Understanding the Effects of
Centre for Geotechnical Research
Shale is the most abundant rock type in a group called Wianamatta. The group is the major geological sequence in the Sydney Basin. Bringelly shale, the major formation of the group is comprised of claystones and siltstones with occasional sandstone layers. Weak cementation and presence of swelling clay minerals have caused a number of problems with structures founded on Bringelly shale. The results of triaxial tests on reconstituted specimens are compared with the behaviour of a natural rock specimen. In order to study the de-structural behaviour of the shale, the material was subjected to high stress so that low porosity structure can be produced. Same materials that have experienced higher stress show significantly reduced frictional strength and pattern of behaviour inconsistent with critical state concepts.
Keywords: Sydney Basin, Bringelly shale, structure and bonding, rocks
Engineers have often encountered significant problems involving shales and other argillaceous rocks. Shale comprises the upper rock layer for the majority of suburban Sydney, covering a total of approximately 1125 km2 . In the current study the engineering performance of the claystone/siltstone that comprises the majority of the Bringelly shale has been investigated. This involved index and mechanical properties (William and Airey, 1999a, 1999b). Due to the poor core recovery from conventional diamond core drilling with water flush, obtaining suitable specimens for laboratory tests has been a difficult task. The swelling behaviour of Bringelly shale is a consequence of about 22% of the shale constituents being comprised of swelling clay minerals. There is some evidence of recrystallisation of mica at particle contacts, but there is no evidence of induration and only small amounts of siderite and organic matter that can act as cementing agents are present. To investigate the mechanical behaviour of the shale, tests on reconstituted specimens were performed so that lower bound values for strength and stiffness can be determined. Responses from reconstituted specimens at stresses more typical of engineering soil is compared to the behaviour of a natural rock specimen. This paper is concerned with the different behaviour shown by the reconstituted material at different porosities, and the implication of their different behaviours compared to that of the natural shale.
Block samples of the rock shale were obtained from four different sites (Figure 1) namely Mulgoa (M), Badgerys Creek (B), Kemps Creek (K), and Horsley Park (H).
Figure 1. Location map of the study area
The block samples were obtained following ongoing excavation of the quarry floor, and the core samples by diamond drilling below the quarry floor. It has been found that the physical and index properties of the shale are similar at all four locations. The grain size distribution of the material is shown in Figure 2. The figure shows 55%, 41%, and 4% of clay, silt, and sand respectively.
Figure 2. Grain size curve of the tested material
Two methods were adopted for preparing specimens from pulverised rock. The first involved mixing the crushed shale with water to form slurry at moisture content close to the liquid limit. The mix was then compressed one-dimensionally to give a final vertical stress of 80 kPa. The second method was used to prepare sample with low porosity. Dry soil was compressed vertically to reach a target void ratio of 0.15.
Two triaxial cells were used in the tests, a conventional cell for tests with effective confining pressure up to 1 Mpa, and a rock cell for tests with confining pressure up to 60 Mpa. Extruded specimens and a core specimen were then placed to the designated triaxial cell where samples were isotropically consolidated to the desired stress in stages. In the conventional triaxial cell, samples were subjected to standard drained and undrained triaxial tests while in the rock cell samples were tested drained as the drainage only occurred from the base of the sample.
The responses in isotropic compression are shown in Figure 3. Points representing the equilibrium states of void ratio, e and mean effective stress, p' = s1(1 + 2s3)/3, are plotted and the isotropic normal compression line (INCL) has been drawn on the figure. This figure shows that at higher values of p' ( >10,000 kPa) the curve flattens out. For mean effective stresses less than 10,000 kPa, the INCL is given by:
with the parameters l = 0.07 and N = 0.85. For mean effective stresses higher than 10,000 kPa the INCL line is given by:
where A and B are constants that can be chosen so that the INCL is continous with no change of slope at p' = 10,000 kPa
Figure 3. Isotropic compression of Bringelly shale
It can be seen from Figure 3 that the slope of the isotropic unloading curve reduces as the maximum effective stress increases, as does the slope of the INCL. The samples that were prepared by the dry-press method had experienced a maximum effective stress of approximately 30 MPa and a minimum void ratio of 0.15 prior to extrusion from the mould. However, this void ratio increased to 0.33 due to yield prior to reloading in the triaxial cell. The natural rock on the other hand has measured a minimum void ratio of 0.15, on reloading the void ratio reduced to 0.12 due to a compressible response.
