The assessment of the influence of tunnelling on buildings and other structures has become an important and costly environmental issue. A large proportion of the petitions against the Jubilee Line Extension were settlement related.

The key steps in assessing building response to tunnelling are:

(i) prediction of settlement;

It is common practice for each of the above steps to be carried out in isolation and certain simplifications are introduced; e.g. it is assumed that the individual effects of separate tunnels can be simply added to achieve the effect of multiple tunnels in step (i), and the influence of building stiffness is ignored in steps (i) and (ii).

This extensive two dimensional finite element study, completed in April of 1996, considered many important aspects of modelling of tunnelling in stiff clay, and also culminated in the presentation of new design charts to integrate steps (i) and (ii) above.

First, the importance of modelling the nonlinear elastic nature of stiff clay was demonstrated. A linear elastic model (isotropic or anisotropic) pre yield, with a Mohr Coulomb yield surface and plastic potential defining plasticity, predicts a surface settlement profile which is too shallow and too wide when compared with field data. The introduction of a nonlinear elastic model (isotropic) pre yield goes some way to correcting this error.

Employing nonlinear elastic perfectly plastic soil models (Jardine et al, 1986; and Puzrin and Burland, 1995), the interaction between two tunnels was investigated. Tunnels located one directly above the other, and two side by side at the same axis depth, and diagonally orientated tunnels were all modelled. The rest period between the construction of the two was varied, as was the physical spacing of the two. The significance of modelling nonlinear permeability was also studied, employing a model which varied permeability with mean effective stress (Vaughan, 1994). It was shown that interaction is more significant for two tunnels located side by side than one above the other, but that for all geometries if interaction was to be expected then the surface settlement profile would not be the simple addition of two greenfield short term profiles - first because the presence of a tunnel affected the shape and size of the profile developing above a second tunnel, regardless of the time interval between excavation of the two, and also because of the additional settlement occurring during the rest period.

The analyses of soil / structure interaction represented surface structures as beam elements with specified axial and bending stiffness. Over 100 finite element analyses were used to investigate the influence of tunnel excavation at varying depth and lateral positions on surface buildings of varying stiffness and size. The buildings were shown to modify the expected greenfield surface settlement profile (interpreted in terms of the sagging and hogging deflection ratio, Burland and Wroth, 1974) and the horizontal displacements (interpreted as horizontal compressive and tensile strain, Boscardin and Cording, 1989). These results were combined to develop design charts which predict the expected modification to the conventionally predicted greenfield deflection and strain for a surface building of given size, relative position (relative to tunnel), and relative bending and axial stiffness (relative to the soil).

This research also lead to the successful use of a plastic limiting tension yield surface to model the cracking of masonry panels. Several masonry panels were analysed, subjecting them to distortions (deflection and strain) at their bases, and predicting the tensile strain within the panel which was compared with the elastic beam theory predictions of likely damage (Burland, 1995).

Boscardin, M. D. and Cording, E. J. (1989). Building response to excavation induced settlement. Journal of Geotechnical Engng., ASCE, Vol 115, No1, pp 1 - 21.

Burland, J. B. (1995). Assessment of risk of damage to buildings due to tunnelling and excavation, Proc. 1^st Int. Conf. Earthquake Geot. Eng., IS-Tokyo '95.

Burland, J. B. and Wroth, C. P. (1974). Settlement of buildings and associated damage. Proc. Conf. 'Settlement of Structures', Cambridge, Pentech Press (published 1975, London), pp 611 - 651.

Jardine, R. J., Potts, D. M., Fourie, A. B., and Burland, J. B. (1986). Studies of the influence of non-linear stress-strain characteristics in soil-structure interaction. Geotechnique 36, No. 3, pp 377-396.

Puzrin, A. S. and Burland, J. B. (1995). Nonlinear model of small strain behaviour of soils. Internal Report, Imperial College, London.

Vaughan, P. R. (1994). Assumption, prediction and reality in geotechnical engineering. Geotechnique 44, No. 4, pp 573 - 609.

