ADVANCED LABORATORY CHARACTERISATION OF LONDON CLAY
Thesis submitted to University of London in partial fulfilment for the degree of Doctor of Philosophy and for the Diploma of Imperial College London
APOLLONIA GASPARRE July 2005 Department of Civil and Environmental Engineering Imperial College London London SW7 2BU
ABSTRACT New findings about the geology of London Clay (King, 1981) have highlighted the importance of investigating the relationship between geology and engineering behaviour for stratified soils. Recent events, such as the Heathrow tunnel collapse in 1994 and the poorly predicted ground movements at St. James Park during the construction of the Jubilee line extension have also highlighted a local need to revise the general proprieties of the material with which engineers in London deal. This research aimed at finding a framework for the London Clay relating the engineering proprieties of this material to its geological features.
High quality samples from different depths in London Clay were tested in their intact and reconstituted states using oedometer and advanced triaxial apparatus.
The lithological units of the London Clay at the site have been
accounted for in analysing the mechanical response of the clay.
The structure and the nature of the clay from different strata were investigated microscopically and correlated with its large and small strain mechanical response. Shallower units showed a more open structure and higher clay content than deeper units.
Samples from the same units had the same mechanical
behaviour and engineering parameters, regardless their depth within the stratum, but differences were found between the different units, which reflected the differences in the nature and structure of clay from each stratum. The behaviour in both compression and shearing seemed to be dominated by the structure of the clay as well as by its nature, so that clay from units having a more packed and orientated structure showed a stiffer response and higher strengths than the clay from units with a more open structure. The behaviour of the clay was also investigated in the elastic region and the elastic parameters confirmed the effects of lithology on sample behaviour.
ACKNOWLEDGMENT I share the success of this research work with my supervisor Dr M.R. Coop, who is a great teacher and an enthusiastic supporter. His continuous presence has been precious throughout this project.
I would also like to thank all the research group involved in the London Clay Project. In particular, thanks to Prof. R.J. Jardine for his advises and suggestions, to Will, who helped me through the difficult process of understanding the geology of London Clay and to Satoshi and Minh, who shared with me many experiences on site and in the laboratory. I would also like to thank Akihiro and Pedro for their help in the sampling process.
I am grateful to Prof. D. Hight, Prof. Chandler, Dr. M. De Freitas, and Dr. J. Standing for sharing with me some of their knowledge on London Clay and to Dr. J. Huggett for her analysis on the microstructure of the clay.
Special thanks to the technicians Steve, Alan and Graham, because this research would not have been possible without their constant assistance. I also express my gratitude to the MSc students Ana, Naeem, Jimmy and Eduardo, who performed some of the tests used in this research and to Giovanny, who often supervised my tests while I was away and helped with the interpretation of the bender element signal.
Working at Imperial College has been a great experience and I would like to thank all the research staff and students in the Soil Mechanics Section for making this place so special.
TABLE OF CONTENT ABSTRACT
TABLE OF CONTENT
LIST OF TABLES
LIST OF FIGURES
LIST OF SYMBOLS
1.1 Background of the research
1.3 Structure of the thesis 2
2.1.5 Sedimentation and post-sedimentation structure
2.1.6 Degree of structure
2.2 Large strain behaviour
2.2.1 Normalising parameters
2.2.2 The Sensitivity Framework
2.2.3 Post yield behaviour
2.2.4 Anisotropic destructuration
2.2.5 Destructuration in swelling
2.2.6 Effect of weathering
2.3 Large strain strength
2.4 Yielding behaviour
2.4.1 Y1 surface
2.4.2 Y2 surface
2.4.3 Y3 surface
2.5 Small strain behaviour
2.5.1 Elastic parameters
(a) Shear modulus (b) Interpreting bender element signals
(c) Other elastic parameters 2.5.2 Influence of recent stress history
2.6 Creep 2.6.1 Effects at small strains
2.7 Strain rate effects 2.8 The influence of fissures 3
3.1 Introduction 3.2 The London Clay formation 3.2.1 Depositional processes: London Basin and Hampshire Basin (a) Lithological units 3.2.2 Post-depositional processes (a) Influence of the Alpine orogeny
87 88 89 91 91
3.3 Mineralogy of the London Clay 3.4 Macrofabric 3.5 London Clay proprieties in west London 3.5.1 Geology (a) Lithological units
92 93 94 94 95
3.5.2 Index proprieties
3.5.3 In situ stresses and ko
3.5.5 Cone penetrometer tests at T5
3.5.6 Shear strength
3.5.7 Anisotropy and stiffness 4
4.3.2 Conventional stress path cell
4.3.3 Stress path apparatus for 100mm samples
(a) Axial and radial LVDTs
(b) Mid-height probe
(c) Bender elements
4.3.4 The medium pressure apparatus
4.3.5 High pressure triaxial apparatus
4.3.6 Calibration and accuracy
4.3.7 Load cell connection
5.2.1 Rotary Core Samples
5.2.2 Block samples
(a) Sampling process 5.3
Natural Samples- triaxial tests
153 154 154
(a) Rotary cores
5.3.2 Preparation of the cell
5.3.3 Preparation of the sample
5.3.4 Expected effective stress
5.3.5 Test procedures (a)
Unconsolidated undrained tests
Consolidated drained or undrained tests from isotropic state
(c) Investigation of the influence of the recent stress history (d) Tests from the in situ stress state 5.3.6 Effect of temperature 5.4
Natural samples –oedometer tests
158 158 159 160 161 164 166
5.4.1 Sample preparation
5.4.2 Testing procedures
5.5 Reconstituted samples
5.5.1 Triaxial tests (a)
(b) Sample preparation
(c) Test procedures
5.5.2 Oedometer tests
5.6 Analysis of the data
5.6.1 Calculations and corrections (a) Water content
(b) Specific volume
(c) Area correction
(d) Membrane correction
(e) Volumetric and shear strains
5.6.2 Shear plane analysis
5.7 Nomenclature of the tests
DESCRIPTION OF THE SOIL
6.2 Sample descriptions
6.3 Microstructure of the London Clay 6.3.1 SEM analysis
(a) Unit A3
(b) Unit B2
(c) Unit C (d) Comparison between different lithological units 6.3.2 Chemical micro-analysis 6.4 Sample characterization
198 199 200 201
6.4.1 Specific gravity Gs
6.4.2 Water content distribution
6.4.3 Atterberg limits
LARGE STRAIN BEHAVIOUR
7.2 Intrinsic proprieties: reconstituted samples
231 231 231
7.2.1 Behaviour in compression
7.2.2 Shearing behaviour
(a) Critical state line
(b) Normalised shearing behaviour 7.3 Natural samples 7.3.1 Behaviour in compression
235 235 235
(a) Stress Sensitivity
(b) Swell Sensitivity
7.3.2 Lithological units and compressibility
7.3.3 Normalised compression behaviour
(a) New normalisation
7.3.4 Destructuration due to swelling
7.3.5 Shearing behaviour
(a) Shear plane characteristics
(b) Pore pressure distribution
(c) Shear strength
(d) Sample size effect
(e) Sample quality
7.3.6 Strength envelopes and lithological units 7.3.7 Influence of pre-existing fissures
(a) Fissures due to drying
(b) Strength on fissures
(c) Lithological units and fissures 7.4 Structure and destructuration of natural samples 7.4.1 Normalised strength
255 255 256
7.4.2 Destructuration in swelling
7.4.3 Destructuration due to anisotropic compression
SMALL STRAIN BEHAVIOUR
8.2 Lithological Unit C
8.2.1 Bender element tests 8.2.2 Static probes
8.2.3 Monotonic loading tests
8.2.4 Elastic parameters
8.2.5 Kinematic surfaces
(a) Y1 surface
(b) Y2 surface
8.2.6 Stiffness degradation 8.3 Unit B2 8.3.1 Sub-Unit B2(c) (a) Bender elements tests (b) Static probes
348 349 350 350 350
(c) Elastic parameters
(d) Kinematic surfaces
(e) Stiffness degradation 8.3.2 Sub-Unit B2(a) (f) Bender element tests (a) Static probes
354 354 355 356
(b) Elastic parameters
(c) Kinematic surfaces
(d) Stiffness degradation
8.4 Unit A3 8.4.1 Bender elements tests
8.4.2 Static probes
8.4.3 Elastic parameters
8.4.4 Kinematic surfaces
8.4.5 Strain rate dependency
8.4.6 Stiffness degradation 8.5 Influence of the lithological unit
8.5.1 Elastic parameters
8.5.2 Kinematic Surfaces
(a) Y1 surface
(b) Y2 surface
8.5.3 Strain energy
8.6 Effect of fissures on the elastic parameters 9
EFFECTS OF RECENT STRESS HISTORY
9.2 Case 1: short approach stress path
372 473 473 474
9.2.1 Creep allowed
9.2.2 Creep not allowed
9.3 Case2: long approach stress path 9.4 Effects of angle of rotation on the kinematic surfaces
9.4.1 Shear modulus
9.4.2 Elastic surface
9.4.3 Y2 surface
9.4.4 Effect of creep
10.1 Suggestions for future work
Calculations of the in situ stress state and approach stress path
Measurements of the elastic parameters APPENDIX 7.1 Shear planes
LIST OF TABLES Table 2.1: Normalised secant moduli at 50% maximum deviatoric stress on London Clay samples (Costa-Filho, 1984, data from Sandroni, 1977) Table 2.2: Normalised secant moduli at 50% maximum deviatoric stress on Lias clay samples (Costa-Filho, 1984, data from Maguire, 1975) Table 3.1: Index proprieties of London Clay at Wraysbury (Skempton et al., 1969) Table 3.2: Index proprieties of London Clay at Ashford Common (Bishop et al., 1965) Table 3.3: Shear modulus ratios for London Clay (Wongsaroj et al., 2004) Table 4.1: Details of bender elements Table 4.2: Summary of key features of typical laboratory instrumentation used in this project Table 5.1: Summary of all tests performed in natural samples Table 5.2: Coordinates of borehole and block samples Table 5.3: Division of the borehole sample into lithological units Table 6.1: X-ray analysis on samples from different lithological units Table 6.2: Specific gravity of grains at T5 and Ashford Common Table 6.3: Index proprieties Table 7.1: Parameters for reconstituted samples isotropically compressed Table 7.2: Parameters of reconstituted samples one-dimensionally compressed Table 7.3: Yield stresses and Stress Sensitivity for samples from different lithological units Table 8.1: Shear moduli during the consolidation stress paths for samples from Unit C (*refer to Figure 8.1) Table 8.2: Elastic parameters derived from static probes and monotonic loadings of samples from Unit C Table 8.3: Strain energy at the in situ and yield stresses for Unit C Table 8.4: Shear modulus during the approach stress paths for samples from SubUnit B2(c) (*refer to Figure 8.18) Table 8.5: Elastic parameters derived from static probes and monotonic loadings on samples from Sub-Unit B2(c) Table 8.6: Incremental strain energy for samples from Sub-Unit B2(c) xi
Table 8.7: Shear moduli during the approach stress paths for the samples from Sub-Unit B2(a) (*refer to Figure 8.38) Table 8.8: Elastic parameters derived from static probes and monotonic loadings of samples from Sub-Unit B2(a) Table 8.9: Incremental strain energy for samples from Sub-Unit B2(a) Table 8.10: Shear moduli for samples consolidated along the approach stress paths used for Unit A3 (*refer to Figure 8.55) Table 8.11: Elastic parameters derived from static probes and monotonic loadings of samples from Unit A3 Table 8.12: Elastic parameters derived from static probes and monotonic loadings of samples from Sub-Unit B2(a) consolidated to the in situ stress state of Unit A3 Table 8.13: Incremental strain energy for samples from Unit A3 and Sub-Unit B2(a) consolidated to the in situ state of Unit A3 Table 8.14: Average of the independent elastic parameters measured in the static and dynamic probes on samples from different lithological units Table 9.1: Strains developed during the approach stress paths for Probes 17-23o e and 17-157o e (refer to Figure 9.1) Table 9.2: Strains developed during the approach stress paths for Probes 17.3-75o c and 17.3-105o c (refer to Figure 9.5) Table 9.3: Strains developed during the approach stress paths for Probes 17.3-L30o e and 17.3-L150o e (refer to Figure 9.9) Table 9.4: Elastic parameters for probes on Samples 17SH and 17.3SH
LIST OF FIGURES Figure 2.1: Structure of the main clay units (Veniale, 1983; Cotecchia, 1996) Figure 2.2: Classification of fabric (Sides & Barden, 1970) Figure 2.3: Schematic diagram showing enhanced resistance of natural clays in compression Figure 2.4: Typical compression curves for (a) clays with sedimentation structure and (b) clays with post-sedimentation structure (Cotecchia & Chandler 1997) Figure 2.5 Influence of structure: proposed normalising parameters Figure 2.6: Sedimentation compression curves for normally consolidated clays (Skempton, 1970) Figure 2.7: The intrinsic and sedimentation compression lines (Burland, 1990) Figure 2.8: (a) In situ states for normally consolidated clays (Skempton, 1970) and (b) interpretation of the data indicating sensitivity (Cotecchia and Chandler, 2000) Figure 2.9: Geometrical definition of the strength sensitivity (Cotecchia & Chandler, 1997) Figure 2.10: Stress and Strength Sensitivity relationships for clays having in situ states on the right of the ICL (Chandler 2000). Figure 2.11: Stress and Strength Sensitivity relationships for clays having in situ states on the left of the ICL (Chandler 2000). Figure 2.12: The Sensitivity framework (Cotecchia & Chandler 1997) Figure 2.13: State boundary surface of reconstituted and undisturbed Pappadai clay consolidated to stresses before yield (Cotecchia, 1996). Figure 2.14: Compression curves of (a) clay with a stable structure (Coop & Cotecchia, 1995) and (b) clay with a meta-stable structure (Burland, 1990) Figure 2.15: Normalised SBS for samples compressed before and beyond gross yield: (a) Bothkennar Clay (Jardine & Smith, 1991); (b) Pappadai Clay (Cotecchia, 1996); Valericca Clay (Amorosi & Rampello, 1998). Figure 2.16: Unique SBS for Pappadai Clay normalised by structure Figure 2.17: Isotropic and k0 compression for Pisa Clay (Baudet & Stallebrass, 2004, data from Callisto 1996) Figure 2.18: Destructuration of Bothkennar Clay (a) isotropic and k0 compression (b) shearing behaviour (Jardine & Smith, 1991) xiii
