Peter Day, University of Stellenbosch and consultant (Jones, Wagener, Pty Ltd), South Africa
Challenges and shortcomings in geotechnical engineering practice in the context of a developing country
Peter Day, a South African consulting engineer and academic, presented the Terzaghi Oration, an excellent paper focussing on the relationship between research and practice in geotechnical engineering, and in particular on the difficulty over transfer of information between the two.
Peter’s Terzaghi Oration focussed on the gap between research and practice. Although our ability to understand and evaluate ground behaviour has improved significantly over the years, the number of publications has increased exponentially to the extent that we are unable to translate all the findings into practice. As a result geotechnical failures tend to be the result of our inability to apply available knowledge in practice, or simply of not recognising critical design situations, rather than a lack of knowledge. The consequence of this is that the profession needs to shift its focus from the creation of knowledge to facilitating its implementation.
A Critical Review of Knowledge Development and Implementation in Geotechnical Engineering
The state of practice and the uptake of research findings is lagging further and further behind the state of the art in geotechnical engineering (particularly in developing countries). This manifests itself in many ways ranging from failure of the most basic geotechnical works to ignorance of recent developments by many practitioners.
Examples were given of four projects in South Africa where failures or unexpected geotechnical issues were encountered, causing escalations in costs. These encompassed well known issues on expansive soils, a slip caused by unfavourably dipping shale beds and inadequate site investigation resulting from programme pressure. Each failure could easily have been foreseen as the knowledge was available, but for one reason or another was not put in to practice.
The proliferation of technical publications is driven by researchers who may not have worked in practice themselves, and there appears to be a preference for research into analytical procedures and laboratory testing as opposed to field experiments, construction methods or engineering practice. The result is an increase in research findings that are not being applied in practice. In addition the rate at which research is being produced far exceeds the rate at which the findings are being incorporated into geotechnical practice. The benefit of geotechnical research can only be realised when it is applied in practice, and for this to happen it needs to be converted into documents that practitioners are able to use and are sufficiently authoritative to rely on. CIRIA, BRE, FHA and GEO are given as good examples of establishments that produce authoritative guidelines extensively used in practice.
Established practitioners are likely to stay with what is familiar rather than embrace new ideas, but the CPD system forces a degree of exposure to new knowledge that the average professional may not otherwise have experienced. However, the uptake and application of new knowledge is dependent on it providing some advantage, and in a cost driven environment it is not necessarily recognised that good engineering provides good value, although that applies less to the larger projects.
Development of Codes and Standards
In the past South Africa has adopted many British standards, but also had regard to North American and Australian standards, the latter due to similarities in climatic conditions. It is currently evaluating options for the development of a local geotechnical design standard, which has led to critical examination of the purposes of such codes. The main purpose is of course that of standardisation, but benefits also include the setting of the norms for the profession, and that they represent a distillation of existing knowledge on which there is consensus. In this sense, codes and standards lag behind the introduction of new developments in the industry.
In many countries the method of design is prescribed by national standards, but countries like South Africa, which do not have a geotechnical design code, face a choice of which design methods to adopt. The four main geotechnical design methods that have been considered by South Africa are working stress design, limit states design, reliability based design and mobilised strength design. A comparative study of these has been undertaken as part of the assessment being carried out on the suitability of reliability based design for routine design purposes and possible incorporation into a South African geotechnical design code. This comparative study was based on the existing Eurocode-compliant solutions and assessed a very limited range of design conditions, but significant conclusions included:
- in limit state design reliability indices are reasonably constant for all the types of structures considered;
- global factors of safety vary significantly across the range of problems and material properties considered, supporting the well-established realisation that Factor of Safety is a poor measure of the reliability of a structure;
- the global factors of safety obtained using characteristic values of material properties are closer to accepted values while those obtained using mean values are significantly higher. This supports the view that there is no significant difference between characteristic values and the “responsibly-conservative” values typically used in working stress design;
- a working stress design prepared to achieve the minimum acceptable Factor of Safety using mean parameter values will have an unacceptable reliability index. This situation becomes worse as f’ increases;
- Reliability analysis is a valuable tool to be used in conjunction with limit states design procedures. The comparisons undertaken suggest that our current limit states design procedures are robust and not far off the mark.
The agreement between the results of limit states design and reliability-based design in the case of simple every-day geotechnical problems is encouraging. The development of simple methods of undertaking reliability-based design calculations opens the way for their use in practice, particularly for problems with explicit solutions. It is clear that working stress design methods are flawed and the use of such methods should be discouraged.
For countries that do not have geotechnical design codes of their own, the adoption of limit states design codes from developed countries remains a good option. Reliability based design can be introduced to complement limit states design. However, the adoption of reliability-based design as the “default” or “preferred” method could present problems with the potential for misuse because of its complexity or abuse on account of the flexibility of the method and the dependence of the outcome on the input parameters.
