Report on a seminar 'Consolidation of soft ground' held at the University of Sheffield on 15 January 1998, by Charles Hird.
This seminar was sponsored by the Engineering & Physical Sciences Research Council with the aim of disseminating the results of recent research, mainly that funded by EPSRC. It was attended by 79 practising engineers and 13 academics. The level of industrial interest testifies to the continuing difficulty of predicting deformations due to consolidation in soft ground, especially where variable conditions are encountered in alluvial or estuarine deposits.
In the first of four presentations, John Powell of the Building Research Establishment discussed the use of the piezocone in determining consolidation characteristics and stressed the benefits of speed and economy in site profiling. While discussing pore pressure dissipation tests, he showed results from the EPSRC soft clay test site at Bothkennar using several filter positions, Figure 1; these dissipation patterns are in fair agreement with certain theoretical models (Teh and Houlsby, 1991) giving some confidence for the potential of interpretation.
With good technique repeatable results from dissipation tests can be obtained and interpreted within the theoretical framework to give values of the coefficient of consolidation in the horizontal direction, ch. On the other hand, very different results can be obtained in overconsolidated clays, where suctions may be generated during penetration of the piezocone as a result of the high overconsolidation ratio and filter position; similar behaviour can result from a poor deairing technique.
When interpreting dissipation data, a plot of pore pressure versus the square root of time may prove useful in determining the initial pore pressure (by extrapolating the straight line portion to zero time) which is particularly important if interpretation is to be based on a short duration of dissipation. Logging speeds need to be sufficiently fast to capture the early part of the dissipation curve. Experience worldwide (Robertson et al, 1992) has indicated that, at present, it should be possible to determine ch to within half an order of magnitude and then to estimate cv to within one order of magnitude. Where this inaccuracy is too large and other types of measurement are necessary, the piezocone still has an important role in showing up the variations that are present within a profile or changes that have occurred with time.
Darren Russell of Mott MacDonald, a former research student at Sheffield University, discussed the use of finite element methods in embankment design against the background of uncertainties about the soil properties, including strength and secondary compression. The importance of appreciating the risks and applying the observational method was stressed, as well as the usefulness of an initial finite element prediction. The observational method allows for the construction sequence to be adjusted if the subsoil is found to consolidate too quickly or too slowly.
A procedure (Hird et al, 1992 and Hird et al, 1995) for modelling the effect of vertical drains in two-dimensional finite element analyses was then described and illustrated for two embankments at Porto Tolle (Italy) and Whitehall Creek (London). For the second of these, the finite element mesh is shown in Figure 2, to illustrate that only a small number of 'drains' need be inserted in the mesh. Matching of the average degree of consolidation with that which would occur with radial drainage to the vertical drains is achieved by manipulating the permeabilites of the soil. The predictions of both vertical and lateral movements were relatively good, despite being based on limited information, and permitted the observational method to be applied with much larger action trigger levels for lateral movement than would otherwise have been possible (Russell, 1996).
The performance of the approach embankments for bridges constructed in association with the Second Severn Crossing was described by David Nash and Sarah Ryde from Bristol University. The embankments were up to 9m high and were constructed on highly variable laminated estuarine deposits in which vertical drains were installed.
These subsoils behaved very differently from more homogeneous marine clays and excess pore pressures typically remained small throughout construction, ie the foundation response was virtually fully drained. Although settlements of up to 1,750mm occurred, lateral movements at the toe were small (less than 10% of the surface settlement). Due to creep, the settlements are continuing despite the use of a surcharge, Figure 3.
The variability of estuarine soils makes characterisation difficult and in this case history there were significant uncertainties surrounding strength, overconsolidation ratio, coefficient of secondary compression and, most important of all, permeability. Fortunately, reasonably reliable permeability values could be estimated from the performance of the trial embankment constructed nearby at Avonmouth (Murray, 1971).
Finite element analysis of the embankment construction at two locations was carried out retrospectively, employing the above-mentioned procedure for modelling the vertical drains and approximating the longitudinal section as a plane strain problem. Satisfactory results for pore pressure and settlement were obtained, although lateral movements were not accurately modelled. In response to the continuing settlements (Figure 3), a finite difference program was developed to incorporate the modelling of creep both during and after consolidation. This program predicts settlement, on the assumption that lateral movement is negligible, with both vertical flow and radial flow to vertical drains.
In the final presentation Charles Hird described research at Sheffield University on smear effects in laminated soils, ie soils comprising thin alternating layers of clay and sand (or silt). Smear was defined as the disturbance in the soil adjacent to a rigid penetrating object such as a piezocone or the mandrel for installing a vertical drain. Radial flow permeability tests were conducted in small scale physical models in which a central vertical drain had been installed and the results were interpreted in terms of an idealised uniform smear zone. Reductions of horizontal permeability in the smear zone were substantial (an order of magnitude higher than suggested in the literature) and were shown to depend on the thickness of the more permeable sand layers, their spacing and the mandrel driving speed.
