Previous   |    Next

doi:10.2204/iodp.proc.314315316.214.2011

Methods and materials

Yue et al. (submitted) provide a more detailed description of the procedures we used for testing permeability and imaging grain fabric. All of the specimens consist of hemipelagic mud with varying degrees of consolidation. We tested eight samples from the accreted trench wedge facies and three samples from the upper Shikoku Basin hemipelagic facies (Fig. F2). The dip of bedding surfaces relative to the horizontal plane is based on nearby measurements on split core (see the “Expedition 316 Site C0006” [Expedition 316 Scientists, 2009b] and “Expedition 316 Site C0007” [Expedition 316 Scientists, 2009c] chapters).

Specimen preparation

The whole-round samples were capped and taped in their plastic core liners on board the D/V Chikyu, sealed with wet sponges in aluminum vacuum bags to prevent moisture loss, and stored at 4°C until immediately prior to trimming. To extract each specimen, the plastic core liner was cut lengthwise along two lines 180° apart using a hacksaw. The core liner was then removed to extrude the sample. Cylindrical specimens for permeability tests in the vertical (along-core) direction were trimmed using a wire saw and soil lathe. Specimen length after trimming was ~4.4–5.9 cm and averaged 5.3 cm (Table T1). Specimen diameter was ~3.4–4.1 cm and averaged 3.8 cm. These dimensions were measured at several points using a caliper to a resolution of 0.03 cm and averaged to obtain the values used for subsequent calculations. Specimens for tests in the horizontal (cross-core) direction were trimmed from material immediately below the specimen for vertical (along-core) testing. These specimens were trimmed perpendicular to the core axis. Bedding dip with respect to the core axis was not taken into account during trimming.

Initial porosity was calculated from gravimetric water content of the specimen trimmings by assuming 100% pore water saturation. Values of specific gravity of the mineral solids have been imported from shipboard measurements of the closest adjacent specimen (see the “Expedition 316 Site C0006” [Expedition 316 Scientists, 2009b] and “Expedition 316 Site C0007” [Expedition 316 Scientists, 2009c] chapters). Gravimetric water content of the specimen trimmings was determined by measuring the ratio of the mass of water to the mass of mineral solids determined by oven-drying the trimmings at 105°C until constant mass was reached (generally within 24 h) in accordance with shipboard measurement protocols (see the “Expedition 316 methods” chapter [Expedition 316 Scientists, 2009a]). A correction for salt content as a percentage of the total dry weight was applied for all calculations of gravimetric water content and porosity, using

where

• W_c = corrected dry weight,

• M_t = total mass of the saturated specimen,

• M_d = mass of the dried specimen, and

• r = salinity (per mil).

For salt corrections on pretest trimmings, we assumed an average interstitial salinity value of 35‰, and for post-test trimmings, we picked a value of 25‰ to match the concentration of the simulated seawater that was used to saturate specimens during the tests. Additional information in Table T1 includes Skempton’s B-value to assess specimen saturation after backpressuring (see subsequent discussion), shipboard values of porosity and water content, and the post-test values of water content and porosity calculated from oven-dried trimmings after the specimens had been consolidated to ~0.55 MPa effective stress.

Constant-flow apparatus

Constant-flow, flow-through permeability tests were used to determine hydraulic conductivity in the vertical and horizontal core directions. A withdrawal-infuse syringe pump (KDS Scientific, Model 260) was used to simultaneously inject and extract pore fluid from the top and bottom of the specimen. The system consists of an acrylic confining cell to contain the specimen and provide isotropic effective confining stress, a constant flow syringe pump, one differential pressure transducer to measure hydraulic head difference between the specimen top cap and bottom cap, and an air/water interface panel for regulating the confining fluid pressure and pore fluid backpressure (Fig. F3). Signals from the differential pressure transducer are acquired to obtain hydraulic head difference through the specimen at a precision of ±1 cm H₂O over a range spanning ±1000 cm H₂O. A digital interface is used for readout and storage of values of effective isotropic confining stress (σ′), hydraulic head difference (Δh), and time duration measurements made during each test run. The flow pump holds two syringes (Hamilton GasTight Series 1000) and has the capability to cycle continuously back and forth in a push-pull action. As one syringe is infusing pore fluid into the specimen the other withdraws an equal volume of fluid from the other end of the specimen at the same rate. At the end of the set volume the direction is automatically reversed and the next cycle begins. With the use of three-way valves, the pump can empty and refill syringes for a continuous dispense. Volumetric flow rate (Q) for the series of tests described here ranged from a minimum of 7.0 × 10^–5 cm³/min to a maximum of 8.0 × 10^–3 cm³/min.

