Because the Earth is a complex three-dimensional object, with numerous structures and hydrogeomorphologies that are diffi cult to imagine and to characterise, it is important to develop concepts and techniques that will support the identifi cation and description of its spatial arrangements in manipulative two-dimensional representations.
Although this challenge can be approached by a hybrid of disciplines, it has prompted the development of the scientific fi eld of soil mechanics and, in particular, the area of site characterisation.
Site characterisation and insitu test interpretation have evolved from basic empirical recommendations into a sophisticated area demanding a thorough knowledge of material behaviour and numerical modelling.
The state of the art report for this plenary session presents a critical appraisal of the understanding and assessment of the stress-strain-time and strength characteristics of natural soil conditions and explores new interpretation methods capable of measuring soil properties shaped by effects of microstructure, stiffness non-linearity, small and large strain anisotropy, weathering and destructuration, consolidation characteristics and rate dependency.
With modern testing techniques and more rigorous methods of analysis, present design practice of insitu testing interpretation has moved from early experience gathered in reconstituted clay and sand to a more general fi eld that covers a variety of natural geomaterials, comprising clay deposits, granular soils, intermediate permeability silts and bonded hard soils and soft rocks.
A brief review of general recommendations for different geomaterials is outlined in the report as follows:
Theoretical and empirical approaches are mainly related to high plasticity clays in which simple rate-independent, perfectly-plastic isotropic soil models are generally adopted in the interpretation of insitu test data.
Whenever the experience of soft clays is extended to stiff overconsolidated clays, interpretation of test data should follow specific recommendations.
Current practice in characterising soft clay sediments clearly indicates that signifi ant progress has been made in the development of analytical and numerical methods adopted for the interpretation of pressuremeter, cone and flow penetrometer mechanisms.
Interpretation of test data as inverse boundary value problems allows engineers to use insitu tests to predict features of behaviour: nonlinearity is captured by a combination of seismic and pressure meter tests, small strain anisotropy by seismic tests with waves polarised in different directions and large strain anisotropy by vane tests.
Strain rate effects and the resulting increase in undrained shear resistance are recognised as important and can be evaluated from both analy tical and numerical methods.
Despite recognition of limitations on the interpretation of insitu tests, it is in the diffi cult conditions of soft clay that both laboratory and insitu tests offer the best and most reliable combination of test procedures and interpretation methods available for assessing constitutive parameters for engineering design problems.
Without the possibility of retrieving undisturbed samples, soil classifi cation from insitu test results in granular soils should preferably rely on at least two independent measurements.
A measure of the ratio of the elastic stiffness to ultimate strength, expressed as G o/q c and G o/N 60 , has shown to be fairly sensitive to cementation and ageing and is therefore useful for identifying the characteristic behaviour of geomaterials.
The ratio of G o/q c predicted from critical state approach is shown in Figure 1, in which fresh uncemented sands are characterised by a region defi ned through upper and lower boundaries. Values of q c and G o profiles that fall outside and above this region suggest possible effects of stress history, degree of cementation and ageing.
Since G o and q c are controlled by void ratio, mean stress, compressibility and structure and are therefore different functions of the same variables, it is possible to anticipate that as a ratio these two measurements can be useful in prediction soil properties.
Acknowledging that critical state parameters and initial soil state are in the root of the so-called 'state parameter', , an obvious approach is to correlate the G o/q c ratio and as shown in Figure 2.
For a given mean stress, the ratio of G o/q c is not very sensitive to variations in strength and stiffness and is shown to decrease with decreasing (ie G o/q c decreases with increasing relative density). The seismic cone is therefore considered useful in assessing soil properties in granular deposits.
The approach using the G o/q c ratio and a number of other correlations based on pressuremeter, cone, cone pressuremeter, dilatometer and SPT (in association to energy measurements) are outlined in the report.
Scope for future research comprises a better understanding of soil crushability and the effects of soil structure, which may require discrete element methods (DEM) to investigate the response of particle assemblage in penetration processes.
The mechanical behaviour of bonded soils is still not fully understood and cannot be modelled even in the laboratory when testing elemental specimens. A combination of poor understanding, lack of constitutive models and complex boundary conditions give rise to interpretation methods of insitu tests that are empirical in nature, with the single exception of the pressuremeter test.
Combinations of G c/q c and G o/N 60 are shown to be useful for the characterisation of the bonded structure of cemented geomaterials.
The small strain stiffness to strength ratio embodied within the G o/q c and G o/N 60 terms is expected to increase with bond/cementation, primarily because the effect of these on G o is stronger than on q c (or N 60 ).Cavity expansion offers a unique possibility of combining the effects of structuration and destructuration in shear mode.
An analytical solution is introduced where the flow rule has been accommodated to describe evidence that dilation of the intact soil is inhibited by the presence of the cement component.
In energy terms it is suggested the total work done by the stresses at the boundary of an element is partly dissipated in friction and partly in disrupting the structure of the soil. Pressure expansion relationships plotted in Figure 3 demonstrate that limit pressure is strongly affected by the degradation of the cohesion component on a bonded soil.
Intermediate permeability silt soils
For intermediate soils, the simplest idealised approach of a broad distinction between drained (gravels and sand) and undrained (clay) conditions for the interpretation of insitu tests cannot be applied since test response can be affected by partial consolidation. Consequently analytical, numerical or empirical correlations can lead to unrealistic assessment of geotechnical properties.
Unfortunately there are no standardised recommendations to guide engineers on how to perform insitu tests or interpret test results in these materials. Since numerical assessment is restricted to the consolidation characteristics of clays, recommendations for intermediate soils are empirically based, established from fi eld observations.
The report provides a critical overview of recent studies in clay and silt soils and describes a method designed to recognise consolidation patterns that may take place during penetration and cavity expansion.
Recent studies place emphasis on the normalisation of penetration results, represented by an analytical 'backbone curve' of penetration resistance against rate effects expressed by a dimensionless velocity.
Cavity expansion has been adopted to couple the rate of penetration to consolidation effects in attempting to identify drainage patterns in field tests.
Centrifuge tests have provided the necessary experimental database against which theoretical concepts have been verified, as illustrated in Figure 4.
A general recommendation for fi eld tests in intermediate permeability soils is to avoid tests that yield dimensionless velocities within the range of 0.001 to 10. In this range, partial drainage is expected to occur and properties assessed from fi eld test interpretation can be overestimated, in particular the undrained shear strength.
Since the insitu behaviour of natural soils is complex, a single general recommendation is to cross-correlate measurements from different tests.
When datasets are combined there is more scope for rational interpretation and, for this reason, emphasis has been placed on correlations with mechanical properties that are based on the combination of independent measurements.
A wide variety of solutions is now available - some are rigorous but others are a rough approximation of soil behaviour and should be viewed only as guide to decision-making.
Regardless of the approach to interpretation of data, analysis should always be supported by background theory. Overall, site characterisation is a subject to be approached by experienced engineers, but geotechnical design can no longer be guided by experience only.
Fernando Schnaid is associate professor of civil engineering at the Federal University of Rio Grande do Sul, Brazil. His research interests in soil mechanics comprise site investigation, insitu test interpretation, foundation design and environmental engineering.
After a period in industry at Fugro-McClelland UK, Schnaid moved into research and has supervised more than 40 graduate students. He has been a visiting professor in Argentina, Australia, Portugal, the UK and Uruguay. He is the author of four textbooks in geomechanics, written in Portuguese, and more than 150 papers.