Report on the BGA and Géotechnique lecture, From laboratory testing to design in sand, held at the Institution of Civil Engineers on 8 October 2003, by Duncan Nicholson, Arup.
The BGA's annual Géotechnique lecture is an opportunity for authors who have published outstanding papers in the journal Géotechnique to present their work to a broader base of engineers.The 2003 lecture was given by Dr Matthew Coop of Imperial College, London, who presented some of his 17-year research on sand behaviour at London's City University and Imperial College.
It is not possible to review the full range of topics covered by Coop, so this report focuses on the recent developments in pile behaviour.
In 1982, at the Rankin Offshore platform off the north west coast of Australia, piles were driven into carbonate sands. Pile capacity was found to be much lower than had been expected from the American Petroleum Institute (API) design procedures developed from experience of working in quartz sand.
Coop compared the behaviour of a carbonate sand, Dogs Bay Sand (DBS), similar to that at the Rankin platform, and a typical quartz sand from Leighton Buzzard (LBS). Soil models were presented which explained this low capacity behaviour. These models are being developed and applied to other sands.
Centrifuge modelling of jacked piles To investigate the reasons for the difference in pile performance, centrifuge tests were carried out by Klotz & Coop (2001). Table 1 summarises the properties of the sands.
The geometry of the centrifuge model pile tests is shown in Figure 1. The 16mm diameter model pile in the centrifuge at 100g simulated a 1.6m diameter prototype pile. The centrifuge samples were prepared by pluviation techniques at loose, medium dense and dense conditions. During the test, the model pile was jacked into the sand and the base resistance, shaft friction and radial stresses were measured at the locations shown in Figure 1.
Typical results for the unit base resistance are shown in Figure 2a for the quartz sand and Figure 2b for the carbonate sand.
The quartz sand showed significant increases in unit base resistance with increasing depth and relative density (Figure 2a). A medium dense overconsolidated sample was prepared by consolidating the sands to higher effective stresses (about 3MPa) and then unloading and allowing the sample to swell. The soil profile produced showed reduced overconsolidation ratio with depth.This sand had similar unit base resistance to the medium dense sand. The behaviour was similar to that predicted by the API design method.
The carbonate sand had lower unit base resistances than the quartz sand and did not show significant increase in resistance with depth or density (Figure 2b). However, the overconsolidated carbonate sand showed the highest base resistance at shallow depths (although this capacity reduced with depth).
The API design method overestimated the carbonate sand pile capacity.
Similar behaviour was encountered with the shaft friction performance. Coop also discussed the effects of pile roughness.
Laboratory triaxial tests were undertaken on the sands, in particular to understand the variation of specific volume (v = 1+e, where e is the voids ratio of the soil), with mean effective stress (p'). Figure 3 shows the critical state models.
The 'normal compression line' (NCL) and the 'critical state line' (CSL) are shown in Figure 3b for the carbonate sand. The CSL for the quartz sand is shown in Figure 3a. The carbonate sand had a maximum specific volume at critical state of about 2.85 at mean effective stresses under 100kPa, whereas the quartz sand had a maximum specific volume at critical state of about 2.0 at mean effective stresses under 1,000kPa. The high specific volume for the carbonate sand reflected the angularity of the particles and the presence of hollow shells.
Both sands exhibited the 'normally consolidated' and 'overconsolidated' behaviour often associated with clays. The carbonate sand was more compressible than the quartz sand and showed a rapid drop in specific volume with an increase in mean effective stress (Figure 3b). The volume changes in the normally consolidated condition reflect the fracturing of particles, proven by the increase in fines content with increasing effective stress.
Sand behaviour during shearing is controlled by its initial state relative to the CSL. At states below the CSL, the soil will dilate when sheared and will have a peak strength. For states above the CSL, the soil will compress and be strain hardening. The stiffness will also depend on the state, and for a given stress level the stiffness will increase with distance from the NCL.
Coop has defined the State Parameter (Rs) = p o/p cs where p ois the insitu state mean effective stress and p cs is the critical state mean effective stress at the same specific volume (Figure 4).
Others defined this parameter as the difference in specific volumes between the current state and the CSL at the same mean effective stress. Coop prefers the stress ratio definition because it avoids the problems of defining critical state at low stress levels. However the stress ratio method does require triaxial testing to very high effective stresses to fully define the CSL.
With the exception of the overconsolidated samples, the initial states for the samples in the centrifuge tests had been reached by different degrees of compaction during sample preparation, followed by a monotonic increase in stress during the spin-up of the centrifuge.
Although the samples were at their maximum stress (Figure 3), the insitu states were mostly to the left of the CSL.
State specific volume and mean effective stress derived from the centrifuge model (Figure 1) is plotted on Figure 3 at four depths in the model pile tests. Mean effective stress increased with depth, and the specific volumes reduced slightly towards the base of the model. The dense samples have lower specific volumes appear 'over consolidated' They plot further from the CSL and have high Rs ratios.
The centrifuge test in dense quartz sand (Figure 2a) shows high unit base resistance (qb) which increases with depth because the insitu stresses are well away from the CSL (Figure 3a). The sand dilates and provides additional capacity. The loose quartz sand is close to the CSL and sand would dilate less when sheared, hence unit base resistance is lower and shows a slower increase with depth.
The centrifuge pile test in dense carbonate sands (Figure 2b) shows a unit base resistance (qb) that does not increase with depth.
This is because the insitu stresses even for dense carbonate sands are close to the CSL (Figure 3b). The sand either dilates slightly or compresses and when loaded, depending on the initial state and cannot provide additional base resistance.
The bearing capacity factor (Nq) is (qb/s v9)where qb is the unit base resistance and s v9 is the vertical effective stress. The plots of Nq against Rs are shown in Figure 5a and Figure 5b for the quartz and carbonate sand respectively. These show unique relationships except for the overconsolidated samples.
This shows overconsolidation in sand has a very significant effect on the bearing capacity. Soil density and friction angle are not sufficient to define bearing capacity but mineralogy has a significant effect.
Centrifuge testing work also has direct application to CPT cone penetration tests. Stress ratio modelling can be interpreted from a wide range of quartz sands. It is anticipated the mineralogy and roughness as well as relative density will have to be considered when interpreting results.
Work is also ongoing to interpret soil stiffness from the unload-reload loop of pressuremeter tests.
At present pile tests are undertaken to assess the capacity for a particular site. However there is a need to carry out suite of appropriate laboratory tests to get the most out of the tests and extend the framework soil modelling for sands.
Coop MR (1990). The mechanics of uncemented carbonate sands, Géotechnique , 40, No 4, 607-626.
Coop MR and Atkinson JH (1993). The mechanics of cemented carbonate sands, Géotechnique, 43, No 1, 53-67.
Lee IK and Coop MR (1995). The intrinsic behaviour of decomposed granite soil, Géotechnique, 45, No 1, 117-130.
Cuccovillo T and Coop MR (1997). Yielding and pre-yielding deformation of structured sands, Géotechnique, 47, No 3, 491508.
Jovicic V and Coop MR (1997). The stiffness of coarse grained soils at small strains, Géotechnique, 47, No 3, 545-561.
Cuccovillo T and Coop MR (1999). On the mechanics of structured sands, Géotechnique, 49, No 6, 741-760.
Klotz EU and Coop MR (2001). An investigation of the effect of soil state on the capacity of driven piles in sands, Géotechnique, 51, No 9, 733-751.
Klotz EU and Coop MR (2002). On the identification of critical state lines for sands, Am Soc. Test. Materials, Geotechnical Testing Journal, 25, No 3, 289-302.