v' or
a-v' curves, have been developed by geotechnical engineers (e.g.,
Casagrande, 1936; Janbu, 1985). Some of these methods generate a yield state that lies off the actual stress path of the test
(Fig. 2).
Consolidation tests can be performed either in uniaxial strain cells (odometers) or in triaxial cells. Tests in odometers are simpler, but in only few of these cells can horizontal stress be measured and most have the problem of sidewall friction. On the other hand, consolidation of remolded sediments and slurries are much more easily done in such cells. Tests in triaxial cells are more complicated but, with computer-controlled servosystems, can be run in many modes and generate no sidewall friction.
Consolidation tests are usually run in steps of sequentially higher but constant vertical stress, with each step maintained for about 24 hr. Such step durations allow dissipation of excess pore-fluid pressure and thus measure primary consolidation, but record only part of the secondary consolidation or creep that persists at exponentially decreasing rate with time (e.g., Bjerrum, 1967). Consolidation tests can also be run at constant strain rates, which generate far more data points, but the rate must be sufficiently slow that excess pore pressures do not become significant. Even slower strain rates result in lower porosities because a component of creep is included. The consolidation tests in this study were run at a constant rate of stress increase. The advantage of this deformation path is that, with increasing consolidation, the compressibility decreases and the strain rate drops, which compensates for the tendency for pore pressure to rise as the permeability decreases. This method also includes some creep, which is not well controlled.
A typical uniaxial consolidation test on an undisturbed sediment sample begins at a stress state well within the yield envelope, usually at some isotropic stress state. Although the initial deformation is theoretically elastic, a very low starting stress may produce an initial strain rate greater than that for linearly elastic because of microcrack closing and also because of initial system compliance. Such an initial response is generally followed by a phase of reconsolidation showing linear relationships among stress and strain parameters. This portion of the deformation path appears to provide the best basis for the determination of the elastic moduli of the sediment related to the physical properties at yield (Fig. 2).
Yield is generally signalled by an increase in strain for an increment of stress as well as by an increase in stress ratio. However, the exact stress state picked as yield is often in question, either because of poor definition of the break in slope or because the stress-strain relationships can be complicated by effects such as cementation. The functional definition of the yield stress in an experimentally consolidated sediment is the maximum effective vertical stress to which that sediment was previously subjected. Tests on experimentally consolidated sediments with unload-reload cycles show that the yield stress can be reasonably well approximated by extending tangents from the linear elastic portion of curves and from the first segment of the subsequent consolidation curves
(Fig. 2). More sophisticated methods, involving the extremum of the derivative of the e-log
v' or
a-v' curves, have been developed by geotechnical engineers (e.g.,
Casagrande, 1936; Janbu, 1985). Some of these methods generate a yield state that lies off the actual stress path of the test
(Fig. 2).
Most consolidation tests are performed in odometers that do not have the capability of measuring horizontal stress and thus provide only
a-v' relationships for the estimation of yield. In tests on highly consolidated sediments, there may be very little difference in axial strain increment from elastic to plastic regimes and the yield stress can be literally impossible to determine. A few more advanced odometers and triaxial cells have the ability to measure
h', providing a much wider range of possible relationships with which to estimate yield. A few of these include
vs. m',
vs. a, and
h' vs.
v'. The best estimate of yield stress relies on as many relationships as possible.
In its most rudimentary geological application, the yield stress is compared with the calculated effective vertical stress applied to the sample when it was collected, assuming a hydrostatic pore-fluid pressure. If the two are equal, the sediment is termed normally consolidated. Very often in geologic settings, the measured yield stress is greater than the calculated in situ effective vertical stress with hydrostatic P, or
vh'. This implies that the sample has been subjected to a greater
v' in the past than the in situ stress when collected, and the sediment is termed
overconsolidated. The usage of this term is also similar to that in geotechnical engineering, but the term underconsolidation is unique to geological applications. An underconsolidated sample refers to the condition in which the yield stress is less than
vh' because of P greater than hydrostatic pressure, although the
sediment may actually be normally consolidated with respect to the applied effective vertical stress.
Just as there is a diagnostic relationship between yield stress and
v', there is a parallel relationship between porosity and
v'. Thus, porosity can also be used as an indicator of consolidation state if a reference curve relating
and
v' for the sediment is available. Normally consolidated sediments follow this relationship, whereas overconsolidated sediments have a lower porosity at the same
v'. Thus, under some conditions, yield stress and porosity can be parallel indicators of consolidation state.
Occasionally, naturally consolidated sediments behave similarly to their experimentally consolidated analogs, but too often they do not and the reasons for the differences are seldom fully appreciated. Reasons for these differences include cementation, inappropriate initial test stress states, and inapplicable natural stress paths. In addition there are serious questions concerning the effect of geologic time on the strains and stress ratio associated with consolidation.
