Meeting chairman Alan Powderham of Mott MacDonald set the scene for the audience of academics, consultants and contractors by asking:
Is modern soil mechanics sufficiently focused on design?
Should geotechnical engineering follow a predictive or preventative route?
What is the place of the theorist in this most practical of professions?
Three presentations were made before the question was debated.
Dr Brian Simpson, Arup Geotechnics
Simpson rephrased the question: 'Have recent developments in fundamental understanding of soil behaviour, or new analytical techniques, had a significant impact on our decisions about what will be built?' Would they have a significant impact in the foreseeable future?
In support of his affirmative answer to both questions, Simpson covered a range of topics:
critical state concepts - understanding safe states
emphasis on particle crushing and its effect on strength and stiffness
finite element analysis
chemistry of contaminated soils
pressuremeter and piezocone analysis
mechanics of reinforced soil
intelligent application of statistical methods
relating displacements to mobilised strength
Simpson went on to describe some of the above. The 400m Millennium Tower, planned for construction in London, will have 70m long base grouted piles founded in the Woolwich and Reading beds. The very high loads which will be carried by the piles will take into consideration the potential for particle crushing.
The analysis of complicated geotechnical systems, in particular three dimensional situations, can be analysed using numerical methods. Simpson showed an example of the analysis of a tunnel intersection (Figure 1) where a three dimensional finite element analysis was used to predict stresses and deformations.
He discussed advances in the understanding of the small strain stiffness of overconsolidated clays and described his bricks on strings model (Simpson, 1992). The model can be used to provide realistic predictions of the behaviour of stiff clays, such as London Clay, and has been calibrated against case histories including the British Library. Such models can also be used to predict settlements above tunnels, and the importance of stiffness anisotropy was emphasised. Not all the questions have been answered with regard to soil modelling, but advanced constitutive models coupled with numerical analyses can be used to aid decision making as has been the case for the magnitude of deformations caused by the Westminster box at the Jubilee Line in London.
While Simpson believes modern theoretical soil mechanics have aided design, he has some concerns:
Robust designs must still be achieved.
Simple theory must be taught to allow engineers to design, but the engineer must be aware of the limitations of such theories.
Design engineers must not lose contact with field measurements and there must be a continuous loop of field observations and theoretical development.
Dr Ken Fleming, Kvaerner Cementation Foundations
Fleming discussed the difficulties facing the geotechnical designer from the poor quality site investigation data that was normally available. Inadequate index testing, crude sampling, inadequate borehole depths, insufficient water records, SPT tests carried out in blowing holes, inappropriate laboratory testing and poor interpretation were highlighted. Fleming speculated that some of these problems may be due to lack of money resulting in the lowest contractor's tender being selected and a lack of exploratory holes.
A typical problem facing designers was where pile design was based on soil conditions before pile installation, not after. Fleming suggested this was only the start of the problem: there was a lack of basic understanding of soil mechanics. Soil mechanics should be considered as a branch of materials science and should draw on the knowledge of other materials such as steel concrete and textiles. He discussed the lack of understanding of the behaviour of soil with time and suggested that more research was required to look at the behaviour of soil within the stress/strain/time axes.
Soil deformations are usually split into instantaneous, consolidation, secondary and creep deformations. Fleming suggested that these distinctions are muddled and may not need to be. The use of linear fractional (hyperbolic) methods was put forward as perhaps more than traditional methods to predict all phases of soil deformation (Fleming 1992). An example of its use to analyse settlement of the Monadnock building in Chicago was shown (Figure 2, Skempton et al, 1955).
Fleming highlighted other areas where he considered there was a lack of understanding and inadequate attempts to gain understanding:
What deformations occur when pore water pressures have fully dissipated?
Dilation along shear planes is an important subject but poorly understood.
What happens to pore pressures on a shearing plane and how does this modify effective strengths?
Do we really know what governs peak and residual strengths; do they really exist or do we produce them by our methods of testing?
Soil behaviour is non-linear and the strength of soil depends on the method of testing. Is it possible to get sensible stiffnesses from laboratory tests which have totally different failure mechanisms from those observed in the field?
Fleming concluded by suggesting that although we now had the ability to acquire a vast amount of data, this had not been properly used. 'It is essential to rethink our fondly held theories. Soil mechanics need to return to its roots, which are firmly in materials science. We have travelled too far along paths of sequential thought without going back to verify assumptions. The answer to many problems will not be found by tinkering with existing theories but in broader understanding and original thought,' he said.
Malcolm Puller, consulting civil and geotechnical engineer
Puller revisited the 1939 James Forrest Lecture by Karl Terzaghi, when the latter reviewed the previous 25 years by considering five topics:
Earth pressure on lateral supports. Coulomb's law had been misused for 150 years. Terzaghi introduced pore pressures to Coulomb's law and changed the angle of repose to the angle of shearing resistance dependent on relative density. Terzaghi considered seepage pressure, curved failure surfaces, active and passive zones and the influence of wall deflections.