Backpressure saturation

Prior to testing, all permeant lines and porous stones were saturated with simulated seawater (25 g NaCl to 1 L tap water). A specimen was placed on the pedestal, the top cap was applied, and a latex membrane was placed on the specimen using a vacuum membrane expander. The confining chamber was then sealed and the cell was filled with tap water. Saturation of the specimen was achieved by ramping pore fluid backpressure to 0.48 MPa (70 psi) using the panel board (air/water interface) while also ramping the confining pressure to maintain an effective isotropic confining stress of 0.034 MPa (5 psi). Elevated backpressure was maintained for at least 24 h. Saturation of the specimen was checked by increasing the confining pressure (σ) to 0.55 MPa (80 psi) and measuring the corresponding pore pressure (u) response, which yields Skempton’s B-value (B = Δu/Δσ).

Specimens were considered saturated if B ≥ 0.95 (Table T1). Once saturation was achieved, the cell pressure was increased to consolidate the specimen at an isotropic effective stress of 0.55 MPa. Pore water was allowed to drain during consolidation from both the top and bottom of the specimen by opening the top and bottom valves on the confining cell system. The volume of pore water expelled was measured using the backpressure pipette and monitored for equilibrium to calculate the corresponding volume change of the specimen.

Constant-flow permeation

Constant-flow tests were performed for each of the 21 specimens (11 trimmed vertically, or parallel to the core axis, and 10 trimmed horizontally, or perpendicular to the core axis) at 0.55 MPa (80 psi). Tests at this effective stress were run using four flow rates; two tests were conducted with a top-to-bottom flow direction (denoted subsequently as a negative flow value) and two tests were conducted with a bottom-to-top flow direction (denoted as a positive flow value) to obtain replicate permeability values (Fig. F4). Transient response from the differential pressure transducer was monitored for steady-state head difference (Δh_s). Values of applied discharge velocity (v) and steady-state hydraulic gradient (i_s) were plotted to assess consistency among the four test runs and linearity in their relation (see the “Appendix”). Coefficient of determination (R²) calculated by least-squared linear regression of these relations were not less than 0.9835, indicating good repeatability among the four flow tests conducted at each flow rate and the applicability of Darcy’s law (Equation 2) for calculating hydraulic conductivity.

Data analysis

Hydraulic conductivity (cm/s) was calculated for each specimen using Darcy’s law

where

• Q = applied volumetric flow rate (cm³/s),

• i_s = steady-state hydraulic gradient equal to the ratio of the steady-state head difference (Δh_s) to the length over which that head difference occurs (ΔL) (taken as the initial height of the specimen), and

• A = cross-sectional flow area (cm²; taken as the initial specimen area).

Corresponding discharge velocity is v = Q/A. Hydraulic conductivity values (K = m/s) were converted to intrinsic permeability (k = m²) values using

where

• µ = viscosity of permeant (0.001 Pa·s),

• ρ = density of permeant (1027 kg/m³), and

• g = gravitational acceleration (9.81 m/s²).

Grain fabric imaging

Specimens for grain fabric imaging were cut from the whole-round samples while trimming cylinders for the flow-through tests using a razor blade at vertical orientation and horizontal orientation relative to the axis of the cylindrical samples (Fig. F5). Grain fabric of wet, uncoated, and unfixed specimens was imaged using an FEI Quanta 600 FEG scanning electron microscope (SEM). The instrument operates in environmental mode (ESEM) at 30 kV, with the specimen chamber pressure set at 700 Pa. Water vapor (~98% humidity) from a built-in reservoir keeps the specimen from losing moisture. The temperature of the cooling stage was set to 2°C. The specimens were imaged with a gaseous backscattered electron detector, spot 3.0 at a working distance of ~10 mm. This combination generates an imaging resolution of ~4 nm, and the dimensions of the field of view are ~145 × 130 µm with 2000× magnification. Specimens were placed in the holder on the stage with the imaged surface facing upward. “Center stage” and “Tilt” commands of the ESEM controlling software were used to manually adjust the imaging face to an orientation as close to perpendicular as possible to the imaging beam. All the image files were saved with color gray mode in TIF format (Fig. F6A).