GROUND MOVEMENT AND STRUCTURES RESPONSE

The prediction of ground movements and the response of structures and services to tunnelling has been a source of controversy in recent years. Although methods of prediction have improved, the recent collapse of part of the tunnels of the Heathrow Express Project has drawn more attention to the response of London clay during tunnelling. The above research project, which was started 2 years ago, aims to improve our knowledge and understanding of how the ground and buildings respond to tunnelling. This is to be achieved through the monitoring of a live tunnelling project and the analysis of the data obtained.

The Heathrow Express Project is a high speed link between Paddington and Heathrow Airport, which involves the construction of a 5.7m internal diameter, 6.8km long tunnel. A co-ordinated and planned programme of field measurements was set up as the tunnelling started which provides a wealth of monitoring data along the line of the tunnel. The instrumentation includes biaxial electro levels, extensometers, inclinometers, piezometers and precise levelling. The main areas of the research are:

1. Green field sites; with several cross-sections of surface levelling over the tunnels.

2. The Heathrow Central Terminal Area (where the recent collapse took place); including major buildings with different types of foundations and structural stiffness.This area was originally planned to be excavated using the New Austrian Tunnelling Method/(NATM).

3. Underground Tunnels; Heathrow Express line passes under the Piccadilly Line tunnels and stations, and under fuel and service pipes in the airport area. Tunnelling is now complete at the Terminal 4 site.

The field measurements will be interpreted to advance the state of knowledge concerning ground movements due to tunnelling and their mitigation. The original aim was to establish a better understanding of the following aspects so that future designs could be made with greater reliability and economy. The research plan may now be affected by the results of the current investigations into the failure of the station tunnels, which occurred on 25 October 1994.

1. Subsidence trough. The existing predictive methods give reasonable estimates for the initial vertical surface displacements for single tunnels beneath green field sites.There are considerable uncertainties as to whether superposition can be used for multiple tunnels and large openings. The prediction of lateral displacements and subsurface displacements are also less certain, but are important for assessing building response. Thus studies of the development, shape and amount of subsidence patterns will form part of this research.

2. Soil-Structure interaction. There are few reliable case records of the influence of building stiffness on the shape and magnitude of the subsidence. In most cases, a conservative approach is adopted in which the stiffness of the building is neglected. This can lead to over-estimates of building damage and unnecessary expenses on work designed to minimise movements.

3. The New Austrian Tunnelling Method (NATM). This method was used to construct the 9.5m external diameter stations and some parts of the main tunnels. The method has not been used before in London clay and new parameters will have to be defined for its application in such strata. A detailed study of the ground deformations around the tunnels will help in judging the performance of this method in London Clay.

4. Whilst a number of traditional and novel methods are available for protection and remedial work, there are few well documented studies of the effectiveness of the various measures. Because of the sensitivity of the buildings in the Central Terminal Area, fracture grouting will be used and its performance will be systematically assessed.

The majority of piles in cohesive materials are designed using the total stress method where the unit shaft friction is correlated against the undrained shear strength of the soil, i.e: t = a s_u. This is a simple but crude design method which fails to allow for factors such as soil overconsolidation and sensitivity; pile type, length and stiffness; method of installation and direction and rate of loading. In addition, the measurement of a unique value of s_u is fraught with difficulties.

Designers of foundations for offshore structures face further uncertainties since the a correlations are based on load tests on relatively short, stiff, low capacity onshore piles. In contrast, North Sea piles are often 2m in diameter and can be 100m long. These piles are not load tested, hence the validity in extrapolating the results of the existing database is questionable.

Current research has focused on the need for a more fundamental approach to pile design using the principle of effective stress to predict the stress changes in the ground. This will enable the variations caused by different soil conditions and pile types to be analysed in a more rational manner. The understanding of displacement pile behaviour has been increased during the last decade by the collection of high quality effective stress test data, most notably by Oxford University, the Norwegian Geotechnical Institute, a Joint Industry large diameter pile testing programme and Imperial College (IC). This work has been supplemented by numerical modelling studies at MIT and IC and ring shear interface research.

