Report on the BGA meeting Long term performance of an embankment on soft organic soil stabilised by soil mix columns, held at the Institution of Civil Engineers, London on 11 May 2005, by Jerry Love, GCG.
The meeting examined recent research into the long term performance of embankments founded on soft organic soil stabilised by deep soil mix columns, which formed part of a multinational study programme known as EuroSoilStab.
The EuroSoilStab project ran from 1997 to 2000 (GE May 2002).
It led to publication of a design guide summarising best practice for stabilisation of soft organic soils based on experiences at six test sites across Europe. Two were in Finland, one in Sweden, one in the UK and two in the Netherlands, involving many different partners across Europe.
The work in the UK, which involved deep wet mixing, was principally carried out by Building Research Establishment and Keller. Tony Butcher of BRE began by describing the different types of mixing process. In deep dry mixing, which originated in Scandinavia, paddles are rotated into the ground driven by compressed air in the typical range 5 bar to 15 bar. All the equipment, including the silo, is mounted on a single rig.
This process can be combined with the shallower mass dry mixing process, which is appropriate for mass stabilisation of soft ground to depths of 3m to 4m. For this, the equipment, looking a little like an oversize outboard motor propeller, is mounted on the end of an excavator arm. Ground as soft as 5kPa undrained shear strength may be stabilised in this way.
By contrast the deep wet mixing process, which is mainly used in Japan, involves a separate colloidal mixer in combination with a slightly larger tracked rig. Mixing occurs both on the way down and on the way up, as liquid is injected just above the rotating paddle.
BRE's work was carried out by the BRE at a test site between the River Darent and Dartford marshes, just east of London. Ground conditions were 6m of very soft organic clay.
The clay had an organic content of up to 80%, a moisture content of up to 400% and a unit weight as low as 10kN/m 3. The undrained strength of the clay typically lay in the range 10kPa to 20kPa, although a thin surface crust existed of higher strength material (Figure 1).
Given that lime does not perform well as a binder in organic material, a principal aim was to investigate whether alternative waste products could be incorporated into the binder.
BRE carried out laboratory tests to identify a suitable binder design. The binder had to cope with the effects of high sulfates (3g/litre to 6g/litre), low pH (3.5 to 5.5), high organic contents and a target strength gain of up to 100kPa to 150kPa.
The high sulfate content and low pH both pointed towards use of an increased slag or pulverised fuel ash content in the cement and an increase in the dosage of reactive solids.
The tests showed the most suitable binder to be a mixture of Ordinary Portland Cement (CEM I) and granulated ground blast furnace slag, using a total binder content of 350kg/m 3 mix with a water/binder ratio of 0.5.
This gave slower strength gain, but higher overall strength with time.
The trial embankment was designed to explore varying column diameters (0.9m to 1.2m), varying depth of columns (either full depth or partial depth) and different methods of construction of the columns.
A column spacing of 2m centre to centre was used throughout, except for one part of the trial embankment which had no columns beneath it, to act as a control section (Figure 2).
The maximum embankment height was 3m and the depth of soft organic soil below the embankment was 5.5m.
Using fi nite element analysis BRE predicted settlement of 560mm for the unimproved section, and settlements from 200mm to 300mm for the sections with columns. Instrumentation comprised surface settlement points, inclino meters, magnet extensometers, piezometers and earth pressure cells both on the columns and on the soil between them.
Long term monitoring of the trial embankment revealed the results in Figures 3 and 4. The untreated section settled by about 900mm in the long term. By comparison the sections with columns settled in the range 450mm to 550mm. The partial depth columns provided the least improvement, followed by the 900mm diameter full depth columns, with the 1.2m diameter, full depth columns providing the most improvement as expected.
Maximum lateral movement at the toe of the untreated section (AA) on Figure 4 was about 140mm.
Similar movement was also seen at the section (BB) where partial depth columns were used. But this reduced to 90mm at the section (CC) where full depth 1.2m diameter columns were placed.
The secondary consolidation coeffi cient, c , was seen to reduce slightly in the treated sections.
Some of the columns were later exposed to inspect the degree of mixing achieved (Figure 5). The soil was not found to be mixed to a homogeneous material.
Samples were also taken to measure pH, sulfate levels and c u values, both in the soil between columns and from within the column itself (Figure 6). The value of c u was raised to a maximum of around 60kPa, rather than the 100kPa to 150kPa hoped for.
Analyses by X-ray diffraction and scanning electron microscope were also done. The X-ray diffraction showed increased levels of calcite and some ettringite, as expected.
Discussion Independent consultant David Greenwood suggested a higher strength gain might be achieved by using a lower water cement ratio if a plasticiser was used, and that going down and up more than once might achieve better mixing.
Asked how other dry and mass mixing methods compared, Butcher said these were found to be just as inhomogeneous as the wet mix, but this did not appear to affect performance.
Tony Bracegirdle of GCG asked whether temperature had been measured, since very high temperatures can occur during jet grouting for example, and this may have affected the instrumentation.
He also questioned whether pH might revert in the long term.
Butcher said temperature had not been measured and the pH had not changed in the four-year trial.
In response to a question about monitoring during installation, Butcher said rates of material delivery, rotation, penetration and withdrawal should all be monitored.
Another questioner queried the apparently small difference between the strengths measured in the soil and in the columns. Butcher said the strengths were measured by hand vane and may not be truly representative since strong grout pockets would not be penetrated.
David Baker of Balfour Beatty said it had unsuccessfully tried the wet mix process on CTRL, but the dry mix process had worked.
John Chantler of Pell Frischmann queried whether reducing settlements from 900mm to 500mm was useful, since the latter was still a lot of settlement.
A representative from Mott MacDonald said that on a job last year it had successfully achieved c u of 400kPa and good homogeneity using dry mixing.
Butcher said the work described at the meeting dated back to 1997 and the technique had probably improved significantly since then.