Natural sediments are usually cemented, which refers to the condition where sediment strength is increased without an increase in applied stress and most often with little reduction in porosity. Cementation is widely recognized by geotechnical engineers, who refer to it as structuration, microstructure, bonding, and aging (e.g., Burland, 1990). Cementation includes such effects as electrochemical bonding and secondary mineralization but is usually recognized by its effects on the sediment mechanical behavior as displayed on reconsolidation test curves (Fig. 3).
Cementation leads to a yield stress that exceeds the
v' ever applied to that sediment and thus to an expanded yield envelope. In the low-stress environment of geotechnical engineering, this expanded yield envelope is assumed to be geometrically similar to that of the uncemented sediment
(Lerouiel and Vaughan, 1990), but this assumes that the anisotropy of cement strength is the same as the intergranular component of strength. It remains to be seen whether this assumption can be applied to the higher stresses of geological environments.
Destruction of cement ("decementation" or "destructuring") leads to the reduction of yield strength and to the contraction of the yield envelope. Cement can be destroyed by a variety of processes including subyield stress and fatigue (Burland, 1990). Disturbance of DSDP/ODP cores during drilling and handling, especially those from shallower depths, is almost certainly a cause of cement destruction.
Cemented sediments are identified by a number of characteristically shaped curves generated during reconsolidation-consolidation tests (e.g.,
Janbu, 1985). The routine v-v' curve shows a sharper break in slope at yield, followed by a very rapid increase in strain with increasing stress, but this effect can be rather subtle. More pronounced effects of cementation are seen on the
h'-v',
-m', and
-v curves. For example, the
-m' curve for a cemented sediment shows a rise well above the Ko line before rejoining it as
m' or p' increases (Fig. 3).
The definition of yield stress in a cemented sediment is a contentious issue, as it might be chosen anywhere between the departure from linear elastic conditions and the arrival at Ko condition (Fig. 3). The sharp reduction in the slope of the q-p' curve marks an irreversible destruction of cementation and thus represents yielding, but it may not mark the position on the yield envelope appropriate for uniaxial consolidation. On the other hand, the choice of a yield stress at or near the condition where the stress path assumes a Ko ratio (e.g., Janbu, 1985) would seem to represent the virgin uniaxial condition of a decemented sediment, for which the yield envelope has already contracted. The initial isotropic stress condition from which the test is begun may also affect the nature of reconsolidation relationships (Fig. 3). If, as expected, the elastic relationships during the pre-yield phases of the "overconsolidated" and normally consolidated starting conditions, the stress path will meet the Ko line either at an incorrect yield stress or it will produce curves similar to that for cemented sediments (see Mesri and Hayat, 1993).
Cemented sediments generally have higher porosities than uncemented equivalents subjected to the same consolidation stresses because the stress applied to the sample is shared between the cement and simple intergranular contacts. Both strongly cemented sediments and overpressured (underconsolidated) sediments have higher porosities at a given depth than an uncemented reference sediment under hydrostatic pore pressure. In the former case, however, the yield stress will be higher than that for the reference whereas in the latter case the yield stress will be lower. Although this observation and the shape of the test curves serve to qualitatively distinguish the two cases, both cementation and anomalously high pore pressures can occur simultaneously, in which case the contributions of each are difficult or impossible to separate quantitatively (e.g., Karig, 1993).
The effect of strain rate, or, equivalently, of time, is also very important in comparing the results of our laboratory consolidation tests with in situ conditions. Consolidation at geologic loading rates may exceed 2 MPa/m.y. for rapid sedimentation but are at least 1012 times slower than our laboratory rates. Secondary consolidation or creep should thus be largely complete during the natural consolidation caused by progressive sedimentation, whereas is it is only partial during laboratory consolidation. Creep during consolidation can lead to quite large porosity reductions for high-porosity sediments (e.g., Bjerrum, 1967), but appears to decrease as porosity decreases and consolidation stress increases (Graham et al., 1983).
In contrast to the obvious and commonly significant effect of time on porosity during consolidation, there appears to be relatively little effect of time on the stress ratio, at least on a laboratory time scale
(Karig and Hou, 1992). There is much debate as to whether sediments, especially clay-rich
lithologies, maintain their short-term stress ratio over geologic time or tend toward an isotropic stress state (see references in Karig and Morgan, 1994). Thus, the applicability of the laboratory-determined stress ratio for the estimation of in situ
h' has not been adequately tested. Perhaps the most critical factor is that the sediment stress history can markedly effect the in situ
h'. For instance, the sediment might have been subjected to elastic unloading, which could radically change the in situ stress ratio but not the yield stress.