The Imperial College pile consists of a heavily instrumented, 100mm diameter, closed-ended steel pipe, up to 20m in length. The instruments, designed by Dr. Andrew Bond during his Ph.D.work at IC, are able to measure radial total stresses, local shear stresses, axial pile loads, pore pressures and temperature at various levels along the pile. Six years of field testing has proved the instruments to be very reliable and accurate.

The pile is installed by fast jacking and the instruments are monitored during the installation, equalisation and loading stages. Variations are made to the testing procedure to investigate the effects of parameters such as installation rate, length of jack stroke, loading rate, load cycling and time on pile capacity.

Testing programmes have been conducted at the following six sites. Analysis of the results and modelling of soil behaviour is assisted by parallel Ph.D. laboratory research (currently conducted by Tim Connolly and Reiko Kuwano), involving triaxial stress path, hollow cylinder, resonant column, intrinsic properties and ring shear tests. Extensive in situ testing has been performed at each site by the Building Research Establishment.

The research has shown that pile behaviour in clays is heavily dependent on the soil consistency, overconsolidation ratio and sensitivity. Pile characteristics also influence the load capacity, particularly the slenderness ratio and the distance of a soil element from the pile tip (h/R effect). The shaft friction has been shown to be dependent on the soil-interface residual angle of friction which can be determined using laboratory ring shear interface tests.

Tests in sands reveal similar behaviour: shaft capacity depends on pile length, roughness, diameter and distance from the pile tip. The main soil parameters are relative density, stiffness and interface angle of friction.

Large-scale tests on open-ended piles were conducted in dense sand at Dunkirk. The effects of end condition have been analysed allowing the formulation of modified correlations for open-ended piles. Base capacity, plugging, long-term pile set-up and the interaction effects between piles in a group have also been investigated.

The new understanding of pile behaviour has allowed the development of design methods for open and closed-ended piles in clays and sands. The applicability of these methods and their advantages over current methods have been verified against large databases of high quality full-scale pile tests.

Bond, A.J. & Jardine, R.J. (1995) "Shaft capacity of displacement piles in a high OCR clay", Geotechnique, 45(1), pp 3-23.

Chow, F.C. (1995), "Field measurements of stress interactions between displacement piles in sand", Ground Engng, 28(6), pp 36-40.

Chow, F.C., Jardine, R.J., Brucy, F. & Nauroy, J.F. (1996), "The effects of time on the capacity of pipe piles in dense marine sand", Proc. 28th Offshore Technol. Conf., Houston (in press).

Lehane, B.M., Jardine, R.J., Bond, A.J. & Frank, R. (1992), "Mechanisms of shaft friction in sand from instrumented pile tests", J. Geotech. Engng, ASCE 119(1) pp 19-35.

Lehane, B.M. & Jardine, R.J. (1994), "Displacement pile behaviour in glacial clay", Can. Geotech. J., 31(1), pp 79-90.

Lehane, B.M. & Jardine, R.J. (1994), "Displacement-pile behaviour in a soft marine clay", Can. Geotech. J., 31(2), pp 181-191.

Jardine, R.J. & Overy, R.F. (1996), "Axial capacity of offshore driven piles in dense sand", Proc. 28th Offshore Technol. Conf., Houston (in press).

A STUDY OF STRUCTURAL EFFECTS ON THE MECHANICAL BEHAVIOUR OF NATURAL STIFF CLAYS

The research is intended to contribute to the general understanding of the mechanical behaviour of natural structured soils. In particular the study will investigate whether it is possible to normalise the strength and yield envelopes of structured stiff clays which have been one-dimensionally consolidated both to states below the yield point under oedometric compression and also beyond yield.