Digital images were processed using software known as ImageJ (available at rsbweb.nih.gov/ij/index.html). Our processing steps were the following:

1. Contrast enhancement by linear stretching of the gray-level histogram in order to use 256 gray level values;

2. Median filter by moving each pixel value to the median values of nine closest pixels (to reduce noise);

3. Mean filter by replacing each pixel with the neighborhood mean. The size of the neighborhood is specified by entering its radius in the dialog box (to preserve subtle details);

4. Median hybrid filter by moving each pixel to the median values of the middle horizontal 3 pixels, center vertical 3 pixels, and center pixel of those 9 closest pixels (to reduce noise while preserving linear features);

5. Threshold adjustment by picking up one point of gray-level histogram (to select objects);

6. Make binary to transform the gray image to white and black image (e.g., Fig. F6B);

7. Overlap the image onto the original image and set its alpha value (transparency) to 60% in CorelDraw 11 software (Fig. F6C) and then separate objects that touch by manual adjustment with eraser tool (Fig. F6D);

8. Median filter with ImageJ to remove objects <9 pixels in size (because measurements on small objects are mostly biased);

9. Fill the holes on the objects; and

10. Measure automatically to obtain the long-axis and short-axis dimensions and long-axis orientation of an object.

The software can automatically determine the long or short axis (apparent dimensions) of the objects in the two-dimensional image. The results are automatically saved in a text file after the measurement.

Characterization of microfabric anisotropy

Grain fabric was characterized statistically using rose diagrams to depict orientations of the apparent long particle axes. In petrography, SEM, and transmission electron microscopy studies, most investigators measure between 100 and 500 grains per thin or ultrathin section (Krumbein, 1935; Friedman, 1958; van der Plas, 1962; Griffiths, 1967; Chiou et al., 1991). We generally counted between 200 and 500 grains (Table T2). Each particle orientation (apparent long axis) was assigned to an angle between 0° and 180°. For the vertical section, the core axis is oriented at 90°. Rose diagrams were constructed using Rozeta software (www.softpedia.com/get/Science-CAD/Rozeta.shtml). This software automatically counts the number of particles according to their orientation and combines data into bins of 10° intervals. In addition to the rose diagram, the number of values in each bin was summed and normalized to 100%. Cumulative frequency curves of the normalized bin percentages were constructed to show the distribution of grain orientation and calculate graphical statistics (Chiou et al., 1991).

Various statistical methods can be used to characterize the degree of preferred grain orientation, such as the formulas of Folk and Ward (1957), Martínez-Nistal et al. (1999), and Zaniewski and van der Meer (2005). The Folk and Ward (1957) formula was developed originally to graphically compute values of sorting (standard deviation) for grain size data. The equivalent equation for standard deviation of grain orientation (d) is

where ϕ₈₄, ϕ₁₆, ϕ₉₅, and ϕ₅ represent the angle of orientation (in degrees) at the 84th, 16th, 95th, and 5th percentiles, respectively, on the cumulative frequency curve. This graphic technique avoids the laborious calculations required by moment statistics (Chiou et al., 1991). If the fabric of sediment shows strong preferred orientation, then the sorting of orientation angles will be more tightly clustered and the cumulative frequency curve will be steeper around the median. Numerically, the largest value of d is 72.3° (i.e., a case in which ϕ₁₆ and ϕ₅ = 0° and ϕ₈₄ and ϕ₉₅ = 180°). We normalized each standard deviation to this maximum d value by calculating the “index of microfabric orientation” (i) as

The closer the value of i is to 1, the more the particles are aligned in a preferred direction. For a highly random arrangement of particles, the cumulative curve generally has a slope <0.75 near the median, the standard deviation of orientation is >35°, and the index of orientation is <0.51. For well-oriented clay particles, the slope of the cumulative curve is generally >1.00 near the median, the standard deviation of grain orientation is <25°, and the index of microfabric orientation is >0.65 (Yue et al., submitted). To compare i values from imaging surfaces that were cut parallel and perpendicular to the core axis, we calculated the orientation index ratio (i_h/i_v).

Top of page   |    Previous   |    Next