264
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
(Chow and Airey 2011) and Statnamic pile tests (Brown and
Hyde 2008) where the pile is displaced at rates of the order of 1
m/s. In these cases, rate effects cause the capacities derived
from the tests to exceed static values; an issue which is usually
dealt with by the inclusion of damping co-efficients in the
analyses. (Brown and Powell 2013) These damping co-
efficients are not always uniform throughout each test, but may
vary with strain level. This highlights the importance of
understanding the strain level dependence of rate effects in
order to improve the accuracy of dynamic testing in the field.
2 MATERIAL TESTING AND PROCEDURES
The tests were carried out on reconstituted speswhite kaolin, the
properties of which are shown in Table 1.
Table 1. Properties of the speswhite kaolin used
Property
Value
Plastic limit, w
P
(%)
32.5
Liquid limit, w
L
(%)
65.0
Plasticity index, PI (%)
32.5
Clay fraction (%)
80
Activity (%)
40.6
Specific surface area (m
2
/g)*
36.7
Permeability, k (mm/s)
#
1.17 x 10
-6
c
v
(m
2
/year)
+
23.52
MCSL
0.9
λ
0.101
N
2.678
* Determined from methylene blue spot testing
#
Determined at an effective stress of 300 kPa
+
Determined for a 100 kPa stress increment
The samples were first prepared as slurry with a moisture
content of 120 % using de-aired, de-ionised water before being
one dimensionally consolidated to an effective stress of 180 kPa
for three days. These were then trimmed to 200 mm length and
100 mm diameter to create triaxial samples. Once installed in
the triaxial apparatus, the sample was saturated to an effective
stress of 50 kPa at a back pressure of 300 kPa and then
reconsolidated to an effective stress of 300 kPa to restore
isotropic conditions. Sample drainage was facilitated by using
vertical filter paper drains on the surface of the sample,
connected to both the top and bottom drainage valves. These
were required as the use of lubricated end platens in the testing
meant that conventional drainage was not possible, and had the
additional benefit of significantly reducing consolidation times.
2.1
Testing apparatus
The tests were carried out in a GDS advanced
electromechanical dynamic triaxial rig specially modified to
carry out high speed monotonic tests. The rig is capable of axial
displacement rates of 100 mm/s, and during high speed testing it
is controlled by a GDS digital system capable of controlling the
axial displacement within a time interval of 0.1 milliseconds.
Both the back and cell pressures were provided by GDS
pressure controllers. Lubricated end platens of a similar design
to those proposed by Rowe and Barden (1964) were used in
order to minimise the inhomogeneity caused by end restraint
conditions. As these can introduce errors into the measurement
of small strains using external methods, these were measured
using Hall effect transducers, two axial and one radial, mounted
directly on the sample providing a resolution of 1×10
-6
% strain.
Pore pressures were monitored using a mid-height pore pressure
transducer mounted on the surface of the sample.
2.2
Testing programme
The testing programme consisted of triaxial tests at shear strain
rates from 0.333 to 60,000 %/hr in order to investigate strain
rate effects over as large a range as possible. These were carried
out at a comparatively low effective stress of 300 kPa as
previous studies have shown that greater rate effects are
observed at higher moisture contents. (Bea 1982, Brown and
Hyde 2008, Chow and Airey 2011) Throughout the testing
programme, the samples were allowed to drain through the filter
paper drains in order to allow rate effects due to consolidation
to be investigated.
3 RESULTS AND DISCUSSION
3.1
Observed rate effects and their modelling
The rate effects observed at the various strain rates are shown in
Figure 2, with a shear strain rate of 100 %/hr taken as the
reference rate. In order to allow comparison with other studies
using differing materials and sample sizes, the strain rates have
been converted into the normalised dimensionless velocities
used by Randolph and Hope (2004) as shown in Equation 1.
v
c
vd V
(1)
where
V
is the normalised dimensionless velocity,
v
is the
strain rate applied (in m/year),
d
is the sample diameter (in m)
and
c
v
is the co-efficient of consolidation in m
2
/year. As can be
seen, the curve follows the behaviour expected with time for
consolidation effects dominating up to
V
= 11, after which
undrained viscous effects are significant.
In order to quantitatively assess the data, the rate effects
model proposed by Randolph and Hope (2004) shown in
Equation 2 has been fitted to the data using least mean square
regression.
0
ref
1
0
1
d
ref
V
V
sinh
V
V sinh
)10(ln
λ
1
cV 1
b 1
q
q
(2)
where
b
,
c
, and
d
are curve fitting parameters used to model
the time for consolidation effects and
λ
is the rate effect per log
cycle increase in strain rate used to model viscous effects.
V
ref
is
the normalised velocity associated with chosen reference rate
and
V
0
is the point after which time for consolidation effects are
negligible. This process was also repeated for the measured rate
effects at selected strains to identify the variation in rate effects
at different strain levels as shown in Figure 3. The parameters
obtained from the curve fitting process are shown in Table 2.
The fitting parameters at peak strength (which occurs at
varying strain levels) show that the rate effect per log cycle was
found to be 22.5% which is higher than previous studies. This
highlights the need to develop a framework to predict rate
effects based on the current soil state and properties of the
material in question.