For natural clays the strength behaviour depends on structure aswell as on void ratio and stress state, so an investigation of the structural effects is being carried out. Particular care is taken to identify whether major structural modifications occur whilst compressing the clay to different initial volume - stress states and if these modifications induce any discontinuity in the trend of strength behaviour. In order to do this a laboratory test programme on undisturbed and reconstituted stiff clay samples is being carried out :

a) Compression Test Programme :

-Constant Strain Rate (CRS) Oedometer tests, loading and unloading stages with s_vmax= 25MPa

-End of Primary Oedometer tests, loading and unloading stages with s_vmax = 7000 kPa

- 24h Oedometer tests, loading and unloading with s_vmax = 7000 kPa

-Swelling tests starting from the initial stress state of the undisturbed samples

The above investigation of the differences in clay compressibility for a variety of test procedures is a fundamental step in understanding what governs the clay's response. In particular, it will be seen if different levels of destructuring occur in CRS tests conducted at different speeds. Any difference might be expected to give rise to different responses during the unloading stages. An analysis of the strain rate dependency of the clay's behaviour is also being carried out. The swelling tests will investigate whether secondary swelling destructures the soil. Scanning electron micrographs will be used to look for visual evidence of structure.

b) Triaxial Test Programme

-Isotropically compressed to 1200kPa; drained and undrained shearing.

-Ko compressed to 3000kPa;drained and undrained shearing.

The test data will be used to define the strength and yield envelopes for a large range of yield stress ratios both below and above the yield point. The post-peak data in particular are expected to be useful in assessing the importance of structure.

The first soil being tested in this programme is a Plio-Pleistocene Subappenine Blue Clay from Taranto, Southern Italy. The laboratory tests are being conducted on undisturbed block samples of this stiff and overconsolidated marine clay taken from a depth of 25m. In each case detailed comparisons are being made between the behaviour of the reconstituted and undisturbed clay. A field investigation is being carried out to give the soil profile and the in-situ stresses.

Natural soil deposits close to the surface exist at relatively low moisture contents over much of the earth. Saturated or unsaturated, these deposits often have negative pore water pressures. In order to predict the soil behaviour accurately it is necessary for the laboratory testing to monitor and control not just applied stresses but also these pore pressures, or soil suctions.

Control of matrix suction in the laboratory has, historically, involved the elevation of the air pressure and application of the axis translation technique. This technique may however inhibit the movement of moisture within the sample and can affect desaturation.

Recent developments (Ridley, 1993) have meant that the measurement of soil suction can be undertaken quickly, accurately and directly under atmospheric conditions, in both laboratory and field.

For the main experimental programme a suction controlled oedometer has been developed which operates under atmospheric conditions, fig 1. The osmotic potential of a large molecular weight salt (polyethylene glycol, PEG) is applied across a semi-permeable membrane to control matrix suctions. Pore water can pass freely through the membrane but the larger PEG molecules are retained allowing suctions to controlled. The relationship between concentration of PEG salt and suction has been defined for this application. Features incorporated into the oedometer include independent, continuous recording of:

The stress path followed during testing can be defined and controlled by the measured vertical stress, strain or suction.

The influence of suction on the compressibility and swelling of a range of soils is currently being assessed. Intrinsic properties and soil moisture characteristics are defined for each soil. Routine soil classification tests, mineralogy and SEM studies have also been carried out.

DELAGE.P, VICOL.T, SURAJ DE SILVA.G.P.R. Suction Controlled Testing of Non-Saturated Soils with an Osmotic Controlled Consolidometer. 7^thICES Dallas 1993

DINEEN.K & BURLAND.J.B.B. A New Approach to Osmotically Controlled Oedometer Testing. 1^st International Conference on Unsaturated Soils, Paris, 1995

KASSIFF.G & BEN SHALOM.A. Experimental Relationship between Swell Pressure and Suction. Geotechnique 21, No.3, pp 245-255

RIDLEY.A.M. The Measurement of Soil Moisture Suction. PhD Thesis, University of London, 1993

WILLIAMS.J & SHAYKEWICH.C.F. An evaluation of polyethylene glycol (PEG) 6000 and PEG 20000 in the osmotic control of soil water matric potential. Canadian Journal of Soil Science 102(6), pp394-398, 1969

THE BEHAVIOUR OF OFFSHORE PILES

The aim of this project is to develop a realistic numerical method of analysis for predicting the behaviour of an offshore pile group subjected to a full set of design loadings. The research will extend the successful two dimensional method used for studies of the Hutton TLP foundations into a full 3 dimensional analysis that can deal with turning moments, horizontal forces and non-linear pile group interaction.

