Each Isotropic Consolidation Undrained Triaxial Deformation experiment consists of two parts. First, the specimens were saturated and subsequently consolidated under isotropic stress conditions to the desired effective consolidation pressures. They were then deformed axially to failure through axial loading under undrained conditions. In natural, undisturbed soil samples the degree o f saturation depends on the field conditions such as the depth from which the sample was collected with respect to the ground water table, capillary saturation, time since the last significant precipitation event, and evaporation. Saturation in the laboratory was therefore necessary to make the test conditions between different samples as similar as possible.
Soils are generally a multiphase system consisting o f a mineral phase (the mineral skeleton) and a fluid phase (the pore fluid) which consists o f air and/or water (Figure 5.2.7) (Lambe & Whitman, 1979). The nature of the fluid phase will influence the characteristics o f the mineral surfaces and affect the process o f force transmission at particle contacts. As a consequence, it may be expected that an increment o f stress applied to a soil mass would be carried in part by the mineral skeleton as an effective stress, and in part by the pore fluid as a pore fluid pressure (Terzaghi & Peck, 1967; also see Appendix I). In a fully water-saturated soil the compressibility o f the soil skeleton is substantially greater than that o f the pore fluid at the prevailing stress states (Atkinson & Bransby, 1978). Thus, essentially all o f a stress increment applied to a fully saturated soil is carried by the pore fluid during undrained isotropic compression (Skempton, 1954; Bishop, 1976; Lambe & Whitman, 1979). Since both soil grains and the pore water are effectively incompressible, the volume o f a fully saturated sample can only change in response to deformation if pore water is displaced from the pore space. Under such conditions pore fluid pressure is reduced and the effective stress increases. The variation of soil volume with time following application of a load will therefore be governed by complex interactions between the total stress, pore fluid pressure, permeability and compressibility. This time-dependent process of volume change as pore water being expelled out from the pores is known as consolidation (Lambe & Whitman, 1979) and is responsible for the increase in effective stress.
For the Barbados samples isotropic consolidation was undertaken by increasing the confining fluid pressure incrementally until the desired pressure was achieved (BS 1377, 1990). Conversely, in all o f the tests on the Taiwan samples, cell pressure was not increased incrementally but applied directly at the desired magnitude for consolidation. The change in procedure was due to the low permeability o f the Taiwan material. When
a sample is isotropically compressed under undrained conditions the applied stress increment will be taken by the pore fluid and will generate a pore fluid pressure (under fully saturated condition, this is equivalent in magnitude to the increase in the total confining stress). Subsequently, when one end of the sample is opened to the volume gauge a head difference will be established across the sample and flow will be initiated. If the sample has a very low permeability equilibration o f pore fluid pressures takes an extremely long time and can result in an asymmetric distribution o f compressional strain in the sample. This can be partly overcome using a larger stress interval.
During consolidation, without using side drainage, only one end o f the specimen was open to the volume gauge which permitted pore fluid to drain to a constant back pressure in the volume gauge. The pore fluid pressure at the undrained end o f the sample was monitored and taken as the true pore fluid pressure within the specimen. The back pressure served to maintain the pore water saturation in the sample by keeping trace amounts o f air in solution. The test results are unaffected by the magnitude o f the back pressure, because these are dependent solely on the prevailing state o f effective stress (Atkinson & Bransby 1978).
During the consolidation process the effective stress gradually increased as pore water was expelled from the specimen, causing a reduction in pore fluid pressure. At the end o f consolidation, when the undrained pore fluid pressure was equalised with the applied back pressure, the drainage line was closed and the sample was deformed axially at a constant rate of displacement while the confining fluid pressure was held constant. It was noted in Addis & Jones (1989) that in testing porous sedimentary rock samples the axial displacement rate within a certain range (4*10"^ m min."^ to lO'^ m min."^) does not have a noticeable effect on the material behaviour for this type o f deformation. However, for naturally structured materials with low permeability such as those studied here, the displacement rate has to be slow enough to maintain the pore fluid pressures at both ends in an equilibrium state. If the displacement rate is too fast, this will lead to a non-uniform pore fluid pressure distribution across the sample and thus non-uniform effective stress states, causing inaccurate results. An axial displacement rate of 5*10"^ m min.’^ was used for Barbados tests and a rate o f 2*10"^ m min.'^ for Taiwan experiments.
For all the experiments reported here during undrained triaxial deformation the deviatoric stress increased until the sample failed^. The behaviour o f natural soils, however, rarely follows the ideal behaviour predicted by published models such as Cam Clay or the
^Failure was deemed to have occured which the sample continued to shorten at constant axial stress and pore pressure.
critical state. B eca u se o f this, the tests w ere each taken to as large an axial strain as possible.
Volumes Weights
(a) (b)
Figure 5 .2 .7 R elationships am ong soil phase; (a ) elem ent o f natural soil; (b) elem ent separated into phases.