The effective stress paths from a series of undrained tests on isotropically normally consolidated samples are shown in Figure 4 in a plot of deviator stress q = s1-s3 versus mean effective stress, p' = s1(1 + 2s3)/3. For these samples, which have not experienced effective stress greater than 1 MPa, a critical state line (CSL) can be given by M=1.14 and a corresponding effective friction angle f=28.5°.
Figure 4. Effective stress paths of normally consolidated samples
It is expected that samples with a given overconsolidation ratio (OCR) will behave similarly once allowance is made for differences in confining stress level (e.g. Atkinson and Bransby, 1978). Figure 5 shows deviator stress normalised by the effective consolidation stress, p/c at the start of shearing for a series of drained tests. The tests involved reconstituted and natural rock samples. The reconstituted ones have been divided into two groups, those with relatively low pre-consolidation stress, p'max £ 6000 kPa, and those with p'max = 60,000 kPa.
From Figure 5 it can be seen that similar and normalisable responses are obtained for effective confining stresses up to 1 MPa. It was also found that the response of a normally consolidated sample with p/c = 6 MPa was similar to that at lower stresses. It is therefore believed that for maximum consolidation stresses of up to 6 MPa the normalised responses for a given OCR are unique, as reported in numerous studies on reconstituted clays (e.g. Atkinson and Bransby, 1978). Figure 5 shows that the responses of the samples that have been compressed with a maximum stress of 60 MPa
Figure 5. Influence of OCR on normalised deviator stress, axial strain responses from drained tests
all lie significantly below the curves of the same OCR where the samples have experienced a maximum stress of 1 MPa. Furthermore, the curve of reloading for the natural rock lies just below the curves that experienced a maximum stress of 60 MPa. This may indicate that both the final strength and the normalised stiffness are lower for the natural and highly compressed samples. The ultimate friction angle for the natural rock sample and highly compressed samples are 18° and 21.70 respectively, significantly less than the critical state friction angle, f=28.5°, measured for the lower density samples. When the confining stress reaches 60 MPa for the reconstituted samples, the void ratio is about 0.15. This void ratio is corresponding to the void ratio of the natural rock prior to reloading. With such low void ratios it can be expected that the potential for further volume change on shearing is reduced and this is shown in Figure 6.
Figure 6. Influence of OCR and pre-consolidation stress on volumetric strain responses
For normally consolidated samples the compressive volume strain for a sample with p/c = 60 MPa and a natural rock sample is about half and one quarter that for a sample with p/c = 1 MPa respectively. It may also be noticed that the tendency for volume expansion when OCR = 10 is greatly reduced when p/c = 60 MPa. For low pre-consolidation stresses the material behaves as would be expected from the concepts of critical state soil mechanics, but the patterns of behaviour of the highly compressed samples are not consistent with the existence of a critical state line which would have required much greater dilation from the highly compressed sample with OCR = 10. One observation from these results is that the concept of an OCR as a means of normalising the results has little use for such highly compressed soils. This is discussed further below.
One of the assumptions of many models based on critical state soil mechanics is that sections through (q, p', e) space at constant e are similar in shape (e.g. Atkinson and Bransby, 1978). These sections can be explored by plotting q and p' normalised by the equivalent p/e on the INCL at the same e. To determine p/e the INCL as given by Equations 1 and 2 has been used. The resulting normalised plot is shown in Figure 7.