The numerical model will be developed from an existing, powerful, elasto-plastic finite element package (ICFEP). The model will be designed to incorporate realistic soil constitutive relationships that reproduce, as far as possible, the soil characteristics seen in a series of laboratory test programmes. Key features will include steep non-linearity at small strains, a critical state type description of soil yielding and modified frictional behaviour at the pile-soil interface. The analyses will also be set out to incorporate effective stress conditions around the piles that are compatible with the results of the parallel studies of installation effects and any other available high quality data.

The high costs and large CPU times of a full 3-D analysis makes it impractical to undetake a comprehensive study of the lateral loading on a pile in a non-linear small strain, elasto-plastic medium using a full 3-D analysis. The method proposed is a semi-analytical finite element process applicable to axisymmetric bodies. A fourier series representation for force and displacement in the meridional plane is used to transform the3-D problem into a series of uncoupled 2-D problems.

A programme of numerical analysis of real geomechanical problems has been carried out at Imperial College since 1979 under direction of Prof.D.M.Potts. A variety of non-linear elasto-plastic finite element techniques have been used within the program developed at lmperial College. As part of this research a number of embankment dams and embankments have been analysed, partly as a research exercise but primarily as part of design consultancy. Results of some of these analyses have been calibrated against field measurements. These analyses have included predictions of deformation and problems of stability in drained and undrained embankments, including strain-softening and progressive failure (Potts et al, 1990).

The stress-strain behaviour of fill materials prior to failure has generally been modelled by non-linear incremental-elastic techniques. Plastic behaviour has only been modelled at and post-peak.

Some problems have been experienced with this approach, both in deriving satisfactory numerical parameters from test data, and in reproducing predicted displacement patterns which agree with prototype displacements in both the vertical and horizontal direction simulataneously. The problems are greatest with granular fills such as rockfill. It is suspected that part of the problem lies in the failure to model plastic effects pre-peak.

Depending on progress with tasks one and two, the programme would be extended to look at some of the problems which have already been identified from present work on embankments, which includes the modelling of undrained behaviour of fills, the importance of soil geometry in controlling deformations, cracking by hydraulic fracture, effects of progressive failure in undrained and partly-drained embankments, and other limitations of assuming rigid body mechanics in stability analysis.

Potts, D.M., Dounias, G.T. and Vaughan, P.R. (1990). Finite element analysis of progressive failure of Carsington embankment, Geotechnique, 40, No.l, 79-101.

The behaviour of soils is time dependant as strains increase with time under constant stress conditions ("creep") and stresses decrease under constant strain conditions ("relaxation"). Moreover soils exhibit an increase in strength and stiffness with time ("ageing"). Usually the forrner phenomenon is disregarded for granular soils whilst the latter requires a geological period of time to develop, though Daramola (1980) showed that a few weeks are sufficient to cause a substantial increase in strength and stiffness in silica sand.

It appears that the creep phenomenon is not negligible in calcareous soils and that ageing takes place faster. Engineers have often observed on chalk fills a considerable increase in strength already after a few days. The aim of this research is to quantify the influence of time on the behaviour of a calcareous sand obtained by manual destructuration of a natural calcarenite sampled in the south of Italy (Gravina-Bari). A mathematical expression will be developed to reproduce the experimental observations. This expression will be included afterwards in a more general constitutive model of soil behaviour. This model will not only reproduce the usual stress-strain behaviour of soils but also take into account the effects of time on such a esponse.

The experimental investigation is performed with a computer controlled triaxial cell with a maximum cell pressure of ~5MPa. The cell is displacement controlled and is equipped with local strain gauges, which allow the measurement of axial and radial strains directly on the soil specimen. The axial strains are measured by means of a pair of inclinometers with electrolevels, the resolution of which has recently been improved by an electronic filter. The radial strains are measured by means of a strain gauged radial belt.

Structured soils are widely present throughout the world and in Europe particularly. In recent years considerable progress has been made in the understanding of the mechanical behaviour of structured geomaterials, by means of extensive experimental investigations on artificial and natural structured soils. It has been shown that the engineering response of structured soils with very different geological origins can be explained within the same conceptual framework. More recently some mathematical models have been proposed to reproduce their main features.