Figure 2.19: Behaviour of London Clay swelled to 1/6, 1/12, 1/16 the initial vertical effective stress (a) destructuration in one-dimensional swelling (b) normalised shear stress-horizontal displacement; (c) stress paths in constantheight direct shear box tests (Takahashi et al. 2005) Figure 2.20: Fissuring in the London Clay (Chandler & Apted, 1988) Figure 2.21: Variation of water content with different levels of weathered strata (Lias Clay, Chandler, 1972) Figure 2.22: Idealised relationship between effective overburden pressure and water content during the geological history of an overconsolidated clay (Chandler, 1972) Figure 2.23: Shear behaviour of London Clay samples at different levels of weathering (a) Undrained triaxial compression tests, (b) normalised stress paths (Chandler & Apted, 1988) Figure 2.24: Effects of weathering on Pappadai clay (a) Normalised state boundary surfaces of both the natural and the reconstituted samples (b) isotropic and one-dimensional compression behaviour of both the weathered (yellow) and the unweathered (grey) clay (Cafaro & Cotecchia, 2001) Figure 2.25: Effect of clay particles on the critical state friction angle and on the residual friction angle (Lupini et al. 1981) Figure 2.26: Idealised undrained shearing behaviour of overconsolidated clays with (a) low plasticity and (b) high plasticity (Jardine et al, 2004) Figure 2.27: Strength of stiff plastic clays (Jardine et al. 2004) Figure 2.28: Localization of strains and pore pressure distribution in London Clay (Sandroni, 1977) Figure 2.29: Scheme of multiple yield surfaces (Jardine, 1992) Figure 2.30: Definition of Y2 for Bothkennar Clay from drained cyclic tests (Smith et al. 1992) Figure 2.31: Normalised undrained stress paths for triaxial compression tests on Lower Cromer till samples consolidated to different values of K (Gens, 1982) Figure 2.32: Bounds for the elastic parameters and planes and lines representing special types of materials (Pickering, 1970) Figure 2.33: Configuration for measurement of stiffness of a cross-anisotropic soil under axi-symmetric loading (Pennington et al., 1997)
Figure 2.34: Effect of recent stress history on current stiffness (Atkinson et al. 1990) Figure 2.35: Stiffness response for tests for recent stress history of reconstituted London Clay (Atkinson et al., 1990) Figure 2.36: Compression paths and small strain stiffnesses for natural and reconstituted London Clay samples (Jardine, 1992) Figure 2.37: Stress probes and normalised elastic parameters for Gault clay (Lings et al. 2000) Figure 2.38: Stress dependent stiffness of a single Bothkennar clay specimen subjected to two different loading paths (Clayton & Heymann, 2001) Figure 2.39: Stiffness degradation curves of a Bothkennar clay subjected to two different loading paths (a) schematised stress paths (b) stiffness response (Clayton & Heymann, 2001) Figure 2.40: Stiffness degradation curves of London clay subjected to two different loading paths (a) schematised stress paths (b) stiffness response (Clayton & Heymann, 2001) Figure 2.41: Schematic behaviour in compression after ageing (Tatsuoka et al., 1998) Figure 2.42: Stress-strain curves after restarting loading at a constant strain rate (Tatsuoka et al., 1998) Figure 2.43: Effect of undrained creep on the shearing behaviour of Fujimori clay (Momoya, 1998, Tatsuoka et al, 1998) Figure 2.44: Effects of drained creep on subsequent undrained shearing for undisturbed Vallericca clay and Metramo silty sand (Santucci de Magistris, 1998 and Tatsuoka et al, 1998) Figure 2.45: Creep effect on fast rate shearing of undisturbed London Clay (Sandroni, 1977) Figure 2.46: Influence of time on (a) the shear stiffness of carbonate sand samples (Jovicic & Coop, 1997); (b) small strain Young’s Modulus of Metramo silty sand (Santucci de Magistris, 1998 and Tatsuoka et al. 1998) Figure 2.47: Development of the kinematic yield surfaces with time (Tatsuoka et al., 1998) Figure 2.48: Effect of strain rate change on Chuba Gravel (Tatsuoka et al. 1998) Figure 2.49: Effect of strain rate change on Vallericca clay (Tatsuoka et al. 1998) xv
Figure 2.50: Effect of strain rate change on Hostun Sand (Tatsuoka et al. 1998) Figure 2.51: Effect of strain rate on the very small strain Young’s Modulus (Tatsuoka et al. 1998) Figure 2.52: Stress-strain curve of Vallericca Clay at very small strains during (a) loading and (b) unloading at different rates (Tatsuoka et al. 1998, data from Santucci de Magistris, 1998) Figure 2.53: Strength envelopes on London Clay samples of different dimensions (Bishop, 1972) Figure 2.54: Strengths of samples of different diameters and samples that sheared along pre-existing fissures (Marsland & Butler, 1967) Figure 2.55: Stress-strain behaviours of intact samples and samples sheared along pre-existing fissures (Webb, 1964) Figure 3.1: Late Palaeocene geology (King, 1981) Figure 3.2: The North Sea Basin and London Clay formation (King, 1981) Figure 3.3: Eocene stratigraphy of the London Clay Formation in Southern Britain (King, 1981) Figure 3.4: The London Clay formation: idealised depositional sequences linked to sea level changes (King, 1981) Figure 3.5: Palaeocene and Eocene sections of the London Clay Formation (King, 1981) Figure 3.6: Main features of the lithological units in the London Clay (King, 1981) Figure 3.7: Identification of lithological units by water content (Hight et al., 2003) Figure 3.8: Correlation between the informal lithological division suggested by the BGS (2004) and King (1981) (BGS, 2004) Figure 3.9: Correlation between boreholes at different sites in the London Basin (BGS, 2004) Figure 3.10: Envelope of particle sizes for London Clay (King, 1991) Figure 3.11: Stratigraphical variation in lithology and clay mineralogy (σ’?p >σ’?v and YSR>OCR. 2.1.6
Degree of structure
The enhanced resistance of natural clays to compression is also reflected in shear strengths of the natural material that plot above the intrinsic State Boundary Surface (SBS) defined by the Critical State Framework (Smith et al. 9
1992; Calabresi & Scarpelli, 1985; Rampello 1989; Burland 1990). The locations of the natural compression curve and the natural SBS in comparison to the locations of the normal compression line and SBS of the reconstituted material can be used as a measure of the influence of the structure on the sample behaviour.
Burland et al. (1996) observed that the ratio of the normalised strength at the critical state (DE/DF in Figure 2.5) could be useful in measuring the influence of structure. They highlighted also that the cohesion is a significant parameter of bonding and the ratio between the cohesion of the natural and reconstituted materials might also be used to evaluate the effect of structure. In comparing the cohesion, though, the curvature of the failure surface of the natural material at very small stresses should be taken into consideration.
Terzaghi (1944) defined sensitivity St as the ratio between the undrained strength of undisturbed clay and the undrained strength of the remoulded clay at the same water content. The sensitivity is generally regarded as the parameter embodying the differences of the microstructures of the natural and the remoulded clay (e.g. Cotecchia, 1996). Schmertmann, (1969) defined “swell sensitivity” Ss as the ratio of the intrinsic to the intact swelling indices C* s/Cs, which can also be used as indicator of structure.
2.2 Large strain behaviour
Figure 2.6 shows the Sedimentation Compression Curves (SCC) derived by Skempton (1944) for a number of natural clays and plotted as relationship between the natural void ratio and the in situ vertical effective stress. Skempton and Northey (1952) showed that, by normalizing the sedimentation compression curves of reconstituted clays by using to the liquidity index, the compression behaviour of reconstituted material could be defined in a unique manner, so that
a unique line emerged. Burland (1990) introduced a normalizing parameter based on mechanical proprieties of the soil, the Void Index Iv : * * e − e100 e − e100 Iv = * = * e100 − e1000 Cc*
where e* 100 and e* 1000 are the intrinsic void ratios for one-dimensional compression corresponding to vertical effective stresses σ’v =100kPa and 1000kPa respectively and C* c
is the intrinsic compression index. In the
normalised graph Iv -σ’v , the compression curves of reconstituted materials then plot on a unique line called the Intrinsic Compression Line (ICL). In the same normalised plane Iv -σ’v , Burland (1990) fitted a regression line through the natural sedimentation compression curves determined by Skempton (1970), and identified a unique Sedimentation Compression Line (SCL) (Figure 2.7). The SCL of the natural soils lies above the ICL as result of the structure developed by the natural soils during the sedimentation process and the distance between the ICL and the SCL, called the “sedimentation sensitivity” (St), is a measure of the acquired strength of the natural sediments with respect to the strength of the reconstituted clay. The distance between the ICL and the SCL can be expressed by the ratio of the gross yield stress σ’?y , which is where the breakdown of the natural structure occurs in compression, to the “equivalent” pressure σ∗ e, which is the vertical effective stress on the ICL corresponding to the void ratio at the gross yield of the clay.
The SCL plotted in Figure 2.7 refers to a set of sedimentation compression curves that were derived by Skempton (1970) for clays of similar composition, with organic contents lower than 10% and carbonate contents lower than 25%, resulting in sensitivities lower than 10. These SCCs plotted very close to each other over a quite narrow band. In Figure 2.8 these sedimentation curves are plotted in the LI-σ’v plane together with the sensitivity values associated with each curve. Clays on the same SCC have the same sensitivity and lower sensitivity clays are towards the lower bound of the band, while high sensitivity clays plot toward the upper bound.
The Sensitivity Framework
Cotecchia & Chandler (1997) developed the concept of ‘Strength Sensitivity’ introduced by Skempton (1970) and defined a framework for clay behaviour based on the sensitivity of clays, using data for the Pappadai clay, an overconsolidated clay from the South of Italy. They noticed that, considering the natural and reconstituted peak strength, qpeak and q*peak, if the Strength Sensitivity is defined by the equation:
the State Boundary Surface of the natural material is then scaled up compared to the SBS of the corresponding reconstituted soil. The strength sensitivity therefore represents the distance between the strength of the intact material and the intrinsic strength (Figure 2.9).
Figures 2.10 and 2.11 illustrate the basic definitions of the Sensitivity Framework, given by Chandler (2000). The author defined the Intrinsic strength line ISuL as a line plotted on a graph of Iv against the undrained strength S* u of the reconstituted material. In this plane, the Strength Sensitivity St was defined as: Su/S* u=St
where Su is the undrained strength of the natural soil. In the plane of vertical effective stress against void index or void ratio, the Stress Sensitivity is the distance between the yield stress of the natural material and the vertical stress on the ICL at the same void ratio (Cotecchia & Chandler, 1997, 2000): Sσ=σ’y /σ∗ e
The ‘in situ stress ratio’, IsSR, and the ‘yield stress ratio’, YSR were also defined geometrically. For states on the right of the ICL (Figure 2.10) the equivalent strength is given by the structural resistance of the soil to the vertical stress
(IsSR) together with the extra resistance required to load the sample to its yield stress: σ’y /σ* e= Sσ=IsSR.YSR
In the Sensitivity framework, Equation 2.5 is also equal to the strength sensitivity, and therefore: Su/S* u=St =IsSR.YSR=Sσ
For clays having states lying to the left of the ICL, σ* e>σ’e, (Figure 2.11) S* u might be greater than Su, so that IsSR could have values less than unity. The equivalent strength has then to be factored up for the reduced in situ stress resulting from erosion: Su=S* u.Sσ .IsSR
Rearranging (2.7): Su/S* u=St =Sσ .IsSR
In terms of the state boundary surface Equations 2.6 and 2.8 suggest that a geometric similarity exists between the SBS of the reconstituted and natural materials, so that the strength sensitivity is proportional to the ratio of gross yield stress to the corresponding gross yield stress on the same clay reconstituted (Figure 2.9):
Sσ= p’iy /p*iy ~p’koiy /p*koiy =St
where the symbols are defined as in Figure 2.9. Cotecchia (1996), and Cotecchia & Chandler (1997, 2000) observed also that on a graph of Iv-σ’v, soils with same strength sensitivity followed same sedimentation curves, whose position corresponds approximately to the sensitivity of the clay (Figure 2.12).
The SCLs of natural clays appeared to be approximately parallel to the reconstituted SCL, which, by definition, has St =1. Dividing the stresses by the equivalent pressure p*e, which corresponds to the effective stress on the intrinsic compression curve at the same specific volume of the soil, the influence of volume can be eliminated (Horvslev, 1937).