One of the main challenges in developing countries is the poor quality of laboratory test results. Peter presented data from studies comparing the results of Atterberg Limit, specific gravity and PSD tests carried out in four accredited laboratories in South Africa. The differences between the results are such that they could seriously affect the classification of the material. In tests on five samples of clayey material the plasticity indices from the control tests were 29% to 75% higher than from commercial laboratories. In conclusion, there is no doubt that poor quality data, in conjunction with inadequate site investigations, and the types of test on offer is hampering the uptake on new technologies.
The problem of inadequate site investigation has been with us for years and is unlikely to disappear soon, and competitive bidding of geotechnical work is here to stay. The best that the profession can do is make it as simple as possible for those issuing bids to correctly specify the type and scope of investigation.
The quality of geotechnical data received from testing laboratories remains a concern. It may be necessary to create a new standard for the testing of fine-grained soils distinct from the standards used for road-building materials.
Limit states design is a step forwards from working stress design. Reliability-based design is an ideal complement to limit states design.
There is a concern that the way we teach geotechnical engineering and the way current codes are formulated tends to focus more on design situations involving rupture of the ground than its deformation.
There is a break in the cycle of new knowledge transfer back and forth between researchers/ academics and practitioners. This communication is the responsibility of both parties.
The Ishihara Lecture
Jonathan Bray, University of California, Berkeley,
A Simplified procedure for estimating liquefaction-induced building settlement
The state-of-the practice for estimating liquefaction-induced building settlement still largely involves using empirical procedures developed to calculate post-liquefaction, one-dimensional, reconsolidation settlement in the free-field away from buildings, and neglects the importance of other mechanisms that contribute to building settlement. These free-field analyses cannot possibly capture shear-induced deformations in the soil beneath shallow foundations.
It is useful to categorize movements as ejecta-induced, shear-induced, or volumetric-induced deformations. Observations clearly show that buildings supported by shallow foundations displace downward more than the 1D volumetric reconsolidation liquefaction-induced settlement for the free-field, level ground case. Shear-induced mechanisms should be considered in procedures estimating liquefaction-induced building settlement. When it occurs, ejecta-induced deformation can govern building settlement. The resulting impacts can be devastating and lead to large settlement. The calculated post-liquefaction bearing capacity factor of safety (FS) is also an important index of seismic performance for buildings with shallow foundations situated on liquefiable soils.
The seismic performance of a site should be more influenced by the characteristics of the ground motions at the ground surface than those for the outcropping “rock” site condition. Therefore, the ground motion intensity parameters corresponding to the surface ground motions are preferred in the proposed simplified procedure for estimating liquefaction-induced building settlement.
Some 1308 non-linear effective stress soil-structure interaction dynamic analyses were performed to identify key parameters controlling settlement for this project.
The proposed simplified procedure for estimating liquefaction-induced building settlement involves these steps:
- Perform a liquefaction triggering assessment, and calculate the safety factor against liquefaction triggering (FSL) for each potentially liquefiable soil layer.
- Calculate the post-liquefaction bearing capacity factor of safety (FS) using the simplified two-layer solution of Meyerhof and Hanna (1978). If the post-liquefaction bearing capacity FS is less than 1.0 for light or low buildings or less than 1.5 for heavy or tall buildings, large movements are possible, and the potential seismic building performance can generally be judged to be unsatisfactory.
- Estimate the likelihood of sediment ejecta developing at the site by using ground failure indices such as LSN, LPI, or the Ishihara (1985) ground failure design chart. If the amount of sediment ejecta is significant, estimate the amount of building settlement as a direct result of loss of ground due to the formation of sediment ejecta.
- Estimate the amount of volumetric-induced building settlement.
- Estimate the shear-induced building settlement (Ds) due to liquefaction below the building according to:See the full paper for definition of terms in the equation and further explanation.
- Estimate the total liquefaction-induced building settlement as:
Dtotal = Dejecta + Dvolumetric-induced + Dshear induced
- Use engineering judgment.
Some important limitations of the procedure which should be
- The equation in 5 above was developed using a subset of potential building configurations and earthquake ground motions. The structures considered are regular (e.g., uniformly loaded) and have heights no greater than 24 m.
- The non-liquefiable crust does not have defects (e.g., utility trenches that could provide preferential paths for ejecta).
- Some volumetric induced liquefaction building settlement occurs during strong shaking, but this procedure categorizes all of this settlement as being due to the shear-induced mechanism, which is conservative.
- Case histories and previous experience are important to consider in developing the final engineering assessment of this complex problem.
- For important projects, perform nonlinear dynamic SSI effective stress analyses.
The performance of this methodology was tested for several well-documented case histories from the Kocaeli and Canterbury earthquakes.