Despite the better understanding of the soil behaviour, the design of engineering structures involving these materials is still very empirical. The geotechnical properties of structured soils are not fully exploited in engineering practice, often resulting either in an unnecessarily expensive design or in failures.

In this numerical research the behaviour of different types of foundations on structured soils is analysed, with particular regard to identifying the effects of the soil structure on the behaviour of the foundations. The analysis is performed with the Imperial College Finite Element Program (ICFEP), developed by Professor D.M. Potts. The soil behaviour is simulated by means of a constitutive model proposed by Lagioia and Nova (1995) for reproducing the stress-strain response of structured soils. The model takes into account the initial undisturbed structure of the soil and the effects of the progressive destructuration as plastic strains are experienced by the material

Problems related with potentially expansive soils are more and more frequent, both due to changes in climatic conditions of the globe and to new developments on arid or semi-arid regions.

The moisture content variation in potentially expansive soil causes volumetric strain: the soil swells or shrinks. In U.K. severe moisture variation is caused by trees, associated with dry summers.

This project intends to investigate the behaviour of potentially expansive soils (e.g. London Clay), in terms of effective stress, during desiccation and rehydration. It will be carried out partly in the field, and partly in the laboratory. It is in continuity with the work of Crilly and Chandler since 1987 at the Imperial College.

In the field, the seasonal suction variation will be studied, using the filter paper method and displacements will be measured to monitor the ground's seasonal movements. In the laboratory triaxial tests will be performed simulating the stress paths followed during desiccation and subsequent rehydration, on undisturbed specimens from the instrumented site. Studying the stress paths followed in the field will enable the volumetric strains to be measured more realistically, and will improve the accuracy with which heave may be predicted.

EFFECT OF CHANGE IN RATE OF SHEARING ON

The purpose of this investigation is to examine the behaviour of soils at earthquake induced shearing velocities. The subject has been investigated by previous researchers at Imperial College and the response of different soil types under earthquake loading was analyzed. However, limitations of the ring shear apparatus restricted their investigations. The volume of the sample cannot be controlled and approaching earthquake induced strain rates this problem is more profound.

Our first task was to put together a new data acquisition system for the automatic logging, analyzing and displaying of the measurements from the load cells, proving rings and transducers of both ring shear apparatuses. Furthermore, in an effort to control the volume of the sample at high strain rates, a new set of rings was developed. The new system was calibrated and tested in order to verify its validity and the new rings were tested with dummy samples so that any additional friction that comes from the system can be accounted for. The new system has given us the advantage of testing at high shearing velocities for a much longer period than the old system.

In the last few months of this research programme different soil types were tested, with varying clay contents and plasticity indices, submerged and non-submerged, normally and over-consollidated in an effort to complete the investigation.

Many different geotechnical problems are affected by the shear coefficient of soil at small strains. These include most static foundation problems aswell as dynamic analyses of soil behaviour during earthquakes, explosions, machine or traffic vibrations or cyclic wind or wave loading

Both static and dynamic methods exist for investigating shear stiffness at small strains. Properties obtained from reliable static tests are comparable to those measured dynamically

The initial dynamic shear modulus (Gmax) is a key soil parameter for many different geotechnical applications, but it is not clear whether soil behaviour at very small strains is isotropic, linear or rate independent, as is often assumed. It is intended to study these factors, and others in the present project.

During the last decade the Resonant column test has been found the most reliable method for determination of dynamic properties of soil samples in the laboratory The non-uniform stress conditions within cylindrical resonantcolumn samples can be avoided by using a hollow cylinder and this revision has been incorporated in recently developed apparatus. Further feasible improvements include providing a static or cyclic torsion shear test capability within the same equipment

Such an apparatus is being commissioned at I. C. which allows the same soil sample to be tested dynamically and statically over a wide range of strain. from 10^-4% up to 10%

The testing programme for the new equipment includes research on a variety of different soil types, as well as on some artificial model materials.

An alternative method of measuring Gmax, at low strain levels, is- the use of piezoceramic bender elements Installed at opposite sides of a soil specimen as a transmitter and receiver they may be used to measure the shearwave velocity and consequently the shear modulus. Computation of Gmax is simple and direct.