The stability of slopes. Terzaghi referred to work by Fellenius and Olsen and considered circular failure surfaces, tension cracks, mobilised shear strength, estimation of the angle of shearing resistance from shear tests, flow slides, liquefaction and critical density.
Piping in dams. Terzaghi discussed the failure of dam foundations due to piping and considered the critical hydraulic gradient downstream and the design of filters.
Settlement of foundations. Terzaghi reviewed his consolidation theory and considered prediction of stresses by Boussinesq's theory, consolidation of clay layers at depth and the effect of underdrainage.
Piled Foundations. Terzaghi considered pile foundations reviewing the use of pile driving formulae. He referred to pile group effect, redistribution of load within the group and the effects of structural stiffness.
Puller suggested a sceptic might feel that soil mechanics had not made as much progress in the period since 1939 as it had done in the preceding 25 years. He examined two of Terzaghi's topics in more detail: earth pressure and pile design.
Earth pressure on lateral support
Various methods are available to the designer to assess lateral pressures: limiting pressures calculated by hand which are identical to those used in 1939; programmes introducing soil and wall stiffness based on springs; or BS8002 which uses mobilisation factors to compute serviceability earth pressures irrespective of wall stiffness. Puller suggested that the temporary works designer found little help in modern soil mechanics, in particular to decide whether to design for long or short term conditions. The temporary works engineer designing propped or anchored systems also has problems: Eurocode 7 does not differentiate between temporary or permanent works and if design is done by hand then methods originated by Terzaghi (1939) are often used. If a numerical analysis is used then the designer must make assumptions as to the soil stiffness and does not have as much confidence in the results as when using the empirical Terzaghi and Peck envelopes. If cohesion is ignored in design, excessive embedment lengths are calculated.
Terzaghi described the use of empirical pile driving formulas. Puller suggested that these methods are still used in practice. In terms of bored pile designs, empirical equations are still used. Effective stress design is available but based on empirical rules.
Puller voiced his concern that the lack of precision of modern geotechnical design, perhaps due to the lack of support from modern theoretical soil mechanics, can distort the use of the observational method as proposed by Peck (Terzaghi and Peck,1948). The method appears to have become a way of remedying shortfalls in design and inherent waste.
Perhaps the most severe test of geotechnical design is in unfamiliar soils. Puller described a recent project in Medinah, Saudi Arabia, observing that while the job did not materially gain from the application of modern soil mechanics theory, it should have done, and it was not for the want of trying.
The job was to construct an unobstructed 17.5m deep excavation in downwash silts and silty sands on three sides of the Holy Mosque in Medinah. Each side was 500m long and the excavation, which was to house an underground car park, was 100m wide (Figure 3). No movement of the excavation was acceptable to the client. The design was a cellular diaphragm, involving 320,000m2 of wall. Several design problems were encountered:
estimates of ground movements due to diaphragm wall excavation (finite element analysis overpredicted movements)
estimation of ground movements due to the bulk excavation (finite element analysis produced incorrect deformed wall shapes even though small strain testing of the silts had been carried out in the UK. The wall stiffness was found difficult to estimate)
accurate assessment of the stresses in the wall (only low stresses were predicted)
accurate assessment of the magnitude and rate of heave of the excavation base (prediction of the rate and extent were excessive)
assessment of the horizontal movement of piles below the mosque
design of vertical rock anchors
design of bearing piles with sleeves to accommodate heave of the formation
Each of the above was solved by reliance on practical experience with resulting overdesign. A full scale excavation trial reduced the reliance on soil mechanics by providing data to ensure that excessive deformations did not occur.
The principal disappointment was the failure to accurately predict the deformed wall shape and when the shape was revised the magnitude of movement was overpredicted. The rate and magnitude of heave were completely wrong. Additional reinforcement was included in the wall design as there was little confidence in the numerical prediction of bending stresses.
Puller concluded that theoretical soil mechanics did serve geotechnical design 60 years ago but rarely does now.
A point raised by Arthur Penman, and echoed by others, was that the amount spent on site investigation was often insufficient and the necessary data for geotechnical design was therefore not obtained. Undisturbed samples could not be obtained and therefore there should be a greater reliance on insitu testing. Fleming agreed that the use of insitu testing was preferable to using inaccurate laboratory testing. Penman also initiated a discussion of the value of the coefficient of earth pressure (K) which should be used in retaining wall design. Simpson emphasised that the value of K should provide a design that was safe and Puller speculated on the effect of wall construction on the value of K.
Chairman Alan Powderham initiated a discussion on the roles of and interaction between research and industry. John Findley suggested that perhaps the perhaps the academic community needs to move back into the field. David Twine suggested that theory was now well advanced and it was now necessary to test it against high quality observations. This data collection was not an activity for academics alone; large amounts of data was collected by industry and not properly disseminated. Perhaps databases of geotechical observations should be developed for use in research and practice.
Duncan Nicholson made a brief contribution describing recent work for the CIRIA publication on the observational method (CIRIA Report RP257). He believes soil mechanics helps in design but can never eliminate uncertainty. The observational method provided a mechanism for dealing with the inherent uncertainty of geotechnical design. Powderham endorsed this view, saying that soil mechanics provided better predictions for use in the observational method. In turn the observational method also provided high quality case history data which could be used to test theory.
