The results of these tests on 12 serpentine mud samples are shown in Table T1, along with the corresponding depths and porosities. For comparison, the results of both methods of analysis, the square root of time model and the finite difference model, are presented. In general, the two methods give similar results, with the values of hydraulic conductivity estimated using the finite difference code being on average ~20% less than the values estimated using the square root of time analysis. There is less difference, on average, between the estimates of specific storage using the two methods.
Repeatability of the measurements was checked on two of the cores, Samples 195-1200D-1H-3, 120-130 cm, and 1H-5, 10-20 cm. Two cores were taken from both of these samples and tests were run on them (Table T1). In both tests, the difference between the two measurements was ~20% when using the results from the finite difference method and ~30% when using the square root of time method. Although a sample size of only two is too small to generalize, this result suggests that the finite difference method does a slightly better job of estimating the flow parameters.
Figure F5A shows hydraulic conductivity as a function of depth. At shallow depths (<5 m), the values are ~10-20 x 10-11 m/s and then decrease rapidly with depth to nearly constant values of ~4 x 10-11 m/s below a depth of 20 mbsf. The specific storage behaves in a similar manner, except that it decreases only by a factor of about one-half instead of by one-fourth as the depth increases. Porosity vs. depth for all the serpentine muds samples measured during Leg 195 (Fig. F5B) shows a similar trend of decreasing values with depth. The porosity values given in Table T1 are from samples located within depths of 1 m from the samples used for the hydraulic conductivity measurements, and so the variability of those measurements may mask the general trend shown in Figure F5B.
The dependence of the hydraulic conductivity and the specific storage on the axial load used during the test was also investigated. As can be seen in Table T1 for Sample 195-1200D-1H-4, 120-130 cm, a larger axial load resulted in decreased values of both the hydraulic conductivity and the specific storage. The larger load results in greater consolidation so that the flow paths through the material are restricted. This consolidation would also reduce the porosity so that the storage is also reduced. The hydraulic conductivity decreased more than the specific storage so that the hydraulic diffusivity, the ratio of the hydraulic conductivity to the specific storage, experienced a net decrease.
The average ratio of anisotropy (vertical/horizontal) of hydraulic conductivity was found to be 0.82 for the seven samples from which both horizontal and vertical cores were taken and tested. This anisotropy may be due to grain fabric aligned with the horizontal plane. This observation might warrant further investigation.
Several of the assumptions made in the above analysis may not be well met when using the Manheim squeezer. Chief among these is the assumption that there is no friction between the cylinder walls and both the piston and the sample sides as the sample is consolidated. This assumption would result in an underestimation of the vertical compressibility because the applied load used in the calculation would actually be reduced by friction. Using Equations 6 and 7, we can see that an underestimation of the vertical compressibility would result in underestimation of the specific storage and the hydraulic conductivity.
We assumed that the pore fluid pressure is atmospheric at the sample bottom as fluid moves into the syringe. It may be that the syringe produced some backpressure due to friction between the syringe plunger and the syringe wall. This backpressure could be measured to determine if it is significant relative to the pore pressure in the sample during the test. Although it would add to the complexity of the testing apparatus, use of displacement transducers to measure the axial displacement would remove the need for using fluid discharge as a proxy for axial displacement.
We attempted to determine whether the volume of the outlet channel might decrease the initial discharge rate; the outlet channel would need to fill before any fluid could be measured in the syringe. We did several runs with the outlet channel filled with deionized water and did not note any significant difference between runs where we did not fill the channel. It may be that the air in the channel was displaced into the syringe, and so if the friction between the syringe wall and piston was low, the air would displace the piston as easily as the water and there would be little error introduced. Even if the air did not displace the syringe piston, the volume of the outlet channel was ~0.5 mL and so would affect the total fluid discharged by only ~5%.
Another assumption is that the sample was fully saturated. We know this was not the case, as some gas bubbles (~1 mL, or 5%-10% of the total discharged volume) were discharged into the syringe during most of the tests. This gas came out of solution when the sample was brought from the ocean bottom to the surface. The sample should be resaturated so that this gas would not be present. The presence of gas would affect the results by increasing the vertical compressibility of the sample by introducing a highly compressible pore fluid, thus increasing the specific storage, and by introducing two-phase flow, reducing the hydraulic conductivity.
Last, we assumed that the hydraulic conductivity and the specific storage do not depend on the degree of consolidation. This assumption is common to all estimations of hydraulic conductivity and specific storage when using a consolidation test. That this assumption is not true is apparent in all the fits of the finite difference model to the measured data. At early times, up to 1000 s in Figure F4B, the finite difference model underestimates the fluid discharge and then later, after 5000 s, overestimates the fluid discharge. This pattern is common to all the finite difference model fits and cannot be removed by variation of the parameters without gross misfit to the data. This pattern can be explained if the hydraulic diffusivity decreases with consolidation. Less of the material is consolidated at early times and so fluid flows more quickly. Our measurements of hydraulic conductivity and specific storage for these samples represent weighted average values obtained over a very wide range of consolidation states. For Sample 195-1200E-5H-3, 51-61 cm, shown in Figure F4, the porosity of the sample decreased from 48% at the start of the test to 28% at the end of the test. It should be noted that this dependence is not apparent using the square root of time method of analysis.