Figure 7. Paths in normalised q' - p' space for natural rock and reconstituted shale at low and high stress.
This figure shows clearly the different shape of the sections at constant e, and of the state boundary surface. The reloaded natural rock sample lies just below the highly compressed samples that lie significantly below those with a maximum stress of 1 MPa. The structural differences between the reconstituted and natural rock samples have reflected upon the reduction of the frictional stress in the natural rock which was indicated by the failure point of the natural rock that is positioned below the CSL of the highly compressed samples. The dry pressed samples that have been compressed to a void ratio of 0.15 and then allowed to swell back show intermediate behaviour. This behaviour is believed to be a consequence of the hetrogeneities created by the sample preparation method, because when sheared these samples reach an ultimate frictional strength identical to the samples that have only been lightly compressed, i.e. M =1.14, f=28.5°. The dry pressed samples also have shear stiffness similar to normally consolidated samples (William et al, 2001). It is apparent that the use of OCR and other conventional means of normalising soil behaviour are not appropriate for the dry pressed samples.
DISCUSSION AND CONCLUSIONS
Direct comparison of our results to other work is difficult since laboratory-induced high pre-consolidation stresses on shale with significant amount of swelling clay minerals have rarely been reported. However, a similar analysis by Picarelli et al (1998, 2003) was carried out on highly plastic intensely fissured clay shales from Italy. Picarelli et al show that the normalised state boundary surface of the intact shale lies below the surface for the reconstituted material tested at higher density and lower pre-consolidation stresses. They also suggested that the mechanism of deformation and strength are controlled by movements along joints and fissures. The results presented in Figures 6 and 7 agree with the test results produced and reported by Picarelli et al., but partially disagree with the mechanism of deformation. In this study deformation and strength believed to be controlled by the frictional strength. This strength is likely to be controlled by the state of particles alignment during consolidation. In this study, it is suggested that the fabric associated with the state of particles alignment and low porosity, created by the high stress, is the main contributor to strength reduction and hence to the deformation mechanisms.
A related mechanism can be proposed for the dry pressed samples. These samples show many of the characteristics of normally consolidated samples, even though they have been highly compressed and then reloaded to a state where they are clearly over-consolidated. It has been suggested that yielding occurring during 1-D unloading has created softened and highly sheared zones. These findings agree with the dilation behaviour that was detected from the natural rock curve (Figure 6). This could be attributed to the creation of sheared bands within which movement is controlling the deformation. It should be noted that there was no visual evidence of in-homogeneity and further microscopic investigation is required to confirm the proposed mechanism.
The natural Bringelly shale has a very low porosity, similar to that produced by the high stresses in this study, and in addition has lower friction angle and significant micro-cracking in the plane of the laminations. Consideration of these factors would suggest that the effective friction angle controlling the strength of the shale is £18°. Microscope studies suggest that bonding is relatively weak, and its affect on strength is not anticipated to be large.
It has been suggested that the reduced fractional strength and pattern of behaviour that deemed to be inconsistent with critical state concept are related to alignment of the clay particles and the promotion of sliding on planar surfaces between these aligned particles. These results are significant for two reasons. First, the difference in frictional strength of the intact shale and the highly compressed reconstituted material is due to the difference in their void ratio and structure. The frictional strength of the natural shale is lower than measured from reconstituted samples at stress levels typical of standard soil mechanics practice. Second, attempts to understand the significance of bonding and structure in the natural shale need knowledge of the same material in a reconstituted state, and this can only be achieved if the reconstituted material is compressed to comparable void ratios and then compared to the natural rock.
Mechanisms have been proposed to explain the different deformation and strength responses. Confirmation of these findings will require further mechanical testing of the natural shale at different reloading stresses, which is in progress. Microscope studies are required to confirm that different mechanisms are active at low and high stress levels.
This study was supported by a grant from the Australian Research Council into the Engineering behaviour of cemented geomaterials.
|© 2004 ejge|