2421
Technical Committee 211 /
Comité technique 211
for structural applications. Standardized guidelines for the
design of this kind of applications are not currently available. If
the previous work allows the construction of standardized and
international test procedures for laboratory mix samples, the
Quality Control (QC) of the execution process is generally
based on the results of Unconfined Compressive Strength
(UCS) tests performed on cored material. As part of the semi-
probabilistic design approach presented in Eurocode 7, it is thus
essential to define the UCS characteristic value that can be
taken into account in the design.
Denies et al. (2013)
discuss
the definition of this value. In the first category of approaches,
the characteristic strength is defined as an X% lower limit value
computed either on the basis of a statistical distribution function
or based on the cumulative frequency curve of the original
experimental dataset of UCS values obtained from tests on
cored samples. A second approach to determine the UCS
characteristic value is the use of the average value of the dataset
in combination with a safety factor. For the first category of
approaches, a value for the X% has to be defined. Actually, one
major issue is the representativeness of the core samples with
regard to the in situ executed material. For the purpose of
investigating this question, the authors present the results of a
study on the influence of soil inclusions and then they discuss
the topic of the scale effect with regard to large scale UCS tests.
The following papers concern the investigation of the
mechanical properties of the soil mix material under the field of
laboratory or in situ experiments and with the help of numerical
modeling.
In such a way to investigate the properties of the soil mix
material,
Szymkiewicz et al. (2013)
have carried out a
parametric study on lab soil-cement mixtures. The influences of
the particle size, the clay content and the water content on the
strength of the material were considered. They propose an
abacus relating the UCS of the specimens to the cement content.
Six zones are identified in the abacus depending on the nature
of the soil. In addition, the authors also propose a formula valid
for granular soils for the estimation of the UCS at 28 curing
days. This formula takes into account the water, the cement and
the fine contents.
In a similar way,
Correia et al. (2013)
have performed
laboratory tests to study the improvement of soft clayey silt
with high organic content by mixing it with a binder made up of
75% Portland cement (PC) and 25% blast furnace slag. They
first give a formula for the assessment of the UCS at 28 days in
function of the binder content and the liquidity index (LI) of the
soil. A normalized UCS is then introduced as follows:
UCS
LI
= UCS x LI. In a second step, the applicability of the
normalized UCS approach is analyzed for seven other cement-
stabilized soft soils with successful result.
If the water/cement (w/c) ratio is often used in attempt to
understand soil-mix properties, it can be found limited since in
practice execution is mostly performed in soils in the presence
of water (unsaturated or saturated conditions). A well-adapted
governing parameter could be then the porosity/cement index
defined as the ratio of porosity to the volumetric cement content
(n/C
iv
).
Rios et al. (2013)
highlight the influence of this index
on the mechanical properties of cemented Porto silty sand.
Unique trend was obtained between the UCS and an adjusted
porosity/cement ratio (n/C
iv
0.21
), proposed by the authors.
Similar observation was also made with indirect tensile
strength. Triaxial tests resulted in two peak strength envelopes
for each predetermined (n/C
iv
)
0.21
and finally, oedometer tests
establish this ratio as the governing parameter of the behavior of
the soil-cement specimen in one-dimensional compression in
lieu of the cement content or the initial void ratio.
A major advance in DMM could be found in the contribution
of
Yi et al. (2013)
with the investigation of the carbonation of
reactive magnesia (MgO) for soil stabilization. Nowadays,
Portland cement (PC) is the most common binder used in the
deep mixing applications. However, there are significant
environmental impacts associated with its production in terms
of high energy consumption and CO
2
emissions. In their
laboratory study, reactive MgO was used as a binder and the
MgO-soil samples were carbonated by CO
2
to improve the
mechanical properties of the soil and reduce the CO
2
emission.
As an evident result, the UCS values of the uncarbonated MgO-
stabilised soils were much lower than those of the PC-stabilised
soils; both mixes took ~28 days to finish most of their strength
development.
Nevertheless,
the
carbonation
process
significantly increased the UCS of MgO-stabilised soils in a
very short time, this latter fast reaching the UCS value of the
28-day PC-stabilised soils, indicating that it could be used to
support a structure just after the completion of the carbonation
procedure.
Another type of binder largely used for soil stabilization is
lime.
Mesri and Moridzadeh (2013)
discuss the results of a
laboratory study focusing on the improvement of the Brenna
clay (high plastic lacustrine clay of North Dakota) by adding
lime. Lime contents varying between 3 and 10 % of the dry
weight of the clay have been considered. The authors observed
a decrease of the measured pH with time and an increase of the
Liquid Limit and the Plasticity Index with time when 5 % of
lime was added. Adding 3 to 8 % of lime, the residual friction
angle (in drained conditions) increases between 3 to 6 %.
Unfortunately the laboratory test results were not compared
with full scale test results.
Extensive laboratory tests have been performed by
Szendefy
(2013)
for the purpose of determining the effect of lime
stabilization on 21 Hungarian clayey soils. In addition, some in
situ stabilized soils have also been analyzed. According to his
study, the improvement of the clayey soil with the lime is
mainly related to the coagulation of the clay particles related to
the cation exchange. Indeed, during the stabilization with lime,
Ca
2+
ions attach to the surface of clay particles. As a result of
this high charging, the clay particles coagulate resulting in a
material characterized by an increased internal friction angle.
The pozzolanic reaction would play then a secondary role in the
stabilization.
Soil stabilization can also be performed with fiber
reinforcement, such as discussed in
Madhusudhan and Baudet
(2013)
. In their study, laboratory tests have been performed to
determine the influence of adding polypropylene fibers on the
shear strength characteristics of completely decomposed granite
(CDG). In Hong Kong, CDG is regularly used for landscaping
and as green cover of existing shotcrete slopes. The test results
clearly indicate an important increase of the UCS when adding
0.5% of fibers and compacting the CDG at the water content
close to the optimum Proctor value. In triaxial drained tests, the
addition of fibers seems to increase the shear strength of the
CDG and its stiffness. Dilation is also reduced.
In Singapore, laboratory tests have been performed by
Xiao
et al. (2013)
in order to determine the characteristics of the
Singapore upper marine clay when mixed with 20 to 50%
Portland cement (PC) and up to 0.32% fibers of different types.
As a result of their study, strength and ductility of cement-
treated clay were improved by fiber reinforcement. There is an
optimum fiber content with regard to performance and
workability of the material. Polyvinyl alcohol (PVA) fibers are
generally more efficient than polypropylene (PP) fibers except
for low cement and water contents. The length of the fibers has
a significant effect on the ductility of the cement-treated clay for
both fiber types. Concerning the strength, the influence of the
fiber length is more significant for PVA reinforcement than for
PP reinforcement.
Cuira et al. (2013)
present the results of numerical models
simulating an axial Static Load Test (SLT) on a soil-cement
column. Numerical and experimental results are compared with
the help of three Finite Element models and one simplified