Cotecchia (1996) found that the
state boundary surface of natural Pappadai Clay samples, consolidated up to gross yield plotted and normalised for the volume by p*e, plotted on a unique curve with a constant size ratio relative to that of the corresponding reconstituted soil equal to St (Figure 2.13). 2.2.3
Post yield behaviour
After the gross yield, σ’y , structural breakdown starts to take place and the compression curves of natural materials tend to bend downwards. The destructuration is a gradual process, so that the yield stress σ’y is often not well defined (Burland et al. 1996), but, as the strains increase after gross yield, the strength sensitivity St is no longer a constant value. It has been observed that the compression curve after yield can converge towards the ICL, demonstrating a “meta-stable” structure that degrades with strains, or move along a line parallel to the ICL, demonstrating the presence of “stable” elements of structure that do not degrade with strains. Figure 2.14 shows the compression curves of Boom Clay, with a meta-stable structure, and Sibari Clay, with a more stable structure. Figure 2.14b also shows the swell sensitivity indices, i.e. Cs/C* s, as compression proceeds. As a result of structural breakdown of Boom clay, the swelling curves of the intact material tend to become parallel to the intrinsic swelling curve of the reconstituted soil. A soil can have both stable and meta-stable elements, so that after yielding the compression curve of the natural material can bend downwards towards the ICL, due to the breakdown of the metastable elements, but can then stabilise on a line which is parallel to the ICL and above it due to the presence of stable elements. Coop et al. (1995) suggested that the meta-stable elements in structure are likely to be associated with bonding, while the stable elements are likely to result from fabric.
The structural breakdown after yielding modifies the normalised soil behaviour. Jardine & Smith (1991) found that the ko consolidation beyond yield reduces the normalised compression strength of Bothkennar clay and Cotecchia (1996) showed that the normalised boundary surface of Pappadai Clay, compressed to a state above the gross yield, lies below that of the clay compressed to states below the gross yield (Figure 2.15). Analogous behaviour was found by Amorosi & Rampello, (1998) testing Vallericca clay. The nonunique state boundary surface in the plane q/p* e-p’/p* e is related to the inability of the equivalent pressure to describe changes in the mechanical behaviour of a structured clay induced by the “destructuration strains” that is the plastic strains that developed during compression and shearing (Cotecchia & Chandler, 1997; Amorosi & Rampello, 1998). Cotecchia & Chandler (1997 and 2000) pointed out that the size of the current gross yield surface of natural soils is not controlled solely by the plastic volumetric strains, as for reconstituted clays, but is controlled both by the volume change and by the change in strength sensitivity. In order to model the ‘post yield state’ of structured material Leroueil et al. (1984) highlighted the need of a more appropriate normalizing parameter that accounts for the progressive process of destructuration and Cotecchia & Chandler (1997) suggested to use the Strength Sensitivity St . They showed that, when the SBS of Pappadai clay was normalised for the structure (St =p’iy /p*iy ), as well as volume using p* e, a unique SBS curve appeared, as shown in Figure 2.16. The Strength Sensitivity changes after yielding with increasing strains due to the progression of structure breakdown. Sophisticated numerical models and normalizing parameters have been suggested (Kavvadas & Amorosi, 2000; Baudet & Stallebrass, 2004) to take into account the progression of the destructuration process as plastic strains develop, and to evaluate the influence of the volumetric and deviatoric components of the “destructuration strains”. Callisto & Rampello (2004) proposed to normalize the mean and deviatoric stresses by a parameter that is similar to the equivalent intrinsic pressure p* e, but which also includes the effect of the progressive breakdown of the structure.
Amorosi & Rampello (1998) suggested that for those soils that retain a “stable structure” after yield, (e.g. Vallericca clay), the state of the soil destructurated 15
under anisotropic compression should be taken as a reference for evaluating the influence of structure rather than the reconstituted state. Baudet & Stallebrass (2004) allowed stable and meta-stable elements of structure to be modelled considering the intrinsic proprieties of the soils, which do not require highpressure tests on natural samples to be performed.
These models have been shown to reproduce well the behaviour of normally consolidated soils, but still are less well able to simulate the behaviour of heavily consolidated clays that form shear planes. 2.2.4
Destructuration is related to the cumulative volumetric and deviatoric plastic strains, the “destructuration strains”, which can occur during both consolidation and shear stages. The mechanism of destructuration depends on the direction of the stress path (Jardine & Smith, 1991; Kavvadas & Amorosi, 2000; Baudet & Stallebrass, 2004) and on the fabric of the soil. The arrangement of the particles influences the capability of the soil to sustain better one or other component of the “destructuration strains”. The influence of the two components of the destructuration strains is still the subject of debate in the literature. Callisto (1996) showed that ko compression of Pisa clay, a soft clay, created more damage to the structure than isotropic compression, as shown in Figure 2.17. The breakdown of structure of the clay under the deviatoric strains involved in the ko compression is faster than that generated under isotropic conditions, where the deviatoric component of strains is much lower. Amorosi (2004) found that Vallericca clay, a stiff clay, destructured more during isotropic compression than during ko compression, implying a reduction of the normalised size of the State Boundary Surface of the clay.
Jardine & Smith (1991) observed that isotropic
compression of Bothkennar clay led to a more ductile response and a larger stress ratio than ko compression (Figure.2.18). Most of the models for structured soils use different proportion of volumetric and shear strains in the destructuration law (e.g. Kavvadas & Amorosi, 2000). Baudet & Stallebrass, (2004) suggested that the plastic shear and volumetric strains are of equal importance in influencing the degradation of structure.
Tests on Vallericca clay (Amorosi & Rampello, 1998) and Bothkennar clay (Smith et al. 1992) showed that the destructuration process is more significant in samples sheared drained than undrained, because during the drained stress paths plastic volumetric strains add to plastic deviatoric strains causing a faster progressive collapse of the natural structure under drained conditions. 2.2.5
Destructuration in swelling
Leroueil & Vaughan (1990) pointed out that swelling might cause changes to the structure of some soils through disruption of interparticle bonding and yield, similar to that induced by compression to very high pressures.
Calabresi & Scarpelli (1985) investigated the behaviour in compression and shearing of two Italian clays, Todi clay and Ancona clay. They noticed that the compression curves of both clays showed that the yield stresses of the samples that had been swelled before re-compression were lower than the yield stresses of the intact samples. Swelling did not show, however, a significant influence on the shearing behaviour of either clay and the normalised stress paths of the swelled samples plotted together with the normalised stress paths of the intact samples.
Takahashi et al. (2004) identified similar behaviour for natural London Clay. They tested samples from Heathrow from two depths, 26m and 37m, in the shear box. The samples that were swelled before shearing showed that the swelling index increased with swelling (Figure 2.19a), suggesting a process of destructuration of the material. However, the influence of this destructuration on the shearing stress paths was not very evident and the peak strengths of the samples swelled before compression plotted together with the peak strengths of the intact material (Figure 2.19c). In the normalised stress-strain response there was no difference between the behaviour of intact samples and samples swelled before re-compression at 37m depth, whereas swelled samples from 26m depth showed a slightly lower strength than intact samples (Figure 2.19b). Jardine et al.
(2004) observed that the difference in lithological units influenced the behaviour in swelling of London Clay samples (see Chapter 3). 2.2.6
Effects of weathering
Evidence of structural degradation has been found for clays subjected to weathering
responsible for the destruction of diagenetic bonds developed during the geological history of clays and this was confirmed by Chandler (1972) for Lias clay, Chandler & Apted (1988) for London Clay, and Cafaro & Cotecchia (2001) for Pappadai clay.
Weathering processes are often recognisable by colour changes of the clay at shallower depths, (e.g. from blue-grey to brown in London Clay and from grey to yellow in Pappadai clay) due to oxidation processes that turn ferrous to ferric oxide. Chandler (1972) observed that the relative proportion of ferrous to ferric oxide can be used to identify the degree of oxidation, the ratio Fe2O3 /FeO being higher for clay strata subjected to higher oxidation. He emphasised that loss of carbonate content and variation of the metallic elements were also indicative of the degree of weathering, although the change of fabric and the distribution of the fissures in the weathered stratum were more representative of the degree of weathering. Chandler & Apted (1988) noticed that London Clay did not show a significant change of mineralogy due to weathering, apart from the loss of carbonate content, but the density of fissures increased dramatically in the weathered material (Figure 2.20). In the weathered stratum of Pappadai clay, Cafaro & Cotecchia (2001) observed only small changes in the mineralogy of the clay, with an increased presence of calcium and iron sulphates, but weathering modified the fabric of the clay, turning to random domains the orientated domains of the fabric of the un-weathered clay.
In Lias clay and London Clay, Chandler (1972) and Chandler & Apted (1988) identified four main zones of weathering, and observed that the depth affected by weathering depended on different factors, which included the presence of overburden material. In both clays a particular feature of weathered zones was
the occurrence of ‘lithorelics’, fragments of relatively un-weathered material set in a matrix of more weathered clay. These fragments retained a water content that was lower than the surrounding material. The water content was therefore assumed to be a fundamental parameter in the investigation of the effect of weathering on the mechanical behaviour of clays. Chandler & Apted (1988) correlated the variation of water content with the variation of strength at different levels of weathered materials (Figure 2.21) and found that weathering increased the water content of London Clay by 4% above that expected if swelling only was involved. Similar behaviour was observed in the Lias clay. Chandler (1972) and Chandler & Apted (1988) noticed that the swelling line through the in situ point of the weathered material intercepts an overburden pressure lower than the overburden pressure of the un-weathered material and concluded that weathering removed some influences of previous consolidation (Figure 2.22). They interpreted the loss of strength of the weathered strata in terms of a reduction of cohesion intercept c’, as shown in Figure 2.23 and observed that the stress path changed with increasing degree of weathering, although the weathered clay still seemed to retain its structure in comparison with the reconstituted material (Figure 2.23a). Cafaro & Cotecchia (2001) found similar behaviour for samples of Pappadai clay. The State Boundary Surface of the weathered clay plotted above the intrinsic SBS of the reconstituted material, but inside the SBS of the un-weathered material, indicating destructuration due to weathering (Figure 2.24a). The compression curves plotted in the v-logp’ plane show that the weathered clay underwent gross yield at lower stresses (Figure 2.24b) indicating weaker bonding and a lower Stress Sensitivity Sσ.
2.3 Large strain strength Based on Skempton work on residual strength (Skempton, 1964), Lupini et al. (1981) introduced the idea that post-peak, soils with lower proportion of platy clay minerals show turbulent shear behaviour, while soils with high content of platy clay minerals show sliding shear behaviour. Based on an analysis of a range of British clays, Vaughan et al. (1978), proposed the plasticity index Ip as a useful parameter to divide soils having turbulent or sliding behaviour. They
observed that for soils with Ip 30% the shearing behaviour was sliding (Figure 2.25). Jardine et al. (2004) pointed out that, although there are many examples of soils that fail to respect this relationship, as other parameters are also involved, in general terms the undrained strength of low plasticity clays is controlled by the water content, as these soils undergo turbulent shearing and their behaviour in an overconsolidated state is basically ductile. Plastic clays, instead, undergo sliding shearing, so their peak undrained stress is controlled by the initial stress before shearing and their shearing behaviour in a overconsolidated state is generally brittle (Figure 2.26, Jardine et al. 2004). The brittleness is thought to be due to the presence of bonding, as this increases the peak strength of the material, but has no influence on the large strain strength.
Burland (1990) defined “post-rupture strength” as being the post-peak strength of stiff clays, which is believed to be that remaining after breakage of interparticle bonds (Figure 2.27). Burland et al. (1996) observed that the postrupture strength envelope tends to lie close to the intrinsic Critical State Line of the reconstituted material. In stiff clays, the localization of strains is thought to be the consequence of the strain softening, and Sandroni (1977) observed that for London Clay the localization of strains coincided with a non-uniform distribution of the pore pressure (Figure 2.28).
Jardine et al. (2004) highlighted that the engineering performance of clays is strongly influenced by their ductile or brittle behaviour and summarised the main factors that determine it, such as stress history, formative history, microstructure, rate effects and fabric.
2.4 Yielding behaviour The yielding of soils is a progressive process, as reflected by the non-linearity of stiffness. In order to simulate this progressive process, several models have been suggested that account for yielding involving multiple kinematic surfaces
(Al-Tabbaa & Muir Wood, 1989; Stallebrass & Taylor, 1997, Simpson 1992; Jardine 1992).
Jardine (1992) proposed a general framework of soil behaviour, consisting of defining three main zones of stress-strain response.
He recognized three
surfaces, named Y1 , Y2 and Y3 , surrounding the stress state at which the soil is located. Y1 and Y2 are kinematic surfaces, and therefore move together with the stress point, while Y3 is thought to remain comparatively immobile and is only affected by large strain events. The multiple yield surfaces are sketched in Figure 2.29. 2.4.1
This is the limit of the zone of linear elastic response. Within this region the strains are linearly dependent on the stresses applied and the load-unload stress paths are expected to coincide giving fully recoverable strains. The soil mass is modelled as elastic particles linked by elastic contacts (Jardine 1992), the particle contacts are therefore thought to be relatively unchanged, with no movements occurring between them. In the past, the resolution of the strains did not allow a region of truly elastic behaviour to be resolved and Porovic & Jardine (1995) noticed that in some soils energy dissipation occurs as dynamic tests are performed, thus the behaviour might not always be fully recoverable.
The magnitude of the strains measured when the Y1 surface is engaged, is very small. The maximum strains might range from about 0.002%, for reconstituted soils (Rolo 2003), to around 0.006% for natural, structured and lightly cemented materials (Jardine 1995; Cuccovillo & Coop 1997; Rolo 2003), but this limit can be much higher for well cemented materials.