The testing programme, for bender elements, includes Ko consolidation of samples in a 300mm diameter oedometer cell, with particular interest in investigating influence of anisotropy on the shear wave velocity.

Data from resonant column tests, torsional tests, and bender element tests, are to be compared to each other, and with data available from other types of tests.

STRESS STRAIN CURVE FOR ROCKS AND SOILS

Nurnerical methods of solution of many boundary value problems in Rock and Soil Mechanics require an analytical simulation of the required stress strain relationship through the entire range of strains - from zero to the strain at and beyond peak strength. In problems of settlements and displacements of structures, the contribution of the zones with very small strains to boundary displacements can be larger than that of zones of contained failure. Therefore, accuracy of simulation is required for both small strain and close-to-failure regions.

A vast amount of data on small strain behaviour of soils and rocks has accurnulated since the introduction of the methods using local strain measurements (strain gauges, electrolevels, proximity and local deformation transducers). However, in the majority of the standard commercial tests, the resolution of strain measurements is not sufficiently high to give reliable data in the small strain region. Therefore, the problem of stress-strain behaviour simulation by an analytical function should be treated differently for two following cases:

Case I. Small strain data are not available.

The function should provide an acceptable accuracy over the whole strain range using a minimum number of parameters available from the standard test.

Case II. Small strain data are available.

The function should be versatile enough to fit these data, not sacrificing accuracy at large strains.

The key requirements for a stress-strain function are:

1. Lowest possible nurnber of constants consistent with accuracy.

2. The constants to have physical meaning.

3. The constants to be easy to derive.

- Two types of normalized stress-strain curve are presented, depending on availability of the test data in small strain region.

- General conditions for small strain behaviour of soils and rocks are formulated, so that when these conditions are satisfied (as for the first curve), prediction can be quite accurate in spite of the lack of pararneters from the small strain region.

- When the small strain data is available, the second relationship, using three parameters defined directly from the stress-strain curve in the small strain region, proved to be the most accurate amongst the best known models.

- The proposed relationships indicated high accuracy of prediction at small and entire strain ranges for large variety of tested soils and rocks under different loading conditions.

The discovery of previous gravity slides within the areas of reservoir banks or abutments, can affect the design of a dam, and may cause the search for a different dam site area. Ignoring the presence of such slides can lead to disaster involving loss of life and financial damages.

The completed research was concerned with slop instabilities in the area of the Thisavros embankment dam in Greece. Since the beginning of its construction (in 1986) movements of slopes near the abutments have been observed. The most dangerous slides have been stabilised by remedial measures (drainage, toe berms, excavations). Construction of the dam is presently continuing but potential landslide problems still exist.

The following main subjects were addressed by the research:

(i) Collection, classification, and comparison of similar case histories.

(ii) Reconstruction of the Thisavros area slides, using data collected by monitoring and observation.

(iii) Back analysis using well established methods of stability analysis and laboratory techniques.

(iv) Investigation of remedial measures and their effectiveness.

Researcher: J.R. Standing

Supervisors: Prof. J.B. Burland and Prof. D.M. Potts

Sponsors: TRL

Soil nailing is primarily used for the construction of excavations and steepening existing slopes. The term is also applied to the reinforcements used for arresting movements in slopes either before or after failure. This project concentrated on their use in forming excavations.

Nails usually comprise either reinforcing bars installed and grouted in a preformed hole or steel tubes or angles driven in the sides of the proposed excavation. The excavation is formed using 'top-down' staged construction techniques.

Current design practice relies on a Mohr-Coulomb approach for determining the skin friction on nails, with normal stresses estimated from overburden pressures and lateral earth pressure coefficients, and interface friction angles determined from direct shear box tests. Site pull-out tests are usually carried out during construction to verify the values adopted for design. Pull-out test results often indicate much higher frictional stresses than those predicted, this is often attributed to the effects of dilation. Very little is understood about the displacements required to generate interface friction.