Dissection of the models revealed the existence of a complex and asymmetrical fabric around the drain and suggested that the smear was concentrated within a zone of much less than twice the drain radius. In two unlayered (all clay) models the effects of smear were relatively modest and the results were consistent with previous research (Onoue et al, 1991). The results of miniature piezocone tests in the same models were also presented. A 1cm2 Fugro piezocone with a tip filter successfully detected sand layers as thin as 1mm, although the pore pressure response was favoured by the small size of the piezocone in relation to the sand layer spacing (20mm) and the large difference between the permeabilities of the sand and clay layers. Sand layers 4mm thick were easily detected and were also evident in the cone resistance profile, Figure 4. Coefficients of consolidation obtained from dissipation tests suffered from a severe scale effect in the layered models and grossly underestimated the overall value of ch. Values of ch in an unlayered model were in closer accord with theory but still appeared to be underestimated by a factor of about 3, based on the time for 50% dissipation.
The final session was a discussion led by David Hight of the Geotechnical Consulting Group who had kindly agreed to interupt his Rankine lecture preparations. Various experiences regarding the use of the piezocone in the field were related and it appeared that both overestimates and underestimates of ch had been obtained outside the limits suggested by John Powell. No clear explanations for the discrepancies emerged, although the difficulties of interpretation in layered deposits may have been a factor. Given the impact of the rate of consolidation on design, the question arose as to how to proceed in the face of such uncertainty.
The solution advocated by Rowe (1972) over 20 years ago, namely to inspect the fabric of continuous cores (or consecutive samples) and test intelligently selected 250mm diameter specimens, does not seem to be widely implemented because of cost and time constraints. Nevertheless, permeability is of such importance that sufficient resources should be allocated to site investigation and in-situ tests should certainly be conducted. Unfortunately, even then smear effects may adversely affect the results. For example, at the Bothkennar test site results obtained using a self-boring permeameter imported from Canada exceeded the values from constant head tests in push- in piezometers, where smear effects were greater, by a factor of about 2 (Leroueil et al, 1992). For profiling, there are good prospects of obtaining the desired knowledge economically and speedily by combining piezocone tests with a limited number of borings (preferably continuous). The main difficulty may then be in deciding how extensive or continuous the more permeable layers in a profile are.
Another way of covering the risks of an inaccurate prediction of the rate of consolidation, at relatively low cost compared to the total cost of construction, is to install vertical drains. In a layered profile, these can be installed at closer centres if the continuity of the more permeable layers is doubtful. Finite element modelling is now feasible but would it be recommended?
The view was expressed that, despite the limitations of currently available models, it would be worthwhile for major projects involving staged construction and that a practising engineer educated in critical state soil mechanics could learn how to do it, given time. This might be the main problem in a commercial setting. The other common problem lies in obtaining sufficiently reliable input information to justify the effort involved.
The last discussion issue was sample quality in relation to the determination of the preconsolidation pressure. David Hight stated the need to improve on the conventional piston sampling techniques and illustrated the point by showing the results of one-dimensional consolidation tests on samples obtained using superior devices such as the Sherbrook and Laval samplers. In these tests the preconsolidation pressure (or vertical yield stress) was much more clearly defined. The type of test (step loading, constant rate of strain or restricted flow) also makes a significant difference, given good sample quality, and needs to be carefully considered.
Hird, CC, Pyrah, IC and Russell, D. (1992). Finite element modelling of vertical drains beneath embankments on soft ground. Gotechnique, Vol 42, No 3, pp 499-511.
Hird, CC, Pyrah, IC, Russell, D and Cinicioglu, F (1995). Modelling the effect of vertical drains in two-dimensional finite element analyses of embankments on soft ground. Canadian Geotechnical Journal, Vol 32, No 5, pp 795-807.
Leroueil, S, Lerat, P, Hight, DW and Powell, JJM (1992). Hydraulic conductivity of a recent estuarine silty clay at Bothkennar. Gotechnique, Vol. 42, No. 2, pp. 275-288.
Lunne, T, Robertson, PK and Powell, JJM (1997). Cone Penetration Testing. Blackie Academic and Professional, London.
Murray, RT (1971). Embankments constructed on soft foundations: settlement study at Avonmouth. Road Research Laboratory, Report LR 419.
Onoue, A, Ting, N, Germaine, JT and Whitman, R. (1991). Permeability of disturbed zone around vertical drains. In: ASCE Geotechnical Special Publication No 27, pp. 879-890.
Robertson, PK, Sully, JP, Woeller, DJ, Lunne, T, Powell, JJM and Gillespie, DG (1992). Estimating coefficient of consolidation from piezocone tests. Canadian Geotechnical Journal, Vol 29, No 4, pp 539-550.
Rowe, PW (1972). Twelth Rankine Lecture: The relevance of soil fabric to site investigation practice. Gotechnique, Vol. 22, No. 2, pp. 195- 300.
Russell, D (1996). The use of the observational method at the Whitewall Creek embankment. ICE Symposium on the Observational Method in Geotechnical Engineering, Thomas Telford, London.
Teh, CI and Houlsby, GT (1991). An analytical study of the cone penetration test in clay. Gotechnique, Vol 4, No 1, pp 17-34.