The strain rate
marginally influences the linear region, which increases as the strain rate increases, although the effect of the increase is only about 10% per log strain rate increment (see Section 2.7). Dynamic tests show strain thresholds larger than those measured in static tests, while strain rates comparable with residual creep rates are thought to reduce the elastic response threshold (Jardine 1985, 1992). Shear wave laboratory tests have demonstrated that soil behaviour may be
anisotropic within the elastic region. Rolo (2003) observed that the elastic surface of Bothkennar clay at 6m depth is about 2kPa in diameter and it is eccentric with regard to the current effective stress. 2.4.2
This is the contour of a zone of non-linear stress-strain behaviour, where plastic strains are produced and the load-unload stress path reversal is hysteretic. Jardine (1992) attributed the loss of energy in the hysteretic loop to small-scale yielding and fretting at the inter-particle contacts, subjected to normal and shear loading. Studies on clays (Jardine 1992, Smith et al. 1992; Georgiannou, 1998) showed that, although there is hysteresis in the Y2 region the behaviour is fully recoverable. Jardine (1995) therefore suggested that the limit of the fully recoverable zone could be mapped by sets of load-unload drained stress paths, which should be seen to close until the Y2 surface is reached. Once Y2 is engaged, at a strain ε crit , the ratio of plastic strain to total strain increases systematically preventing the hysteretic loop closing, as shown in Figure 2.30 for Bothkennar clay. Kuwano (1999) found that, in sands, the hysteresis loop fails to close as soon as the stress state passes Y2 as irreversible plastic strains develop. Both clays and sands show a sharp change in the direction of the strain vector when reaching the Y2 surface (Smith et al., 1992, Kuwano 1999), which could provide a more uniform means to define the limit of the Y2 zone. In his strain-based criterion to delineate the boundary of the Y2 area, Jardine (1985) assumed the critical strain ε crit to correspond to the larger value between the axial and radial strains. Values of ε crit of 0.01% were reported for Magnus clay, 0.04% for intact London Clay, 0.005% for reconstituted materials, up to 0.03% for uncemented natural soils and around 0.07% for cemented materials (Jardine, 1985; Georgiannou, 1988; Smith et al., 1992; Kuwano, 1999). Simpson et al., (1979) also used a strain-based criterion to characterize the contour of the small strain region but defined it using generalised strain.
Burland & Georgiannou (1991) and Puzrin & Burland (1998) developed energy-based empirical criteria for defining the small strain region corresponding 22
to Y2 . Burland & Georgiannou (1991) used values of incremental strain energy that assumed fixed energy values of zero at the in situ stress state and considered changes of strain energy for each stress path outgoing from the in situ state. They defined the incremental strain energy ∆U as: ∆σ’a.∆ε a+2∆σ’r.∆ε r=∆U
where ∆σ’a. and ∆σ’r. are the axial and radial stress increments and ∆ε a and ∆ε r are the axial and radial strain increments.
The authors identified contours of constant incremental strain energy around the in situ stress state, which are scaled up compared to the values of the total strain energy but are similar to them in shape. Puzrin & Burland (1998) postulated that the boundary of Y2 was defined by a contour of constant incremental strain energy such that the work done by the increments of stresses to reach it along any stress path is constant. 2.4.3
The zone within Y3 is defined as the area of irrecoverable plastic strains that become more significant approaching the limit of this area, which is defined by a sharp change in the stress-strain response. In Zone III, the soil particles are thought to slide relative to each other. The limit of Zone III occurs at strains that increase with OCR (Hight et al., 1987) and this limit is also defined as the Local Boundary Surface (LBS). It coincides with the boundary that Gens (1982) identified inside the State Boundary Surface (SBS). He observed that undrained stress paths of reconstituted samples consolidated with different k ratios traced a series of “elastic boundaries” that undrained stress paths would not cross (Figure 2.31). Normally consolidated samples could move from one LBS to another by developing large volumetric strains along drained stress paths that were directed outside the current LBS.
The Y3 surface is relatively immobile and its orientation is influenced by the isotropic or anisotropic nature of the soil. Leroueil & Vaughan (1990), noticed 23
that for initially isotropic soils the yield surface was centred about the isotropic axis, but for anisotropically consolidated soils, the surface was centred about the k=σ’h /σ’v line. The Y3 surface can be interpreted by identifying changes in tangent stiffness and direction changes of effective stress paths in undrained tests and changes in strain increment direction for drained tests.
For all normalisable soils an outer State Boundary Surface is defined in normalised stress space using the equivalent pressure p* e. In materials with a stable structure, such as reconstituted clays, this surface could be reached by drained stress paths that will either stay on it or travel along it. In soils with metastable structures, as mentioned in Section 2.2.2, the stress path might be redirected inwards as it travels towards the intrinsic state (Baudet & Stallebrass, 2004).
2.5 Small strain behaviour Field measurements of the behaviour of geotechnical structures and the need for improving geotechnical design have highlighted the importance of accurately simulating the behaviour of soils at very small strains (e.g. Puzrin & Burland 1998). Studies presented by Simpson et al. (1979, 1981) of the movements around excavations in London Clay indicated that the strains associated with undrained excavations and foundations in London Clay are generally very small. This highlighted the significance of studying soil behaviour at small strains. 2.5.1
Resolving the strain measurements in the laboratory at strains as small as 0.0001% has demonstrated that in some soils it is possible to identify a region of purely elastic response, corresponding to Zone I defined by Jardine (1992). Inside this region, the response of soil to static and dynamic loads can be predicted by applying Hooke’s law.
In a continuous solid material the stress increment ∆σ is related to strain increment ∆ε trough twenty-one elastic constants, grouped in the “compliance matrix”, which can be written in general terms as: δε xx C11 C12 C13 C14 C15 C16 δσ xx δε δσ yy C 21 C 22 C23 C 24 C25 C 26 yy δε zz C31 C 32 C33 C 34 C35 C36 δσ zz δγ = C C C C C C • δτ yz 41 42 43 44 45 46 yz δγ zx C 51 C 52 C53 C 54 C55 C 56 δτ zx δτ δγ xy C 61 C 62 C63 C 64 C65 C 66 xy
In soils the vertical direction of deposition and loading allow the identification of an axis of symmetry, and the hypothesis of ‘cross-anisotropy’ may therefore apply, where the horizontal is considered as a plane of isotropy.
Cartesian axes with the z axis vertical, Equation 2.11 can be re-written for a cross-anisotropic material as:
δε xx δε yy δε zz δγ yz δγ zx δγ xy
1 E h − v hh Eh − v hv E = h ⋅ ⋅ ⋅
− v hh Eh 1 Eh − v hv Eh
− v vh Ev − v vh Ev 1 Ev
⋅ ⋅ δσ xx δσ yy ⋅ • δσ zz δτ ⋅ yz δτ zx δτ ⋅ xy 1 Ghh (2.12)
where the parameters shown are defined as follows: Ev : Young modulus in the vertical direction; Eh : Young modulus in the horizontal direction; ν hh : Poisson’s ratio for horizontal strains due to horizontal strains; ν vh : Poisson’s ratio for vertical strains due to horizontal strains; ν hv : Poisson’s ratio for horizontal strains due to vertical strains; Gvh =Ghv : shear moduli in the vertical plane; Ghh : shear modulus in the horizontal plane. 25
These parameters are not all independent. Thermodynamic rules require that for an elastic material the compliance matrix must be symmetric (Love, 1927), and therefore:
vhv v vh = Eh E v
Ghh is dependent on Eh and ν hh because the horizontal plane is a plane of symmetry, and therefore:
Eh 2(1 + v hh )
Thus only five independent parameters are required to describe a crossanisotropic elastic material: Ev , Eh , ν vh , ν hh , Ghv and Equation 2.12 becomes:
δεxx δε yy δεzz δγ = yz δγzx δγxy
1 Eh −vhh Eh −vvh Ev
−vhh Eh 1 Eh −vvh Ev
−vvh Ev −vvh Ev 1 Ev
⋅ δσ' xx δσ'yy ⋅ • δσ'zz (2.15) δτ ⋅ yz δτ zx ⋅ δτxy 2(1+vhh) Eh ⋅
The thermodynamic requirements the strain energy has to be positive also imposes boundaries to the independent parameters (Lings et al. 2000):
Ev (1 − v hh ) − 2v vh2 ≥ 0 Eh
G hv ≤
Ev E E 2v vh (1 + vhh ) + 2 v (1 − v hh ) 2 1 − h vvh2 Eh Ev
− 1 ≤ v hh ≤ 1
Ev , Eh , Ghv must also all be positive.
The bounds to the five independent parameters were shown graphically by Pickering (1970) in a 3-D graph, shown in Figure 2.32. The Poisson’s ratios are indicated with the symbol µ in the graph. The ship’s bow shape represents the permitted space within which all the combinations of the drained elastic parameters plot. In Figure 2.32, the plane ABC represents all uncoupled materials, which undergo no distortional strain with isotropic loading, or volumetric strain with deviatoric loading. The line CD in this plane represents isotropic materials. The plane tangential to the bounding surface on the line AB represents all incompressible materials, so that incompressible elastic materials lie on line AB. Ideal undrained materials deform at constant volume and such materials must therefore lie on the line AB. This line results from the intersection of the uncoupled and the incompressible planes, so that ideal undrained materials are both uncoupled and incompressible. Any set of drained parameters can be converted to a set of undrained parameters as all the points within the permitted space can be mapped onto the undrained line. The points on the undrained line, though, can be reached from an infinite number of drained points within the space and therefore the mapping can only be performed from drained to undrained parameters.
The five independent parameters can be measured in triaxial tests by choosing stress paths that allow simplification of Equation 2.15. In triaxial tests no shear stresses τyz, τzx, τxy can be applied to a cross-anisotropic material and the conditions of the triaxial cell impose that:
dε xx =dε yy =dε r ; dσ’xx =dσ’yy =dσ’r and dσ’zz=dσ’a
Equation 2.15 therefore simplifies to:
1 δε a E v δε = − v r vh E v
− 2vhv ' Ev δσ a ⋅ ' 1 − v hh δσ r Eh
The variables can also be written in terms of triaxial parameters: 1 δε v K = 1 δε s J pq
1 J qp δp' ⋅ 1 δq 3Geq
where K is the bulk modulus, Geq is the equivalent shear modulus and Jqp is the coupling modulus linking changes in deviatoric stress to changes in volumetric strain, while Jpq is the coupling modulus linking changes in mean effective stress to changes in deviatoric strain (Lings et al. 2000). For an elastic material the compliance matrix has to be symmetric, and therefore Jqp =Jpq =J. For an isotropic material there is also no coupling between distortional and volumetric behaviour, and therefore 1/J=0. (a) Shear modulus The simplest method to measure the shear modulus at very small strains is the bender element technique. It was developed by Shirley & Hampton (1977) and consists of measuring the time taken for a shear wave to propagate through the sample. Two piezoceramic plates protrude into the soil specimen on opposite sides. One plate, the transmitter, is excited by means of an applied voltage causing it to vibrate in the direction perpendicular to the plane of the plate, producing a shear wave that propagates through the sample. The plate on the other side of the sample, the receiver, detects the wave arrival and generates a small voltage that is displayed on a digital oscilloscope together with the wave sent. The time difference between the sent wave and the received wave is used to calculate the velocity of the shear wave vs,
where d is the distance between the plates and tarr is the arrival time. The elastic shear modulus Gmax in a defined direction is then calculated from: Gmax=ρvs2
where ρ is the total mass density of the soil.
Depending on the orientation of the plates and on the direction of the shear wave propagation the shear modulus in different directions can be evaluated and therefore the anisotropy of Gmax within the soil measured. Figure 2.33 shows a sketch of the possible configurations of the bender elements. The shear moduli are given a double suffix, the first referring to the direction of propagation of the wave, the second referring to the direction of polarization or particle motion.
Jovicic & Coop (1998) pointed out that, in studying anisotropy, bender elements are a useful means, but that it is important to identify the kind of anisotropy studied and the limit to these measurements imposed by the configuration of the bender element system. The authors distinguished between inherent or strain-induced anisotropy, which is related to the strain history of the soil that is geological or arising from consolidation, and stress-induced anisotropy, which results from the anisotropy of the current stress conditions. The inherent anisotropy can only be separated from the stress-induced components in an isotropic stress condition.
The configuration of the triaxial
apparatus however is axi-symmetric, and when an anisotropic state of stresses is created a cross-anisotropic pattern of strains is created, having the horizontal plane as a plane of isotropy and anisotropy developing only in the vertical plane. Shear waves that propagate along the direction of the axis of symmetry can therefore measure stress-induced effects of anisotropy, but are unable to measure the strain-induced anisotropy in the sample.
In investigating inherent anisotropy
the bender elements should therefore be mounted in the plane of isotropy (Jovicic & Coop, 1998). Commonly the bender elements are mounted in the top and bottom platens of specimens for convenience. Jovicic & Coop (1998) therefore trimmed their samples horizontally and vertically to investigate the inherent anisotropy of London Clay (b) Interpreting bender element signals The interpretation of the bender element signal represents the main difficulty of this technique. Several interpretation methods have been proposed in the literature, but only the two methods used in this research will be reviewed here. These methods gave good agreement and no further investigation of this topic was needed in this project. (i) First Arrival Method This method assumes a plane wave front and the absence of any reflected or refracted waves. The arrival time is estimated as the time between the start of a single shot pulse input to the transmitter element and the first deflection in the output signal from the receiving bender. Single shot square or sine waves can be used, but Viggiani & Atkinson (1995) demonstrated that using a sine wave reduces the subjectivity in interpreting the signal, because the square wave is composed of a spectrum of different frequencies, which makes the interpretation more difficult.