The objectives of the research were to investigate the factors controlling load transfer between nail and soil in different soils. This knowledge could then be used to verify or improve the current method of determining the interface friction and implemented into an amended design method where anticipated displacements could also be estimated. The latter feature is important for cases where soil nailing is used in urban areas.

The research focused on an experimental approach using single model nail element tests and a modified Bishop and Wesley triaxial stress path cell. Parametric numerical analyses carried out prior to final apparatus design indicated that for uniform stress conditions along the nail interface, the sample and nail diameters had to be 150 and 8 mm respectively. This equipment was used to carry out tests on kaolin reconstituted from slurry, following a stress path intended to simulate stress relief conditions experienced during excavation .

In order to monitor stresses on the nail interface it was necessary to develop an instrumented model nail. The final nail version was able to measure axial force and radial stress at three positions along the length of the nail using strain gauge mechanisms.

Tests using kaolin were performed using an earlier version nail capable of measuring only axial force, thus enabling interface shear stresses to be determined. Tests were also performed using `dummy' nails so that they could be terminated at different stages and samples prepared for thin-section analyses. This technique is widely used at Imperial College and it is clear that the micro-fabric generated during installation and loading of piles and anchors has a profound influence on the development of skin friction. In this way it is possible to relate the development of stress acting on the nail and clay fabric around it.

An extensive series of tests on three sand sizes (fine to coarse grain) at three densities (loose to dense) were performed in a specifically contrived cell where boundary stresses were controlled in the same way as the effective stress cell (tests were performed on dry sands). The final version model nail was used for these tests. In both sand and clay testing it was necessary to carry out reference tests without any inclusion to assess the comparative effects.

A comprehensive review of current design practices as used in Europe and the United States has been carried out. The test results are being processed with the aim of developing an amended method of analysis for determining the frictional capacity of soil nails and implementing it into current design methods.

ANISOTROPY OF A GRANULAR MEDIUM

Although principal stress rotation is experienced in almost all geotechnical engineering situations, it is usually neglected in design because of the difficulties in (i) measuring and quantifying the effect experimentally and (ii) formulating theoretical soil models that can realistically account for the observed behaviour. The overall objective of this research was to investigate (i) above by performing sophisticated stress path tests using the Imperial College Hollow Cylinder Apparatus (HCA), Fig.1. This device enables a uniquely wide range of stress paths, involving different amounts of principal stress rotation out of the triaxial plane, to be imposed on soil samples. The research was instigated by recent problems identified with the foundation behaviour of offshore structures which experience large principal stress rotations due to both static and cyclic wave loading.

The IC HCA allows a large hollow cylindrical sample (203 mm inner diameter, 254 mm high, 25.4 mm wall thickness) to be subjected to two forces, axial load (W) and torque (Mt), and different internal and external cell pressures (p_i and p_o), Fig.2. These forces and pressures can be independently controlled. Consequently it is possible to control the magnitude and direction of the three principal stresses. The apparatus also allows all strains to be measured locally (axial, radial, circumferential and torsional shear), with a resolution that provides an insight into small strain behaviour.

The laboratory experiments were performed on an artificial quartz based silt called HPF4, whose grading curve covers the whole silt range of the particle size distribution diagram. This material was chosen for two reasons. Firstly, it has properties similar to those of certain sea-bed sediments and consequently the results have direct relevance to offshore situations. Secondly, little experimental information is available for silt soils and therefore the results provide valuable data spanning the information gap between the behaviour of clay and sand soils.

The research consisted of three main sets of experiments. Each set of experiments was designed to investigate a particular aspect of principal stress rotation:

* The general strength and yielding anisotropy of K_o consolidated HPF4. These tests focused on overall behaviour and were compared with previous studies on sands.

* The directional dependence of the stiffness of lightly overconsolidated HPF4. These tests provided information on the true stiffness anisotropy over the full strain range.

* How the anisotropy established under K_o conditions may be altered by consolidation paths in which the major principal stress direction is rotated. The anisotropy of the "new" material formed this way was compared with that of K_o consolidated HPF4.

The main conclusions drawn from the results considered both pre-failure (small strain) and failure (large strain) characteristics of the silt and were: