869
Technical Committee 104 /
Comité technique 104
the resulting uncertainty in the stress and strength within the
vicinity of the column.
Physical modelling is particularly appropriate to investigate
ground improvement considering the large resources required to
undertake field scale testing and the complexity of numerical
modelling: there is complexity associated with modelling of the
reinforcement soil interface and of the process of reinforcement
installation or construction. In some instances, the research
undertaken has led to the development of very sophisticated
testing technology, associated with the generation of the
reinforcement or mechanism visualisation, which have provided
insights that could not have been gained from other
investigation techniques.
Ground improvement by the use of reinforced columns can
be optimised by minimising the length of columns. They may
not need to penetrate entirely through the soft layer.
Physical model tests reported by Tekin and Ergun (2013)
compared the settlement of surface foundations on clay,
reinforced by sand columns of varying length. The experiments
featured a novel miniature extensometer arrangement. This used
an antenna to detect magnets buried at multiple elevations
within the sand columns. The efficiency was shown to depend
on the length of the columns relative to the zone of loaded soil
beneath the surface foundation. Columns that were too short
settled with the surrounding foundation soil. Columns that
extended deeper than the breadth of the surface foundation
showed significantly less settlement. The vertical strain field
measured by the extensometers supported these observations.
The theme of sand columns partially penetrating through
clay continues in the paper by Sadek and Lattouf (2013). They
performed drained triaxial tests on models of sand columns with
varying volume ratio, relative to the full clay sample. The sand
columns did not extend to the base of the clay. Careful
exhumation of the samples after testing allowed the failure
mechanism to be identified. The samples were treated as a
single soil element in the interpretation. By fitting a Mohr-
Coulomb failure envelope to the ultimate loads, the ‘smeared’
strength of the composite element was determined. These
physical models essentially simulate a building block of a larger
network of ground improvement columns.
The previous two studies focused on the in-service behaviour
of a reinforcing column placed within a pre-bored hole in clay.
Gautray et al (2012) focus more closely on the stone column
construction process, through centrifuge modelling. Their aim
was to examine the changes in column geometry and the pore
pressures in the surrounding soil, as the compaction process
evolves. Their tests include the full process of lance insertion –
expanding a cased hole in the clay – followed by a cyclic
retraction process whilst granular material is filled into the hole.
Their data show the loading of the surrounding soil as the lance
is inserted. The changes in pore pressure are significant. They
are precipitated both by the initial insertion of the lance, and
also by the lateral expansion of the granular column by the
vertically-oscillating lance, during extraction.
After completion of the model construction process, the
granular columns were loaded by a model foundation. The
exchange from a column installation tool to a model foundation
was made possible using the independent tool table of the ETH
Zurich drum centrifuge. The model foundation was pushed into
the ground, over the stone column, and the bearing capacity of
the reinforced ground was identified. In this case, analytical
solutions based on previous studies were able to bracket the
identified capacity.
Two further system for studying column-reinforced ground
are reported. Houda et al. (2013) describe a modular
experimental apparatus that has been developed to allow
parametric studies, including cyclic loading events, to be
imposed on improved ground – using combinations of columns
and geosynthetics.
Takano et al (2013) describe a highly sophisticated system
which allows grout columns to be constructed within a X-ray
CT scanner, providing data of the changes in density
surrounding each grout bulbs, as the column is constructed in
increments. The sample container is also instrumented to
calculate any changes in earth pressure coefficient. Subsequent
centrifuge model tests, using a shaking table, demonstrate that
the increased earth pressure coefficient leads to a reduction in
the tendency to liquefy.
An alternative ground improvement system – the use of a
cellular geosynthetic – is explored by Xu and Wang (2013).
They describe investigations into the bearing capacity of
footings on saturated granular soil, with and without geocell
reinforcement. This ground improvement technology is
relatively new development, and is most suited to reclaimed or
filled ground. The geocell is laid on the ground then covered by
fill. The tests showed that the geocell reinforcement serves to
provide tensile capacity within the composite material. This
changes the bearing capacity by altering the failure mechanism.
The settlement around a surface footing becomes more bowl-
shaped, rather than involving a punching shear mechanism. The
punching shear is prevented by the tensile action of the geocell,
which also serves to prevent tension cracks from opening
adjacent to the footing.
These studies into ground improvement reflect the increasing
need for urban developments and transport corridors to utilise
poor ground, requiring mitigation measures to limit settlements.
2.2
Shallow and deep foundations
Several papers report experimental studies into the behaviour
mechanisms of shallow and deep foundations. These focus on
interaction effects – between pile base resistance and tunnels
(Williamson et al. 2013), between pile base resistance and water
jetting (Shepley and Bolton 2013) and between foundation and
soil stiffness (Arnold and Laue 2013).
While loading of shallow and deep foundations can be
undertaken relatively easily at full scale, physical modelling
offers the possibility to investigate complex installation and
interaction processes as illustrated in the examples below. The
development of miniaturised electronics and sophisticated
computer controlled motion systems, has enabled a
continuously increasing realism of the modelling, providing
invaluable insights into problems related to soil-structure
interaction.
The complex experimental arrangement presented by
Williamson et al. (2013) allows three model piles to be
independently loaded, whilst a model tunnel is ‘constructed’
(through the simulation of volume loss) in the ground beneath.
The effects of tunnel construction on overlying piled
foundations is of increasing relevance as new urban railways –
such as London’s Crossrail project – burrow beneath existing
buildings.
During these centrifuge experiments, the soil movements are
watched intently by an array of cameras, allowing displacement
fields to be generated through particle image velocimetry (PIV).
These results allow the full soil deformation mechanism to be
visualised – rather than gathering only boundary movements
measured by instruments located at the edges of the model. The
displacement field extending from the tunnel construction to
disrupt the existing piles can then be observed.
Pile construction can also be disruptive to existing
infrastructure, particularly if dynamic or percussive installation
methods are used. An alternative is to install piles through a
jacking method, and the limited capacity of hydraulic pile
jacking machines can be countered by the use of water jetting to
reduce the penetration resistance. Shepley and Bolton (2013)
describe centrifuge experiments which investigated the effects
of water jetting.
The interaction between water jetting and the penetration
resistance in sands is complex. Jetting serves to locally raise the
pore water pressure at the pile tip, whilst also potentially
causing migration of fines. Both of these effects will ease