The trace of the received signal is often characterized by an
initial downward deflection called the “near field effect”. This results from the spreading of the wave front and coupling between waves that exhibit the same particle motion but propagate at different velocities (Jovicic et al., 1996). It is more evident at low frequencies where it obscures the true arrival point, creating difficulties in the interpretation of the signal.
Using a sufficiently high frequency
eradicates the near field effect and the arrival time can then be read on the oscilloscope and should not change with frequency, so it can be estimated from the average reading for a set of frequencies. (ii) Phase Velocity Method This consists of sending a continuous wave at different frequencies and analysing the point when the transmitted and the received waves are in phase and 30
exactly 180o out of phase. This corresponds to identifying the maximum crosscorrelation between the signals recorded at two points in space, making the hypothesis of a plane wave and in the absence of reflected and refracted waves. Each half phase corresponds to a mode of vibration in the sample. The number of wave cycles Rd may be written as:
where D is the travel length and λ is the wave length. The frequency f and the travel time t are related to Rd by: Rd=D/λ=f.t
The arrival time can then be derived from Equation 2.24 as the inverse of the slope of the line obtained by plotting Rd against frequency. Viggiani & Atkinson (1995) found that the travel length D is the main source of error in determining Gmax and concluded that D should be the distance between the tips of the bender elements. (c) Other elastic parameters Pennington et al. (1997) and Lings et al. (2000) suggested methods of calculating the elastic parameters by combinations of dynamic and monotonic loading tests. In their work, a ‘three parameters’ calculation was adopted. Pennington et al. (1997) performed stress paths at constant σ’v and constant σ’h that allowed simplification of Equations 2.19 and 2.20 and allowed determination of the independent elastic parameters. From Equation 2.19, Ev and ν vh were directly measured in probes at constant horizontal stress: δσ ' Ev = a δε a δσ 'r =0
δε v vh = − r δε a δσ 'r = 0
The other parameters can be derived from probes at constant vertical stress 31
δσ r' Eh = (1 − v hh ) δε r
δσ a' = 0
δε 2v hv = − a (1 − v hh ) δε r
δσ a' =0
By introducing the parameter Fh , where:
Fh =Eh /(1-ν hh ),
and measuring Ghh with bender element tests, the combination of Equations 2.29 and 2.14, allows Eh to be defined as:
4 Fh Ghh Fh + 2Ghh
From probes at constant mean effective stress and constant deviatoric stress, the bulk modulus K, the equivalent shear modulus Geq and the coupling modulus J can be measured. Lings et al. (2000) also derived equations linking K, Geq and J to the five independent parameters:
3 1 + 2vvh 1 − v hh + 4 2E h Ev
1 1 − 4v vh 1 − v hh +2 Ev Eh 3 1 − vvh 1 − vhh 2 − E Eh v
The formulation proposed above is valid for drained probes, but Lings (2001) also derived the elastic parameters from undrained probes and related them to those derived from drained probes.
Influence of recent stress history
Atkinson et al. (1990) and Stallebrass & Taylor (1997) observed that the stress-strain response of overconsolidated clays depends on their current state and on the loading history followed to reach that state, in particular the relative directions of the current and previous loading paths. Atkinson et al. (1990) defined “recent stress history” as the current path load undertaken by the soil in relation to the previous stress path, which might have the form of a stress path direction change or an extended period of rest. Atkinson (1973) observed that the stiffness in triaxial and plane strain tests increased as a consequence of a sudden change in the direction of the stress path. The author noticed that in conditions of axi-symmetry and for loading paths that remained inside the boundary surface, two samples, loaded undrained in the same direction from the same stress point, had different stress-strain responses if they had reached the stress point by following different stress paths. The response was stiffer for samples having a higher angle of rotation from the previous stress path (Figure 2.34). Costa Filho (1984) observed that for samples of London clay the stiffness depended on the consolidation stress path followed. He performed unconsolidated undrained tests starting from isotropic stress states and undrained tests on samples consolidated to their in situ stress state. The samples consolidated to their anisotropic in situ stress state showed a stiffer response than samples sheared from an isotropic stress state as the reversal in the direction of loading increased the stiffness of the material.
Som (1968) noticed that the compressibility of samples in oedometer tests reduced following a rest period under constant load.
Inspired by these
observations Atkinson et al. (1990) carried out a set of stress probes on reconstituted samples of London Clay. Their stress probe programme is sketched in Figure 2.35. They performed probes at constant p’ in compression and extension and constant q in compression and swelling starting from isotropic stress states to which the samples had been taken by following different approach stress paths. The approach stress paths and the probes were about 90kPa in length. This length seemed to the authors to be sufficient to observe a behaviour that was only influenced by the direction and not by the length of the stress path. 33
Creep was allowed for only three hours before starting the probes, which was long enough so that no volumetric strain changes could be measured. The results of their tests are shown in Figure 2.35 in terms of stiffnesses. The stiffness degradations are curved, indicating an inelastic behaviour even at small strains. The small strain stiffnesses depended on the angle of rotation, being the stiffest for 180o degree rotations and the least stiff for an angle of 0o . For angles of rotation of +90o and –90o the stiffnesses plotted in between. The influence of the rotation angle was not distinguishable after 0.5% strain where all the curves converged to a unique value.
The resolution of the strain measured by Atkinson
et al. (1990) was only about 0.001%, which was good at the time when the experiments were carried out. However, their results should be considered in terms of the degradation of stiffness for the range of strains they presented. Only the tests with an angle of rotation of 0o showed an apparently constant and lower initial stiffness, whereas for other angles of rotation there seemed to be a degradation trend perhaps from similar initial values.
Atkinson et al. (1990) stated that the recent stress history should be included amongst the factors that influence the stress-strain response, together with the current stress state and the overconsolidation ratio. The authors characterised the recent stress history by the angle of rotation between the approach stress path and the current loading path. In this definition, the length of the approach stress path and creep may play significant roles as they are directly associated with the stress path in terms of the strains involved. In this sense, the recent stress history assumes the meaning of a “recent strain history” and the lengths of the approach stress paths, which are where the strains are developed by the sample, become significant.
Jardine (1992) performed a set of tests on London Clay starting from isotropic and anisotropic stress states and investigating the stiffness of the soil and the size of the kinematic surfaces. He noticed that the Y2 surface, which was assumed to coincide with a critical shear strain ε crit =0.04%, did not change in size, but changed in orientation, as a result of the different consolidation paths followed. In particular Jardine compressed intact and reconstituted London Clay samples to
their initial in situ stress points following different stress paths, before shearing to failure, as shown in Figure 2.36. The author found that where there was no reloading stage, the undrained stiffness in compression, Ec at 0.01% strain was higher than the undrained stiffness in extension Ee at the same strain, giving a Y2 surface more elongated in the direction of compression than in extension. Heavily reloaded samples were stiffer in extension than in compression at the same strain of 0.01% and the Y2 surface was more elongated in the direction of extension. Jardine also observed that at 0.1% strain the difference in stiffness reduced significantly and, at large strains, the influence of the recent stress history was erased, in agreement with the observations of Atkinson et al..
Lings et al. (2000) observed that in the heavily consolidated Gault Clay, the approach stress path and the angle of rotation influenced the elastic parameters of the material and their degradation.
They performed two sets of tests (Figure
2.37) on natural samples including different angles of rotation of the approach stress path. The probes did not start from the same stress points and the samples were not taken back to the initial stress point, but followed long stress paths that moved away from the initial stress point. The results were therefore normalised for the respective initial stresses. Creep was allowed before each probe started and the long stress paths included the development of large strains. The stiffnesses and the Poisson’s ratios are plotted against strains in Figure 2.37. The degradation of values seems to depend on the angle of rotation, but, surprisingly, the angle of rotation seems also to influence the elastic values, which, by definition, should be independent of recent stress history.
Clayton and Heymann (2001) performed a set of tests on Bothkennar clay as shown in Figure 2.38. Their undrained shear probes were about 9kPa in length and started from an initial isotropic state to which the sample had been consolidated following three different approach directions (DA, CA and BA in Figure 2.38). The strains involved during the approach stress paths were not reported, but the length in term of stress was about 10kPa and the authors believed this stress path to be sufficiently long considering that for isotropic loading the state boundary surface of Bothkennar clay would have been at about 40kPa.
Creep was allowed before each shear probe until axial and volumetric 35
strain could no longer be measured. The first two shear probes (AB in Figure 2.38) used +90o and –90o angles of rotation from the approach stress paths DA and CA respectively. The maximum axial strains reached during the two probes were about 0.08%. The third shear probe used an angle of rotation of 180o from the approach stress path BA in Figure 2.38 and took the sample to failure. As shown in Figure 2.38, no difference in the stiffness curves from all the shear probes was noticed, regardless the different approach stress paths. To the authors this result demonstrated that the recent stress history had no influence on the soil response if creep was allowed.
They believed that creep might erase any
memory of the approach stress path, so that the time spent at constant stress became the recent stress history for the material.
It is possible that in the tests performed by Clayton & Heymann (2001) the approach stress path was not long enough in term of strains to see the effects of recent stress history, even if it seemed sufficient in term of stresses. Smith et al. (1992) found the diameter of the Y2 surface of Bothkennar clay to be about 10kPa corresponding roughly to 0.02% axial strain. If the stress path had moved outside the Y2 surface, this might have caused large strains to develop so that recent stress history effect might have been seen. However this was not the case for the approach stress paths used by Clayton & Heymann (2001), which kept the sample within Y2 . Clayton & Heymann (2001) also performed shear probes in compression and extension on Bothkennar clay and London clay samples, starting from their corresponding in situ stresses. Creep was allowed at the in situ stress points that were approached from the same direction in each probe. Figures 2.39 and 2.40 show sketches of the approach stress paths, the probe directions and the test results for both clays. For the Bothkannar clay, for which ko 1, the compression path rotated by 150o from the approach stress path direction and the extension path rotated by 30o . Again, the initial stiffness in both cases was the same, but the stiffness degradation was faster in extension than in compression as the compression path moved the soil away from the failure line. Jardine (1992) had already noticed that, for Bothkennar clay, although the overall size of the Y2 surface did not change greatly, the presence of the failure line restrained the extension of Y2 . Clayton & Heymann (2001) concluded that, since the initial stiffness of both Bothkennar clay and London clay did not change for the different stress path rotations, only the outgoing stress path was relevant to the soil response if creep was allowed.
In a truly linear elastic range, however, the stiffness should be independent on the sign of loading (Kuwano, 1999, Rolo, 2003) and both tests shown in Figures 2.39 and 2.40, the elastic stiffness of the soils should be expected to be independent of the angle of rotation from the approach stress path. However, in both cases, the rate of degradation was faster when the stress path moved towards the failure line. This made Clayton & Heymann (2001) conclude that only the outgoing stress path influenced the soil response, but in fact, from the tests performed, it is not clear whether the rate of degradation was influenced also by the angle of rotation as well as by the outgoing direction. The two effects were not distinguishable and could have overlapped preventing any clear conclusion being made.
In investigating the influence of the recent stress history on the stress-strain response, the stress path rotation and the creep should therefore be isolated from other parameters that influence this response, such as the outgoing stress path and the current stress state. When loading from an anisotropic stress state soils show a stiffer response in the stress path that moves away from the failure line and Jardine (1992) noticed that the size of Zone II is restricted by the proximity of the Bounding Surface. The outgoing loading direction might therefore interact with the effects of recent stress history, confusing the real cause of the observed stress-strain response. Stallebrass & Taylor (1997) developed a model that allows the replication of the main features of soil behaviour by modelling elasto-plastic deformations
conventional modified Cam-Clay model. This model extends the bubble model developed by Al-Tabbaa (1987) by introducing a second kinematic surface that takes into account the recent stress history of the soils.
2.6 Creep Creep may be defined as plastic strains that occur under constant effective stress and is one of the most important processes of ageing. Tatsouka et al. (1998) distinguished long term and short term ageing effects, including lithification and weathering on both the geological timescale and on the civil engineering timescale. Lithification may consist of either an increase or decrease in the strength and stiffness of soils. Geologically, it incorporates diagenetic effects, due to intrinsic physical and chemical processes or due to external processes such as desiccation, earthquake, loading or compaction. On an engineering scale, it might be associated in negative terms with destructuration, due to weathering associated with the exposure of soil after excavation or erosion.
Vaughan (1997) noticed that the response of Bothkennar clay in one dimensional compression was stiffer after ageing and suggested that the behaviour of soil after ageing is similar to that sketched in Figure 2.41.
loading rate after ageing is slow, the reloading curve yields on the curve for continuous loading. If the loading rate is high the reloading curve yields at higher stresses than on the curve for continuous loading, showing a temporary overshoot. Vaughan, (1997) called the higher yield observed after ageing “ageing preconsolidation”.
Tatsouka et al. (1998) observed that the soil response becomes stiffer as the ageing period increases. They reported a number of tests on different soils investigating the effects of ageing (drained creep) in shear and defined a postageing stress-strain relationship as shown in Figure 2.42. It consists of three types of behaviour. In the first type the stress-strain curve after ageing re-joins the original primary loading relationship without exhibiting overshooting. In the
second type the reloading stress-strain curve rejoins the original primary loading curve after having exhibited a temporary overshoot. In the third case the stressstrain curve does not re-join the primary loading relationship and exhibits a persistent overshoot, with a noticeably larger peak strength than that obtained from the original primary loading.
The overshooting behaviour, observed in many soils after prolonged ageing, has been attributed to structuration effects sometimes associated with the development of bonding. Mitchell (1960) defined this behaviour as “thixotropic hardening. In loose soils, such as clean sands, no bonding is likely to form, but the interaction of acceleration effects and particle crushing should be taken into account.
