1561
Technical Committee 203 /
Comité technique 203
dense specimens have curves rising sharply as the number of
cycles decreases; and (2) while the effect of relative density is
very pronounced for Toyoura sand, the effect of relative density
on pumice specimens appear to be not as remarkable.
4.2
Effect of confining pressure
Next, the influence of effective confining pressure on the
liquefaction resistance of reconstituted pumice sands was
investigated. For this purpose, dense pumice sand specimens
(initial void ratio,
e
i
=1.90-2.00) were subjected to three different
levels of effective confining pressure,
c
’=35, 100 and 500 kPa
under different levels of cyclic shear stress ratio, CSR
(=
d
/2
c
’). Figure 3 illustrates the confining pressure
dependency of liquefaction resistance for reconstituted pumice.
It can be seen that the curves are almost parallel to each other,
with the liquefaction resistance increasing as the confining
pressure decreases, consistent with the observations made on
natural sands (e.g., Rollins and Seed, 1988). The value of the
correction factor for overburden stress
K
(CSR causing
DA
=5% in 15 cycles under any confining pressure normalised
to the corresponding value of CSR at
c
’=100 kPa) is equal to
1.16 for
c
’=35 kPa and 0.88 for
c
’=500 kPa. These values
appear to coincide with those reported for reconstituted natural
sands (e.g., Boulanger and Idriss, 2004).
4.3
Development of particle crushing during cyclic loading
The level of particle crushing during undrained cyclic testing
has been reported by Orense et al. (2012). They noted that under
the confining pressures considered, pumice undergoes
remarkable particle crushing when subjected to cyclic shear. As
cyclic shearing and particle crushing occur, the soil structure is
gradually stabilized, resulting in higher cyclic shear resistance,
even exceeding that of Toyoura sand. The cyclic shearing and
the associated particle breakage resulted in stable soil structure
for both dense and loose cases, and therefore, the effect of
density was not as remarkable when compared to the cyclic
behaviour of Toyoura sand, a hard-grained sand.
To elucidate further the development of particle crushing
during a cyclic loading, a series of tests were performed such
that the tests were terminated after a specified number of cycles
afterwhich sieve analyses were performed. For these tests,
virgin samples were used at each test. A confining pressure of
c
’=100 kPa was considered, with the void ratio set at
e
i
=1.90-
2.00. For CSR=0.10, the sieve analyses were carried out: (1) on
the virgin samples; (2) after the end of consolidation stage; (3)
after
N
=10 cycles; (4) after
N
=100 cycles; and (5) after
N
=1000
cycles. On the other hand, for CSR=0.20, sieving was done (1)
after
N
=10 cycles; and (2) after
N
=83 cycles where initial
liquefaction (pore pressure ratio,
r
u
=100%) occurred.
The grain size distributions of the specimens after the tests
were determined. Particle crushing occurred, but with the level
of CSR and the number of cycles applied, it was difficult to use
the grading curves to make reasonable comparison. Instead, a
method of evaluating particle crushing originally proposed by
Miura and Yamanouchi (1971) was used which involves the
quantification of the surface area of the particles. The specific
surface of the particles was measured by first sieving the soil
using 2.5 mm, 2.0 mm, 1.18 mm, 0.5 mm, 0.212 mm, 0.15 mm
and 0.063 mm sieve sizes. For this range of particle sizes, the
specific surface area (in mm
2
/mm
3
) is calculated as:
d
ws
m
m
G d
d
F S
3
2
2/
3/4
2/
4
100
(1)
where
d
m
=(
d
1
.
d
2
)
0.5
,
d
1
and
d
2
are adjacent sieve sizes (e.g.,
0.50mm and 0.212 mm),
F
is the % by weight retained on the
sieve,
G
s
is the specific gravity of the particles,
w
is the unit
weight of water and
d
is the dry unit weight of the specimen.
Figure 4 shows the development of the surface area
S
for the
different tests described above. Firstly, it was observed that
consolidation at 100 kPa effective confining pressure did not
induce appreciable particle breakage to the pumice particles;
however, the cyclic shearing did. Secondly, the degree of
particle crushing increased with the amplitude of applied CSR.
For the test with CSR=0.20, the increase in surface area during
the initial stage of cyclic loading was small; however, as the
liquefaction stage was reached (
N
=83), the surface area
increased remarkably because large strains occurred with
associated translation and rotation of particles causing the
higher degree of crushing. For CSR=0.10, the state of
liquefaction did not occur even when
N
=1000 cycles. Particle
breakage was more or less gradual, with almost linear variation
with the logarithm of
N.
4.4
Comparison between laboratory and field data
Cone penetration tests (CPT) and seismic dilatometer tests
(sDMT) were performed at the Mikkelsen Rd site and Carrs Rd
site to supplement the undrained cyclic triaxial tests conducted
on the undisturbed samples taken from these sites. The field
tests were performed as near as possible to the sampling site.
Correlations between the cyclic resistance obtained from the
laboratory tests and the in-situ parameters were performed to
confirm which method was appropriate for pumice. Note that
undisturbed soil samples were obtained at three elevations at
Mikkelsen Rd site, while samples from Carrs Rd site were taken
only at a single depth; hence, emphasis is placed on the former.
In addition, the results presented herein may be appropriate only
for the two sites investigated and further tests are necessary to
confirm their applicability to other pumiceous sites.
0
0.1
0.2
0.3
0.4
0.5
0.6
1
10
100
1000
C
yclic
S
hearS
tress
R
atio,C
S
R
Number of cycles, N
35 kPa
100 kPa
500 kPa
Figure 3: Comparison of liquefaction resistance curves for reconstituted
dense pumice sands under different
c
’
.
In the comparison, the liquefaction resistance of the
undisturbed samples is specified in terms of the magnitude of
80
85
90
95
100
0
1
10
100
1000
S
urface
area,S
(m
m
2
/m
m
3
)
No. of cycles, N
�
d
/2
�
c
’=0.20
�
d
/2
�
c
’=0.10
Consolidation
Figure 4. Relationships between specific surface area and number of
cycles during cyclic undrained tests.
