INTERPRETATIONS AND DISCUSSION

Although only one core section was available for testing, a number of useful insights can be drawn concerning the mechanical behavior of this nannofossil clay, and some constraints can be placed on the in situ stress conditions at the sample depth. Some of these implications have significant bearing on the general application of laboratory measurements to the evaluation of the geological environment.

One fundamental observation from these consolidation tests is that the measured yield stress is very low for the depth from which the sample was collected, which implies a high pore-fluid pressure as well as a very low degree of cementation. The lack of any of the characteristics of cemented sediments in the response curves during these tests was further evidence of an uncemented in situ condition. The low yield stress and lack of cementation cannot have been caused by coring disturbance, in which case a higher porosity would have been preserved.

The consolidation test on the disaggregated equivalent lithology was designed to test this conclusion by generating consolidation parameters for an uncemented "reference" state. If the core sample were totally uncemented, it might be expected to have a porosity similar to the disaggregated equivalent at its yield stress and at higher values of _v'. Instead, the disaggregated equivalent had significantly higher porosities and compressibility than the core sample at stresses above yield (Fig. 7). Clearly the disaggregated equivalent cannot be used as an uncemented reference condition, at least before understanding the reasons for these differences. Therefore, before discussing any implications concerning the stress conditions and stress history at the sample depth in Hole 897D from test results on the core sample, the results of the reference consolidation test must first be considered.

A similar comparison of consolidation parameters can be generated from results of tests on a silty clay and its disaggregated equivalent from the Nankai Trough (Karig, 1993, and Karig, unpublished data). Samples of this silty clay from the trench fill (DSDP Site 582) were highly cemented, whereas other samples of this same lithology, from beneath the accretionary prism (ODP Site 808) were decemented (Karig, 1993). This array of data provides the basis for comparison between laboratory and field consolidation data as well as between the silty clay and nannofossil clay. Some of the consolidation tests were performed in a uniaxial cell, whereas others were done in the triaxial cell. Consolidation parameters for the disaggregated Nankai clay obtained from both cells were quite similar and demonstrated that the difference was not in the method of testing. In addition, replication of consolidation tests on the same core produced effectively identical results.

On the other hand, there are parallel differences between the consolidation parameters from tests on core samples and on their disaggregated equivalents for both a sample of the Nankai decemented silty clay and the calcareous clay of Hole 897D (Fig. 9). Both these natural samples have lower porosities than their disaggregated equivalents at the same consolidation stresses. In addition, both samples not only have lower compressibilities than their disaggregated equivalents, but their compressibilities, as measured in terms of A_c values, are quite similar (Fig. 9). In both comparisons the K_o values for the natural samples during post-yield consolidation were lower than for their disaggregated equivalents.

In striking contrast to the similarity of the behavior of the decemented Nankai clay to the clay from Site 897, a core sample of highly cemented silty clay from Site 582 had higher porosities at yield than its disaggregated equivalent (Fig. 9), but a similar stress ratio. The greater porosity of the cemented sediment was associated with its high yield stress, and attributed to the combined effects of cement and high in situ pore pressure (Karig, 1993). The high post-yield compressibility can be interpreted as reflecting the rapid breakdown of cement.

It is less obvious why there are such large differences in response between the uncemented natural samples and their disaggregated equivalents. Clearly the experimental consolidation of a clay rich sediment does not seem to reproduce the consolidation response of an uncemented natural sample. This could be the result of changes in physical/mechanical properties during the disaggregation process but more likely it reflects differences in secondary consolidation resulting from the very different rates of consolidation in the laboratory and in nature.

A crude idea of the total secondary consolidation that would occur over geological time can be obtained by analyzing the secondary consolidation that occurred during the 26-hr constant stress hold at the end of the consolidation test on the disaggregated sample. Secondary consolidation, as expressed in terms of vertical strain, is thought to be linear with respect to the log of time after cessation of primary consolidation: _v, = C log t, where C is a function of the sediment lithology and of the porosity at the end of primary consolidation (e.g., Wood, 1990). Although the secondary consolidation during test C-34 was of marginal duration, it generated a C of 0.004/cycle of log t, which leads to a porosity reduction of about 5% over geologic time for this sample at a _v' of 9.8 MPa. The porosity reduction of secondary consolidation is largest when a given sediment is at its normally consolidated state, and appears to decrease as the consolidation stress increases (e.g., Graham et al., 1983).

C was not obtained for the disaggregated material at 1 MPa, which is the apparent yield stress of the undisturbed sediment, but it should lead to a porosity reduction for secondary consolidation significantly higher than 5% and might explain the 9% lower porosity of the undisturbed sample at its yield stress (Fig. 9). The lower compressibility of undisturbed sample above yield can be explained as "delayed consolidation" (Bjerrum, 1967), which is a state of apparent overconsolidation generated by secondary consolidation (creep). Such a phase of delayed consolidation is seen following the stress hold during the consolidation of the disaggregated equivalent (Fig. 7). If laboratory consolidation of the core samples had been continued, the porosity-stress curve should have approached that of the disaggregated equivalent.

The effects of creep appear to explain the differences in response between the natural sample and its disaggregated equivalent. The combination of low _c' and the relatively low porosity for the mudstone from 619 mbsf in Hole 897D can reasonably be explained by very high pore-fluid pressure in an effectively uncemented sediment that has undergone extensive creep.

For normally consolidated sediments, P = _v – _c', where _v can be estimated from the integration of shipboard bulk densities over the depth range above the sample. Because porosity rebound is neglected in this approach, the calculated value of _v = 11 MPa is probably a minimum, but leads to P ~ 10 MPa and an overpressure of 5 MPa. The overpressure ratio, , is equal to P/_v, where the mass of overlying water is ignored (Davis et al., 1983), and is greater than 0.9, implying that P is nearly lithostatic.

The stress ratio at yield was about 0.85, but the validity of this value as a measure of in situ _h' is questionable. The first problem is whether or not the value of _h' and the stress ratio at yield during reconsolidation is the same as that at the maximum previous _v'. At the yield stresses measured during the reconsolidation segments of the multiple unload-reload test, _h' and K_o were quite similar to the values at the previous consolidation stresses, indicating that, at least for laboratory conditions, the "in situ" _h' can be accurately estimated.

The natural samples, however, have undergone more complete secondary consolidation as well as having been under stress for >10⁷yr, the effects of which are uncertain. Secondary consolidation did not seem to affect _h' over short periods, but extrapolation of this response to times many orders of magnitude longer is dangerous at best. The higher K_o ratios for the natural samples during delayed consolidation (0.61) than for their disaggregated equivalent (0.57) may reflect the effect of fabric modification. The higher stress ratio measured for the undisturbed samples at their yield condition (0.73), relative to that of either of the other two conditions, could be interpreted either as the effect of regional compression or as relaxation (e.g., Warpinski, 1989). Evidence for Tertiary regional compression was presented by the Shipboard Scientific Party (1994b).

A related problem is how best to apply laboratory-acquired mechanical parameters to in situ geological conditions. Clearly, laboratory-generated _v'- curves cannot be assumed to represent in situ conditions for either cemented or uncemented sediments. If total geological creep could be estimated as a function of _v', primary consolidation results might be "corrected," but the variable cementation of sediments makes this a very questionable procedure.

To the extent that deformation to yield can be considered elastic and accepting that creep effects during this deformation are small (e.g., Graham et al., 1983), especially if tests are run at very low strain rates, stresses and porosities at yield should be fairly representative of in situ conditions. Parameters acquired during post-yield deformation do not represent natural geological conditions, but are probably applicable to problems of settlement from fluid withdrawal or rapid loading.