Figure 2.43 shows two sets of tests on reconstituted samples of Fujinomori clay (Tatsouka et al., 1998). The first set of samples was sheared undrained at a rate of 0.05%/min from an isotropic state to failure. The second set of samples was sheared undrained at the same rate to an anisotropic stress state corresponding to ko =0.5, then allowed to creep for two days and re-sheared undrained to failure. When the undrained shear restarted a zone of high stiffness was seen before a clear yield was observed. The size of this zone was approximately proportional to the pressure level at which the specimen was aged. Similar tests were performed on natural samples of Vallericca clay (Figure 2.44, Tatsouka et al., 1998). The samples were sheared undrained at a rate of 0.009%/min to anisotropic states A and B in Figure 2.44 and aged until the creep axial strain rates became 3x10-5%/min and 1x10-4%/min respectively. As for the Fujinomori clay a high stiffness zone was noticed when the loading was restarted, although in this case the effect on the subsequent stress-strain response seemed to be minor. The author attributed this difference to the lower strains developed by the Vallericca clay during ageing, but a lower shearing rate was also used for the natural samples of Vallericca clay.
The rate of loading used in the tests described above was quite high for both clays and some effects of undissipated or non-uniformly distributed pore water pressure might play a role in the stress-strain response. The effect of creep and 39
the size of the high stiffness zone seemed smaller for the more permeable Metrano silty sand and also for a soft sedimentary sandstone tested by Santucci de Magistris (1998), especially when the rate of reloading was reduced (Figure 2.44).
Sandroni (1977) highlighted the importance of pore pressure dissipation in natural soils that show localization of strains, as these control the pore pressure response. He performed creep tests on a specimen of London Clay, of low permeability, which was sheared at a fast rate (0.2%/min). The shearing was interrupted for 24h without allowing axial rebound and restarted at the same rate. The stress-strain curve of the second stage (Figure 2.45) plotted below the first shear stage. The author attributed this to the equalization of pore pressure that would have occurred between the shear zone and the rest of the specimen during the 24h rest period. 2.6.1
Effects at small strains
For very small strain stiffness the effect of ageing is thought to increase the shear modulus Go and the Young’s Modulus Eo measured in the course of continuous shearing where the structure is continuously changing. Jovicic & Coop (1997) noticed that in Dog’s Bay sand, a low density, poorly graded carbonate sand, the stiffness increased up to 15% over about five days due to ageing phenomena, as shown in Figure 2.46a where the stiffness variation is normalised with respect to the stiffness Go at zero time. The effect of the stiffness increases with time was higher at lower confining pressures and for first loading, while it was less evident as the confining pressure increased and in stages of unloading or reloading.
Tatsuoka et al. (1998) suggest that for sands a re-
structuration takes place due to rearrangement of the inter-particle contacts with time causing the stiffness to increase. The rate of change decreases with time (Figure 2.46b).
Atkinson et al. (1990) included the time spent at a constant stress before loading as a feature of recent stress history and Richardson (1998) pointed out
that the creep adds to the effects of the stress path direction and can be investigated independently.
After prolonged ageing or creep the size and the shape of the kinematic surfaces Y1 and Y2 become less dependent on the previous stress history (Jardine 1985; Jardine et al., 1991; Tatsuoka et al., 1998). Figure 2.47 shows schematically the development of Y2 with the recent stress history. Y2 is thought to re-centre around the current stress state when creep is allowed for a long time, whereas if the creep period is short the effect of the recent stress history becomes more important.
2.7 Strain rate effects The rate dependency in continuous shearing reflects the creep seen under constant load. The overshooting effect noticed after creep has been observed also immediately after a step change in constant strain rate in the course of otherwise monotonic loading. The high stiffness region after creep therefore can be included in a more general framework of the effects of a loading rate change, as it can be thought of as the consequence of an increase of strain rate.
There seems to be a fundamental difference in behaviour between stiff clays and granular materials. Clays seem to follow an isotach model, where the stress is uniquely defined by strain and strain rate (q=f(ε ,dε /dt)). Whereas the behaviour of granular materials seems to be dominated by a temporary effect of change in stress with strain rate, and as the strains increase, the stress-strain curve tends to converge towards a unique stress-strain relationship, independent of the strain rate. The temporary over/undershooting effects increase with strain for granular materials and also became dominant for the stress-strain behaviour of clays at large strains (Jardine, 2004). Tatsuoka et al. (1998) presented a large number of tests on different soils showing that the deviatoric stress suddenly changes due to a step increase or decrease of strain rate. Figures 2.48-2.50 show a few examples from tests on Vallericca clay, dense Chuba Gravel and Hostun sand sheared drained or undrained. In Hostun sand (Figure 2.50), the effect of the
strain rate on the stress-strain relationship was negligible for constant strain rates over a range of 500 times. The overshooting and undershooting behaviour disappeared with further straining and the stress-strain relationship rejoined that which was independent of the strain rate.
The strain rates used to investigate strain rate dependency often seem to be quite high, especially for clays, and therefore undissipated or non-uniformly distributed pore pressures could affect the data. Overshooting and undershooting effects are more evident in sand or dense gravel, where drainage and pore pressure distributions are less problematic. Sandroni (1977) found that the measurements of the actual pore pressure became problematic at high rates of shearing and even the mid-height probe was not free from these effects. From slow and medium rate shear tests on London Clay Sandroni (1977) noticed that up to the maximum pore pressure, the behaviour of London Clay was insensitive to the rate of shearing, but at faster shearing rates the undrained strength of the soil increased. He observed that the failure strains tended to decrease with increasing rate of shearing, but the strains at the maximum were unaffected by the rate of shearing. The author observed that the behaviour of the clay was influenced by the negative pore pressures generated on the shear plane, and at faster shearing rates the time for equalization of pore pressure was smaller. Cotecchia (1996) observed that for the compression behaviour of Pappadai Clay there was a threshold, above which the strain rate dependency became evident.
At very small strains the elastic stiffness seems to be hardly influenced by the strain rate. Figure 2.51 shows the effect of the strain rate on the Young’s Moduli for different materials. The small strain stiffnesses as of hard rock, clean sands and gravel are insensitive to the strain rate and the small strain stiffnesses as of soft rock, silty sand, soft and stiff clays depend marginally on the strain rate. Figure 2.52 illustrates the effects of the strain rate on the Young’s modulus Eo of Vallericca clay. At very small strain rates there seems to be a strain rate dependency, but this becomes smaller as the strain rate increases. Jardine (1992) pointed out that a soil that is still creeping at the end of a loading stage will display a strain rate effect, even at very small strains. In order to avoid confusing interactions, the initial creep rate experienced by the sample as a result of the 42
previous loading should therefore be negligible before starting a new loading stage.
2.8 The influence of fissures Some materials have natural discontinuities, which are often the result of stress release arising from some geological processes (Skempton et al., 1969). Fissures seem to increase the permeability of the material, even if they are apparently closed and are thought to be responsible for the so-called ‘samplesize’ effect on strength measurements. Bishop & Little (1967) observed that the strength envelope of London Clay from Wraysbury measured for 0.75” (19mm) diameter specimens plotted above that measured for 1.5” (38mm) diameter specimens (Figure 2.53). Similar results have been shown by other authors for different clays. Marsland & Butler (1967) carried out a test programme on Barton Clay specimens with diameters between 1.5” and 5” (127mm). They selected “intact lumps” of 1.5” diameter and showed that the strength of these samples was higher than that of the 1.5” samples containing fissures, which, in turn, was higher than the strength of 5” diameter specimens (Figure 2.54). Skempton & Petley (1967) related the difference in strength to the lower effective strength parameters mobilized along the fissures.
Walsh (1965) showed that the bulk modulus of a rock mass with closed cracks was the same as the bulk modulus of an intact mass, but both open or closed cracks affected the elastic parameters of the soil mass.
confirmed that for London Clay the Young’s modulus calculated for specimens of 38mm diameter was 30% higher than that calculated for specimens of 75mm diameter that were 10% higher than that of 125mm diameter specimens. A brief summary of the results from tests on fissured soils was presented by Costa Filho (1984). He pointed out that, although there is a tendency for the moduli of fissured soils measured using small specimens to be higher than those measured using large specimens, this tendency is mainly due to the different initial effective stress.
Costa Filho (1984) showed that when the results were
normalised for the initial effective stress there was no longer any effect.
author considered the normalised secant moduli at 50% maximum deviatoric stress measured by Sandroni (1977) for London Clay samples and by Maguire (1975) for Lias clay (Tables 2.1 and 2.2). Webb (1964) showed that the stressstrain curves corresponding to an intact sample and to a sample that had sheared along a pre-existing fissure were coincident up to 1.5% strain, but the maximum shear strengths of the two specimens were different, being higher for the intact sample (Figure 2.55).
Table 2.1: Normalised secant moduli at 50% maximum deviatoric stress on London Clay samples (Costa-Filho, 1984, data from Sandroni 1977)
Table 2.2: Normalised secant moduli at 50% maximum deviatoric stress on Lias clay (Costa-Filho, 1984, data from Maguire 1975)
Figure 2.1: Structure of the main clay units (Veniale, 1983; Cotecchia, 1996)
Figure 2.2: Classification of fabric (Sides & Barden, 1970)
Structure permitted space NCL * p’ Figure 2.3: Schematic diagram showing enhanced resistance of natural clays in compression
SC curves (Terzaghi, 1941)
SC curves (Terzaghi, 1941)
s*vy=s*vc s’vy=s’vc (a) Sedimentation structure
s*vy=s*vc s’vc s’vy
(b) Post-sedimentation structure
Figure 2.4: Typical compression curves for (a) clays with sedimentation structure and (b) clays with post-sedimentation structure (Cotecchia & Chandler 1997)
sv/s*ve Figure 2.5 Influence of structure: proposed normalising parameters
Figure 2.6: Sedimentation compression curves for normally consolidated clays (Skempton, 1970)
Figure 2.7: The intrinsic and sedimentation compression lines (Burland, 1990)
Figure 2.8: (a) In situ states for normally consolidated clays (Skempton, 1970) and (b) interpretation of the data indicating sensitivity (Cotecchia and Chandler, 2000)
Figure 2.9: Geometrical definition of the strength sensitivity (Cotecchia & Chandler, 1997)
Figure 2.10: Stress and Strength Sensitivity relationships for clays having in situ states on the right of the ICL (Chandler 2000).
Figure 2.11: Stress and Strength Sensitivity relationships for clays having in situ states on the left of the ICL (Chandler 2000).
Figure 2.12: The Sensitivity framework (Cotecchia & Chandler 1997)
Figure 2.13: State boundary surface of reconstituted and undisturbed Pappadai clay consolidated to stresses before yield (Cotecchia, 1996)
(a) Sibari Clay
(b) Boom Clay Figure 2.14: Compression curves of (a) clay with a stable structure (Coop & Cotecchia, 1995) and (b) clay with a meta-stable structure (Burland, 1990)
a:Medium pressure drainedtests b:Medium pressure undrainedtests
Figure 2.15: Normalised SBS for samples compressed before and beyond gross yield: (a) Pappadai Clay (Cotecchia, 1996) (b) Bothkennar Clay (Jardine & Smith, 1991); (c) Valericca Clay (Amorosi & Rampello, 1998).
