15 MPa. Results indicate that the magnitude of dilation following a 10 fold increase in sliding velocity is insensitive to strain history. We investigated the hypothesis that ε is inversely proportional to gouge porosity (hence directly proportional to effective normal stress) by subjecting a sample to 40 MPa normal stress, so as to overcompact it, prior to shearing at 15 MPa normal stress. Results showed that the amount of dilation decreased as a function of increasing displacement after the drop in normal stress. After a slip of ~16 mm the measured dilation was indiscernible from the conventionally loaded sample. These results indicate that porosity of the gouge layer plays an important role in the magnitude of dilation and slip rate dependence of dilation. Our results show that the amount of dilation resulting from an upstep in shear velocity increases with increasing effective normal stress. Porosity of the gouge layer appears to be an important control on dilatancy as shown by increased dilation in samples that have been loaded in excess of the experimental effective stress. The insensitivity of dilation to increasing strain reveals that strain history has little influence on dilation. Further experiments will be conducted to examine the control that layer thickness imposes on dilation, the null hypothesis being that in cases of distributed shear, dilation scales with layer thickness, and in cases of localized shear the magnitude of dilation would be controlled by the shear band and so would not scale with layer thickness.
T42C-08
Frictional and Hydrologic Properties of Clay-Rich Fault Gouge
Understanding the frictional and hydrologic properties of fault gouge is crucial to understanding the generation and nature of earthquakes and the strength of crustal faults. Clay minerals are a major constituent of fault gouge and are of particular interest because they may exhibit exceptionally low friction and low permeability. Previous work has shown that different clay minerals exhibit differences in shear strength, thus it has been suggested that different clay minerals may also exhibit stability differences. We report on laboratory experiments examining the frictional and hydrologic properties of saturated fault gouges dominated by the minerals illite, chlorite, and Ca- montmorillonite. Gouge consists of naturally occurring illite shale and chlorite schist (grain size < 106 μm), and a synthetic mixture of 50% montmorillonite-50% silt-sized quartz. Experiments were conducted in a servo-controlled apparatus within a sealed pressure vessel using the double-direct shear configuration. Confining pressure was 40% of the applied normal stress, which varied between 15-100 MPa. Pre- and post shear permeability was measured by controlling pore pressure normal to both gouge layer contact areas. During shearing, pore pressure was controlled at 5 MPa on one end while changes in pore pressure were measured on the other end. Layers were initially 4 mm thick with nominal contact dimensions of 5 cm by 5 cm. Experiments were conducted under constant shear velocity boundary conditions of 1-300 μm/s. Preliminary results at effective normal stresses of 15-30 MPa indicate that montmorillonite gouge is slightly weaker (coefficient of friction = 0.19-0.23) than illite and chlorite gouge (0.29-0.32). Montmorillonite, chlorite, and illite gouges are consistently velocity strengthening, and (a-b) tends to increase with increased sliding velocity. Under a constant head gradient, layer-normal permeability decreases ~3 orders of magnitude in montmorillonite gouge (from 3x10-16 m2), ~2 in illite gouge (from 5x10-17 m2), and ~1 in chlorite gouge (from 4x10-16 m2) after 13-34 shear strain. Post-shear, chlorite gouge is the most permeable followed by illite and montmorillonite. In all gouges, permeability drops ~1 order of magnitude for each increase of 15-20 MPa of effective stress. Gouges with measured post shear permeabilities below 1x10-19 m2 develop significant overpressure (>550 kPa) after 10-20 shear strain. Ongoing experiments include permeability and friction measurement at higher effective stresses to further investigate frictional and hydrologic variations among different clay minerals and natural fault gouges.