Professor Willaim Powrie defended criticism of recent research discussing the collection and collation of data. He reminded the audience that the extension of existing theories to new ground conditions was of value and that a small part of the money that could be saved in construction costs by using new soil mechanics theories would fund a large amount of research. He closer links between academia and industry.
Dr Ian Farmer listed some of the important issues for industry including soil anisotropy, time effects and brittle and ductile behaviour. It was important that theory was used correctly and perhaps soil mechanics could learn from other material sciences. Perhaps the answer was to have better links between practice and research.
Chris Raisson felt theoretical soil mechanics had benefited large projects where good quality site investigations were carried out and soil parameters were available. Theoretical soil mechanics was of less help to smaller projects where site investigation data might be inadequate.
Dr Dickie Basset suggested that critical state soil mechanics theory had been a benefit to geotechnical design by providing a framework which linked volume changes and shear strength. However, theory should be used as a final check of the adequacy of design. It was difficult to obtain bulk material properties for use in theoretical analysis and often conservative parameters were used.
Written contributions received after the meeting are summarised below.
Professor David Muir Wood, University of Bristol
Muir Wood rephrased the question as 'Does geotechnical design take advantage of theoretical soil mechanics?'. Some practitioners make use of theoretical soil mechanics, but for much of the industry the answer is no. He proposed several reasons for this.
Perhaps engineers are not educated correctly, with too much emphasis on facts rather than understanding through thoughtful scepticism.
Much current research is on deformation characteristics and prediction and observation of the performance of geotechnical structures. Education concentrateson strength and failure and simple elastic behaviour. Engineers need to have a feel for the factors which affect serviceability calculations so the applicability of traditional design methods can be assessed.
Academics need to disseminate their research findings more effectively and practitioners need to be more prepared to learn about new developments.
Theoretical soil mechanics on its own may not lead to good geotechnical engineering, but innovative design is likely to be underpinned by modern theoretical developments. Understanding of and confidence in complex analysis is usually helped if the results can be supported by back of envelope calculations. Improvement of the links between theoretical soil mechanics and geotechnical engineering requires education, but education requires industry to recognise that it needs to be educated.
Dr Trevor Addenbrooke , Imperial College
Recently completed research at the Soil Mechanics Section of Imperial College, London, has led directly to the publication of design charts which indicate the modification a structure might make to the settlement above a tunnelling operation (Potts and Addenbrooke, 1997). This work was sponsored by London Underground because it saw a need for developments in this area, and felt confident that advances could be made by funding of university research. The programme of research involved sophisticated finite element analyses, one of Simpson's highlighted new analytical techniques of tunnel construction in stiff clay, and the influence of existing surface structures on the induced settlement. The analyses incorporated non-linear stress strain behaviour pre-yield, one of Simpson's developments in understanding of soil behaviour. The results of more than 100 analyses of different geometries and building stiffnesses have been combined to produce two design charts. The modern theoretical soil mechanics involved in the research have thus been disseminated back to industry in the form of readily applied, simple design charts which aid in the assessment of building damage due to tunnelling. This is just one example of the very important role which high level numerical research can play in geotechnical design.
Ron Williams, Mott MacDonald
Theoretical soil mechanics cannot be viewed in isolation but rather as one of three elements informing the judgement required for an holistic approach to geotechnical design. A thorough understanding of both engineering geology and precedent experience is of equal if not greater importance.
Unfortunately, the expectation of certainty raised by proponents of theoretical developments has led to considerable disillusion among practitioners when the shortcomings are revealed. However, this should not distract from the substantial strides that have been made and continue to be made. A recent example is the light shone on the failures of laying large diameter tubular steel piles by the theoretical developments of Richard Jardine and his colleagues at Imperial College (Williams, Chow and Jardine, 1997).
In the UK there is a welcome trend for closer liaison between university researchers and geotechical practitioners, but much remains to be done on both sides in developing a culture of collaboration - not least the question of increasing the investment in R&D by the construction industry at large, which lags behind many of our industrial competitors (Department of the Environment, 1996).
Department of the Environment (1996). The funding and provision of research and development in the UK construction sector.
Fleming, WGK (1992). A new method for single pile settlement prediction and analysis. Geotechnique, Vol 42, No 3.
Peck, RB (1969). Advantages and limitations of the observational method in applied soil mechanics. Geotechnique, Vol 19, No 2.
Potts, DM, and Addenbrooke, TI (1997). A structure's influence on tunnelling induced ground movements. Proc Instn Civ Engng, Geotechnical Engineering, due for publication in April 1997.
Simpson, B (1992). Retaining structures: displacement and design. Geotechnique, Vol 42, No 4.
Terzaghi, K (1939). James Forrest Lecture.
Terzaghi, K and Peck, R B (1948). Soil mechanics in engineering practice. New York, John Wiley and Sons.
Williams, RE, Chow, F C and Jardine, R J (1997). Unexpected behaviour of large diameter tubular steel piles. International Conference on Foundation Failures, Singapore.