Figure 2.16: Unique SBS for Pappadai Clay normalised by structure (Cotecchia, 1996)
Figure 2.17: Isotropic and k0 compression for Pisa Clay (Baudet & Stallebrass, 2004, data from Callisto 1996)
(b) Figure 2.18: Destructuration of Bothkennar Clay (a) isotropic and k0 compression (b) shearing behaviour (Jardine & Smith, 1991)
Figure 2.19: Behaviour of London Clay swelled to 1/6, 1/12, 1/16 the initial vertical effective stress (a) destructuration in one-dimensional swelling (b) normalised shear stress-horizontal displacement; (c) stress paths in constantheight direct shear box tests (Takahashi et al. 2005)
Average fissure area (cm)2 0 0
Fissure intensity 10.000
Zone I 10
Figure 2.20: Fissuring in the London Clay (Chandler & Apted, 1988)
Figure 2.21: Variation of water content with different levels of weathered strata (Lias Clay, Chandler, 1972)
Figure 2.22: Idealised relationship between effective overburden pressure and water content during the geological history of an overconsolidated clay (Chandler, 1972)
Figure 2.23: Shear behaviour of London Clay samples at different levels of weathering (a) Undrained triaxial compression tests, (b) normalised stress paths (Chandler & Apted, 1988)
(b) Figure 2.24: Effects of weathering on Pappadai clay (a) Normalised state boundary surfaces of both the natural and the reconstituted samples (b) isotropic and one-dimensional compression behaviour of both the weathered (yellow) and the unweathered (grey) clay (Cafaro & Cotecchia, 2001)
Figure 2.25: Effect of clay particles on the critical state friction angle and on the residual friction angle (Lupini et al. 1981)
Figure 2.26: Idealised undrained shearing behaviour of overconsolidated clays with (a) low plasticity and (b) high plasticity (Jardine et al, 2004)
Figure 2.27: Strength of stiff plastic clays (Jardine et al., 2004)
Figure 2.28: Localization of strains and pore pressure distribution in London Clay (Sandroni, 1977)
Figure 2.29: Scheme of multiple yield surfaces (Jardine, 1992)
Figure 2.30: Definition of Y2 for Bothkennar Clay from drained cyclic tests (Smith et al. 1992)
Figure 2.31: Normalised undrained stress paths for triaxial compression tests on Lower Cromer till samples comsolidated to different values of k (Gens, 1982)
Figure 2.32: Bounds for the elastic parameters and planes and lines representing special types of materials (Pickering, 1970)
Figure 2.33: Configuration for measurement of stiffness of a cross-anisotropic soil under axi-symmetric loading (Pennington et al., 1997)
Figure 2.34: Effect of recent stress history on current stiffness (Atkinson et al. 1990)
Figure 2.35: Stiffness response for tests for recent stress history of reconstituted London Clay (Atkinson et al., 1990) 70
Figure 2.36: Compression paths and small strain stiffnesses for natural and reconstituted London Clay samples (Jardine, 1992)
Figure 2.37: Stress probes and normalised elastic parameters for Gault clay (Lings et al. 2000)
(b) Figure 2.38: Stress probes and stiffness degradation curves for Bothkennar clay (a) schematised stress path (b) stiffness response (Clayton & Heymann, 2001)
(b) Figure 2.39: Stiffness degradation curves of a Bothkennar clay subjected to two different loading paths (a) schematised stress paths (b) stiffness response (Clayton & Heymann, 2001) 74
(b) Figure 2.40: Stiffness degradation curves of London clay subjected to two different loading paths (a) schematised stress paths (b) stiffness response (Clayton & Heymann, 2001)
Figure 2.41: Schematic behaviour in compression after ageing (Tatsuoka et al., 1998)
c : without overshooting before rejoining OR
d: with temporary overshooting before rejoining OR
e: with persistent overshooting and without rejoining OR
Figure 2.42: Stress-strain curves after restarting loading at a constant strain rate (Tatsuoka et al., 1998)
Figure 2.43: Effect of undrained creep on the shearing behaviour of Fujinomori clay (Momoya, 1998, Tatsuoka et al, 1998)
(b) Figure 2.44: Effects of drained creep on subsequent undrained shearing for undisturbed Vallericca clay and Metramo silty sand (Santucci de Magistris, 1998 and Tatsuoka et al, 1998)
Figure 2.45: Creep effect on fast rate shearing of undisturbed London Clay (Sandroni, 1977)
(b) Figure 2.46: Influence of time on (a) the shear stiffness of carbonate sand samples (Jovicic & Coop, 1997); (b) small strain Young’s Modulus of Metramo silty sand (Santucci de Magistris, 1998 and Tatsuoka et al. 1998)
Figure 2.47: Development of kinematic yield surfaces with time (Tatsuoka et al., 1998)
Figure 2.48: Effect of strain rate change on Chuba Gravel (Tatsuoka et al. 1998)
Figure 2.49: Effect of strain rate change on Vallericca clay (Tatsuoka et al. 1998)
Figure 2.50: Effect of strain rate change on Hostun Sand (Tatsuoka et al. 1998)
Figure 2.51: Effect of strain rate on the very small strain Young’s Modulus (Tatsuoka et al. 1998)
(b) Figure 2.52: Stress-strain behaviour of Vallericca Clay at very small strains during (a) loading and (b) unloading at different rates (Tatsuoka et al. 1998, data from Santucci de Magistris, 1998)
Figure 2.53: Strength envelopes on London Clay samples of different dimensions (Bishop, 1972)
Figure 2.54: Strengths of samples of different diameters and samples that sheared along pre-existing fissures (Marsland & Butler, 1967)
Figure 2.55: Stress-strain behaviours of intact samples and samples sheared along pre-existing fissures (Webb, 1964)
3.1 Introduction The samples of London Clay used in the present research were retrieved at Heathrow Airport, west of London, where the works carried out for the new Terminal 5 also made available a quantity of detailed information about the site.
In this chapter, an overview of the geological aspects of the London Clay will be presented for a better understanding of how this influences the mechanical behaviour of this soil. The details of the site where the samples were retrieved will be also described as observed from the site investigation carried out for the T5 project. Most of the data about this site were reported by Hight et al. (2003).
3.2 The London Clay formation “The London Clay is predominantly argillaceous and about 60% of the formation consists of thoroughly bioturbated, slightly calcareous, silty clay to very silty clay” (British Geological Survey, 2004). Beds of clayey silt grading to silty fine-grained sand increase in number and thickness from east to west. The sand and silty grains are of subangular quarz, generally less than 125 microns in diameter. Glauconite grains, up to fine to medium sand grade are dispersed through most of the sandy beds. Glauconite grains are concentrated also in some of the more clayey beds, forming marker horizons.
On the basis of the biostratigraphic characteristics of the London Clay, King (1981, 1991) traced the geological history of this soil as derived from a sequence of depositional strata. A very brief synthesis of his observations will be given here, focusing on the geological history of the London Basin in Southern England. It is believed that the formation of the London Clay is a sequence of deposition-erosion events, but for simplicity, it will be assumed here that erosion 87
or other processes occurred in this soil after the deposition was completed, so that the geological history of the London Clay can be divided into depositional and post-deposition processes.
Depositional processes: the London Basin and Hampshire Basin
King (1981) showed evidence of common lithostratigraphic features over a large part of Northern Europe, which allows the identification of a common origin for the Northern Europe Basin. The plate tectonics in the Northern Atlantic Sea in the late Palaeocene, moving upwards, created the conditions for a sedimentary basin, which originally included a large part of Northern and NorthEastern Europe, England and Scotland (Figure 3.1).
In this basin, in the Early
Eocene, about 50 million years ago, the deposition of the London Clay started due to a sea level rise over part of Southern England up to the Welsh Massif, Northern France, Western and Northern Belgium and Northern Germany, defining an area known as the Northern Sea Basin (Figure 3.2).
regional geological events, mainly linked to sea level changes, distinguished the different geographical zones.
In Southern England, two areas of London Clay Formation are distinguished, the Hampshire Basin and the London Basin. In Figure 3.3 the Eocene stratigraphy of the London Clay in Southern Britain is presented.
In both the
Hampshire Basin and the London Basin, the first material to be deposited into the embryonic sea is called the Harwich Formation, a glauconitic sand that lies above a transgressional series of sands, silts and clays collectively known as the Lambeth Group. The presence of phosphatic nodules, glauconitic grains and intense bioturbation indicate that the Harwich Formation was deposited relatively slowly. The London Clay Formation lies above the Harwich Formation and underlies the Virginia Water Formation, a bioturbated glauconitic sand with clay lamines and lenses, today present at some localities in the North-West Surrey and East Berkshire.
The London Clay was deposited in an environment where the sea level was rising. King (1981) correlated the non-uniformities within the London Clay stratum to transgressive-regressive sea level cycles. The variation of the sea level determined changes in the depositional environment, giving rise to different depositional sequences (stratigraphy) and features. Any fall of the sea level was marked by the deposition of coarser material, more evident along the edges of the depositional basin. The soil of the Hampshire Basin, close to the margin of the depositional basin, was deposited in a shallow marine environment. The high energy setting of this basin determined a varied stratigraphy, where all the sea level cycles are easily recognized by the coarsening upwards of the sequences. The London Basin soil was deposited in a deeper water environment, a low energy environment, where there was not enough time for the sediments to deposit completely before the next sea level rise, so that the stratigraphy of the deposited material is less evident than in the Hampshire Basin. Figure 3.4 shows a sketch of the idealised depositional sequences of the London Clay Formation in the Hampshire Basin and in the London area linked to sea level variation.
The total thickness of the London Clay Formation is between 50m and 150m in the London Basin and 50m and 130m in the Hampshire Basin. In both basins the thickness decreases westwards. The stratigraphy of the London Clay Formation has been found to be very consistent vertically and laterally and it is assumed therefore that continuous layers of similar characteristics persist along the thickness of the London Clay (King, 1991; British Geological Survey, 2004). (a) Lithological units Using a combination of biostratigraphy and lithological variation, King (1981) suggested a division of the London Clay formation into lithological units, corresponding to cycles of sea level rise and fall. The author recognized five principal units, from A to E (Figure 3.5), which are sub-divided into several members according to their bio-chemical characteristics. The units are more easily identified in the Hampshire Basin and near the margins of the London Basin, due to their high-energy environment, although the regularity of the occurrence of the lithological units simplifies their identification throughout the depositional basins. 89
The full sequence of the units, from A to E can be recognised only in some areas of the London Basin, such as in Southern Essex or Hampstead Heath in London (King, 1981). A schematic description of the main lithological characteristics of these units is given in Figure 3.6. In most of the London area only the lower part of the sequence is preserved, Units C and below. Clays from these units are considered in this research and a brief description of the characteristics of those units is given below. Unit A This is the deepest of the lithological units and lies above the Harwich Formation. The upper part of the Harwich Formation was originally included in the London Clay Formation as Unit A1 , although this division, also supported by King (1981), is now neglected and two parts, Units A2 and A3 , are recognized in Unit A.
The lowest unit, A2 , often referred to as a basal bed, is approximately 12m thick. It is non-calcareous, poorly sorted with a high percentage of silt, and has occasional wood fragments and pyrite nodules and contains no claystones. There are numerous partings and lenses of silt and fine sand. Sandy clays and silty clays with diffuse boundaries alternate, reflecting minor sea level changes.
The upper part, Unit A3, has an overall thickness of about 15m and it is divided into two sub-units, A3(1) and A3(2) corresponding to the base and the top parts of the layer. The base is characterised by a homogeneous and slightly calcareous silty clay layer. Silt and sand partings become more common towards the top of the unit and impersistent thin claystone layers occur. T Unit B The total thickness of Unit B is about 25m and two parts can be distinguished in it. The Unit B1 is a 1m thick sandy clay layer and marks the boundary between Units A and B, where occasionally pebbles are recorded.
On top of Unit B1 a glauconite rich layer marks the junction with the Unit B2 , indicating discontinuities in the sedimentation. Unit B2 comprises silty clays with weak silt and sand partings and numerous claystones. Sedimentary cycles are weakly discernible within Unit B2 . Unit C This unit is only occasionally present in the London Basin. It is composed of a basal layer of homogeneous silty clay that becomes sandier, micaceous and lignitic towards the top of the unit.
Standing & Burland (1999, 2005a, 2005b) observed correlations between physical properties of the London Clay and its lithological units. The water content distribution with depth marks the sequence of the lithological units, as measurements in different locations in the London Basin demonstrate (Standing & Burland, 1999; Hight et al., 2003). In Figure 3.7 the water content distribution with depth is plotted for different sites in the London Area with the correspondent division into units.
The British Geological Society (2004) proposed an informal division based on observations from different boreholes across the London Basin. Their informal division is related to the division of King (1981) as shown in Figure 3.8 and the correlation between boreholes in the London Clay is shown in Figure 3.9. The present study will be based only on King’s division. 3.2.2
(a) Influence of the Alpine orogeny The Alpine orogeny compressed the subsiding London Platform and its sediment producing the eastward plunging syncline now identified as the London Basin. This is an extremely gentle syncline and although there are some local faults, dips of more than 3o are rarely encountered. There seems to be evidence that fault-blocks were created during the Alpine orogeny by the wave of energy developed in this process. These blocks involve the whole thickness of the plates
and are believed each to have an area of about 10km2 . They are thought to have slipped relatively to each other by a few metres, causing discontinuities in the London Clay layer. (b) Erosion The erosion of a substantial thickness of the London Clay, that took place in the Tertiary and Pleistocene epochs, led to mechanical overconsolidation of the clay. The amount of erosion has been estimated to range from about 150m in Essex (Skempton, 1961) to 300m in the Wraysbury district (Bishop et al., 1965). The erosion involved all the overlying deposits, and, especially in the Thames Valley, much of the London Clay itself. Along the Thames Valley late Quaternary gravel sheets deposited after erosion (King, 1981). (c) Weathering Weathering followed the erosion. Desiccation affected the near surface of the London Clay producing rough, sub-vertical discontinuities and oxygenated groundwater converted ferrous to ferric oxide, changing the colour of the clay from blue to brown, removed pyrite and dissolved any calcium carbonate cement. Ground freezing even at large depths also occurred.
The thickness of the clay affected by weathering varies between 3 and 6m depending on the lithology. Where the London Clay outcrops, the upper 9m shows a brown colour due to oxidation with a thin transition zone above the bluegrey clay. The clay seems to be strongly weathered to a depth of 1.5m, with a granulated or fragmented texture. Below 3-4m, there is usually little obvious evidence of weathering, except for oxidation, and the structure of the clay becomes increasingly clear with depth.
Where Terrace Gravels cover the clay,
like in the Thames Valley, the effects of weathering are only limited to a very small stratum immediately below the base of the gravel. The clay seems softer and shattered but this physical weathering ceases abruptly at about 1m below the gravel, where it changes to unweathered material (Hight et al., 2003).
3.3 Mineralogy of the London Clay London clay comes from reworked soil from Jurassic shale, Greensand and Chalk and lateritic Eocene soils. It is composed of poorly crystalline kaolinite, illite, chlorite, smectite and montmorillonite. The mineralogy varies greatly within the London Basin, but generally, kaolinite seems to be predominant in the West of the London Basin, while illite in the East of the London Basin and in the Hampshire Basin; smectite and illite dominates in Central London (Huggett, 2005).
Figure 3.10 illustrates the particle size distribution of the London Clay
Formation, and Figure 3.11 shows the mineralogy of the clay at a site in the Hampshire Basin.
engineering importance was emphasised by Ward et al. (1959), Bishop et al. (1965) Ward et al. (1965) and Skempton et al. (1969). Ward et al. (1959) classified the three principal types of discontinuities in London Clay as laminations, backs and fissures. The laminations are characterised by a thin parting of more silty material with, in some cases, a piece of fossilized wood or a shell lying on the surface. The laminations or bedding would correspond to what are now recognized as the boundaries between different lithological units. Backs or joints are large fractures, predominantly vertical, forming a series of intersecting curved surfaces. The fissures are “small fractures existing in clay and siltstone beds, but not crossing the bed or horizons within the bed” (Fookes & Parrish, 1969). Faults and sheeting, which are low-angle joints, were added by Skempton et al. (1969) to the above types of fissures.
The distribution and the orientation of the discontinuities are thought to reflect the structural bedding and the erosion history of the clay. Fookes & Parrish (1969) observed that at three sites, Wraysbury, Whitecliff Bay and Herne Bay, the fissures are randomly orientated and have an irregular area, but the geometry and the surface allow a classification of the possible causes of their formation. Fissures due to shear have smooth surfaces and curved conchoidal geometry, 93
while those due to tension are planar and rough. This type of fissures generally have orientations related to tectonic stress folding at the time of formation and might have been created at the same time as the main syncline during the Alpine orogeny in the Miocene period. Fissures parallel to the bedding are likely to have been influenced by depositional variations within the clay. They are rough and planar and tend to be similar to fissures that are parallel to the present ground surface, whose origin is associated, though, to stress release during erosion. It is not clear the way horizontal fissures are attributed to stress release, as they would be expected to be inclined of 45o -φ’/2 since they are due to passive failure. Ward et al. (1959), Bishop et al. (1965) and Ward et al. (1965), Skempton et al. (1969) discussed the effects of laminations and fissures on standard tests, the implications for plate loading and compression test on fissured clays and considered the effects of sample size on tests results for fissured clays.
3.5 London Clay properties in west London In the early 60’s comprehensive research was carried out by Bishop et al. (1965), Ward et al. (1959, 1965), Webb (1964) and Skempton et al. (1969) on the characteristics of London Clay. The investigations concentrated on samples from South-West London, such as Ashford Common, Prospect Park and Wraysbury Reservoir. The data were also analysed by Wroth (1972) and revisited by Burland (1990). Hight & Jardine (1993) investigated the characteristics of the London Clay in correlation to its geology by considering samples from different sites in Central London and Standing & Burland (2005a and 2005b) emphasised the effects of the geological characteristics of the clay on engineering applications. Hight et al. (2003) added new information on the London Clay on the basis of commercial tests performed for the construction of the new Heathrow Terminal 5. The main properties of the London Clay in West London, as determined by these studies, will be briefly summarised here. 3.5.1
In the west of the London Basin, the London Clay Formation is covered by late Quaternary Terrace Gravels that were deposited after erosion. Geological 94
evidence suggests that perhaps 200m of the upper part of the 55Ma age London Clay Formation were eroded before the deposition of the Terrace Gravel. The geological sections at Heathrow T5 (Figure 3.12), Ashford Common (Figure 3.13) and Wraysbury are very similar. At Ashford Common Bishop et al. (1965) found about 5m of gravel to lie above an estimated thickness of London Clay of between 90 and 120m. The water level was found at about 1.8m above the top of the clay, but measurements of pore water pressure were not available. At T5 Hight et al. (2003) found the London Clay to be overlain by about 6m of terrace gravel, the water level being at 1.5m above the top of the clay. From borings, the total thickness of the London Clay at this site was established to be about 52m and the pore water pressure distribution was found to be hydrostatic. In the area of the T5 site called the “lagoon”, the gravel had been removed during the 1970’s by the Perry Oaks sewage works. (a) Lithological units In Figure 3.12 the lithological division of the London Clay profile at T5 is illustrated (Hight et al., 2003). Units A to C and their sub-units were identified at this site, although the sub-unit of the division C was not clear. Given the regularity of the lithology of the London Clay, a similar division could be applied to Ashford Common if the base level of the clay were known. The estimated total thickness of the clay at this site, between 90 and 120m (Bishop et al., 1965), seems to be unrealistically high, though, compared to the 52m measured at T5. 3.5.2
In Figures 3.13b and 3.14, the index properties of the clay at Ashford Common and at Heathrow T5 are illustrated and in Table 3.1 the index properties at Wrasbury Park are summarised (Skempton et al., 1969). For all the locations, the water contents of the natural samples range between 22.5 and 27% and the liquid and plastic limits range between 60-70% and 24-29% respectively, depending on the depth. The profile of water content with depth has been found to provide a graphical means of identifying the lithological division within the London Clay (Standing & Burland, 2005b; Hight et al., 2003). At different sites in the London Basin, the changes in the trend of the water content with depth 95
have been found to correspond to unit and sub-unit boundaries. On the basis of the water content profile, Hight et al., (2003) suggested a sub-unit division of the main lithological units visually identified by King at T5 (Figure 3.15). 3.5.3
In situ stresses and ko
The heavy overconsolidation of London Clay gives rise to high horizontal effective stresses, determining ko values that are greater than 1. Skempton (1961) and Skempton & La Rochelle (1965) found that in the upper 10m of the London Clay ko varies between 2 and 2.5 and this value tends to decrease with increasing depth, falling to 1.5 at about 30m depth. In Figure 3.16, the ko profiles suggested by Bishop et al. (1965) and Hight et al. (2003) for Ashford Common and T5 are plotted. The profile at Ashford Common was derived from block samples using the equation suggested by Skempton (1961):
pk =p[ko -As(ko -1)]
where p is the in situ vertical effective stress, pk is the initial suction corresponding to the stress at which the samples neither swell nor consolidate in the triaxial apparatus and As is the excess pore water ratio corresponding to the removal of the deviatoric stress and was assumed to be 0.3 (Bishop et al, 1965). The ko profile at T5 was derived from suction measurements on thin-wall samples using a suction probe on site soon after sampling (Ridley & Burland, 1993). Figure 3.17 shows the profile of suction with depth measured on thin-wall samples and on rotary core samples. The suctions measured on the thin-wall samples were assumed to be more representative of the true suctions on site. Although the edge effect causes an increase in suction, if the sample is left in the tube, the centreline effect causes a slight reduction of suction (Georgiannou & Hight, 1994). The suction measured on rotary core samples was found to be lower due to the sampling method, as the fluid in the barrel core wetted the samples reducing their suction. This wetting effect was partially reduced by removing the outer part of the samples immediately after boring, but the suction measured from the rotary core samples was still assumed to be as a lower bound for the suction on site. 96
The permeability of the London Clay has been found to be strongly associated with its lithological units. Hight et al. (2003) summarised the horizontal permeability at different sites in Central London made with self-boring permeameters or self-boring pressuremeters and included the identification of lithological units (Figure 3.18). They observed that the permeabilities of samples from lithological unit A2 were the highest, because of the content of sandy layers in this unit, but they also noticed that an increase of permeability could be expected towards the top of each unit as result of an increasing frequency of silt partings.
Standing and Burland, (2005a), found that erosion of London Clay might increase its mass permeability. The authors observed at St. James’s Park, in Central London, a value of permeability of around 1x10-11 m/s for Unit B2 , which increased to 4x10-11 m/s in areas that had been subjected to erosion. Rapid sudden inflows of water occurred in the upper part of Unit A3 , where there is the larger concentration of sand and silt partings. 3.5.5
Cone penetrometer tests at T5
Cone penetrometer tests were performed at Heathrow T5 by the Building Research Establishment for the present research. The coordinates for the boreholes will be given in Table 5.2.
Fugro subtractive piezocones with face
filters were used. The filters were pre-saturated with glycerine. The cones were robust enough to cope with the claystones, but zero shifts could occur in the readings if claystones were hit. The results of the tests are summarised in Figure 3.19. The cone resistance and the sleeve friction show a fairly uniform increase with depth. A few jumps in the readings were thought to be associated with boundaries of the lithological units. Two main discontinuities, shown as peaks in the readings, were noticed at 18 and 25m depth. Hight et al. (2003) suggested that these levels could be the boundaries of the lithological sub-units and they are indicated in Figure 3.19. However, the presence of claystone layers, frequent in unit B2 , might have induced these peaks. The pore pressure distribution seems to
be more indicative for the identification of the lithological units and seems to mark more clearly the changes in lithology through the depth. 3.5.6
London Clay is a typical example of a fissured stiff plastic clay. It has a brittle behaviour with localization of strains and a well defined post-rupture strength. Figure 3.20 shows typical triaxial stress-strain curves of samples from different depths at T5, sheared undrained after isotropic consolidation to their estimated in situ stresses.
Hight & Jardine (1993) observed that the brittleness tends to
increase with depth.
Due to the fissured nature of this clay, samples tend often to fail along preexisting fissures. In this case they may not mobilise any peak strength, but tend to reach directly the large strain strength (e.g. 11.35m depth in Figure 3.20). Hight et al. (2003) observed that the strength envelope of samples that sheared along pre-existing fissures corresponded to c’=0 and φ ’=20o and matched the post-rupture strength envelope (Figure 3.21). The authors also noticed that some samples,
corresponding to c’=0 and φ ’=12o , which is very close to the values of c’=0 and φ ’=15o found by Bishop et al. (1965) from triaxial tests on block samples from Ashford Common and by Skempton, (1944) from direct shear tests (Figure 3.22). Jardine et al., (2004), pointed out that the strength on fissures is similar in magnitude to the post-rupture strength, which is observed after the breakage of bonding, because fissures are surfaces where little or no relative shear displacement has occurred. The strength on fissures is therefore higher than the residual strength, which is mobilised at very large strains. Hight et al., (2003) also observed that the strength in extension at T5 was lower than that in compression and corresponded to the fissure strength as in extension the failure mode generally tended to coincide with the predominantly sub-horizontal natural fissures.
Skempton (1977) found the strength along fissures and joints to be similar to the strength of normally consolidated reconstituted samples (Figure 3.23). 98
Burland (1990) found a similar pattern, revisiting the data from Bishop et al. (1965) and Ward et al. (1965) from Ashford Common. He noticed that the postrupture envelopes were similar to the intrinsic failure line of the reconstituted material (Figure 3.24).
Figure 3.25 shows the peak strength envelopes for samples from four different depths at Ashford Common (Burland, 1990). The depths for the Levels A to F are indicated in Figure 3.13a and Table 3.2. The strength envelopes of the natural specimens lie above the intrinsic strength of the material, reflecting both the stronger microstructure of the natural soil and its overconsolidation. The strength envelopes of samples from Levels A to E seemed to define a unique line, while samples from Level F showed higher strengths. This could reflect a difference in the lithological unit between samples from Levels A to E and samples from Level F, although the unit boundaries at Ashford Common are unknown because the base of the clay was not loacted.
Hight & Jardine (1993) found that the
changes in lithology and fabric in the London Clay influenced the undrained strength and the failure envelopes. The authors observed that zones of higher sand content, with partings of fine sand or silt, are associated with a higher effective stress failure envelope than the adjacent more plastic clays. This induces a change of the cohesion intercept with depth. The authors also observed that sandier layers occur in the London Clay at depths between 15-18m, 24-27m and below 30m, so that below 25m depth the material can be classified as a hard clay.
The different strength envelopes for samples from different depths in
Central London are shown in Figure 3.26. The strength envelopes measured by Hight et al. (2003) at Heathrow T5 agreed with this scheme.
Standing & Burland, (2005b), correlated the changes in the undrained strength of London Clay with changes in water content and therefore with the lithologcial units. The authors observed that, at St James’s Park, the strength tended to reduce in the lithological units having higher water contents.
The stress paths normalised for volume for samples from different depths at Ashford Common are shown in Figure 3.27, together with the Hvorslev surface of the reconstituted soil. The Strength Sensitivity, St , of London Clay has been 99
quoted to be higher than two (Cotecchia, 1996; Burland, 1990; Webb, 1964) and the normalised state boundary surfaces seem to become horizontal after a normalised stress s’/σ* e of 0.8, although they curve round to the s’ axis at even higher stress levels. 3.5.7
Anisotropy and stiffness
The anisotropic behaviour of London Clay was first addressed by Bishop et al. (1965), Bishop, (1966) and Atkinson (1975) by comparing the large strain strengths of samples cut horizontally and vertically. London Clay samples deform more in the direction of deposition, the vertical, than in the plane of deposition, the horizontal (Atkinson, 1975, Hight et al., 1997, Wongsaroj et al., 2004).
Laboratory test data, illustrated in Figure 3.28, show that the small strain stiffness of London Clay is stress dependent, so that Gmax =Ap’n
with the exponent n varying between 0.36 and 0.61 and usually assumed to be 0.5 (Wongsaroj et al., 2004). The magnitudes of the shear moduli Ghh , Gvh and Ghv change with location and depth, but Ghh is higher than Gvh and Ghv . From bender elements tests on undisturbed samples of London Clay cut horizontally and vertically, Jovicic & Coop (1998) noticed that Gvh was equal to Ghv . The authors also found the ratio Ghh /Gvh to be constant with depth and equal to about 1.5. These observations were confirmed by other measurements in different sites in Central London (Hight et al., 2003; Wongsaroj et al., 2004). Cross-hole and down-hole seismic wave velocities measured at Heathrow T5 at 1m depth intervals from the ground surface, show that at this site the shear modulus Ghh is 50-100% higher than Gvh or Ghv , which were equal (Figure 3.29). These values were found to be consistent with the same moduli at other sites in Central London. Table 3.3 summarises the values of the ratio Ghh /Gvh found in the literature.
Hight et al. (2003) observed that the ratio Ghh /Gvh against depth reflected the influence of lithology on anisotropy (Figure 3.30). The step changes in the distribution of the ratio Ghh /Gvh when plotted against depth were interpreted to indicate the changes of lithological unit, similarly to the water content distribution (Figure 3.15).
from ground level [m]
13 - -21.5
Table 3.1: Index properties of London Clay at Wraysbury (Skempton et al., 1969)