424
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
the peak strength of the OC clay, but greater than the residual
strength.
It is generally accepted that the effective strength of
uncemented saturated clays is frictional and the strength
envelope is nonlinear. Hence the strength envelope will pass
through origin, and so the true cohesive intercept c'=0kPa
(Burland 1990). However, over the typical range of stress levels
met in practice (~50-400kPa) the effective peak strength for NC
and OC clays can be approximated by a linear relationship
between effective normal stress at failure
σ
'
f
and shear strength
τ
f
using the Mohr-Coulomb strength equation:
τ
f
=
σ
'
f
·
tan (
φ
') + c'
(1)
where
φ
' and c' respectively are the tangent drained angle of
shearing resistance and the apparent cohesive intercept, as
illustrated in Figure 1b. For OC clays c'
oc
>0kPa and for NC
clays c'
nc
≈
0kPa. While the angles of shearing resistance for OC
and NC clays are typically found not to differ much.
Generally the frictional resistance to shearing, as expressed
by
φ
', can be expected to decrease as the content of platy clay
minerals increase in the soil mass. With increasing content of
platy clay particles both the liquid limit w
L
and the plasticity
index I
P
will increase, and hence a correlation between
φ
' and
w
L
or I
P
can be expected.
1.2
Existing relationships between effective shear strength
and plasticity index
Several studies have been reported in the literature with regards
to the correlation between the effective angle of shearing
resistance
φ
'
peak
and plasticity index I
P
(Brooker and Ireland
1965, Ladd et al. 1977, Stark and Eid 1997, Terzaghi et al. 1996
among others). These studies are however mainly focused on
normally consolidated reconstituted or undisturbed natural
clays, while only little has been reported for overconsolidated
undisturbed clays.
Figure 2 shows collected data from the literature in a plot of
φ
'
nc
vs. I
P
(single log plot) for primarily NC clays (I
p
range 5-
240%).
φ
'
nc
generally represent a peak secant value with the
assumption that c'
nc
is zero. A very significant scattering of the
data points is seen, e.g. at I
P
=20% the value of
φ
'
nc
is found to
vary between approximately 25deg. and 35deg. However,
despite the significant scatter a trend of reducing
φ
'
nc
with
increasing I
P
is seen, and the data furthermore suggest the
existence of a lower bound value for
φ
'
nc
at given value of I
P.
Figure 2.
φ
'
nc
vs. I
P
for primarily normally consolidated reconstituted
and undisturbed clays after Ladd et al. 1977 (with data from Kenney
1959 and Bjerrum and Simons 1960), Terzaghi et al. 1996 and Brooker
and Ireland 1965.
The shaded area in Figure 2 represents the range of results
reported by Stark and Eid 1997 from a large series of ring shear
tests on 24 different reconstituted normally consolidated natural
soils (I
P
=8-244%, Clay-size fraction CF=10-88%, normal
effective stress
σ
'
n
=50-400). Based on the data, relationships
between
φ
'
nc
and I
P
were proposed which were dependent on
clay-size fraction and normal effective stress, as seen in Figure
3. By taking account of clay-size fraction and stress level Stark
and Eid showed a significantly reduced scatter around the mean
trend lines. A downward shift in the trend lines were observed
with increasing stress levels and increasing clay-size fraction.
The findings by Stark and Eid suggests that the observed scatter
in the reported data found in the literature, as shown in Figure 2,
to a large extent can be explained by variations in stress level
due to a non-linear strength envelope and additionally clay-size
fraction, as both soil mineralogy and clay-size fraction are not
accounted for solely by the variation in the index properties.
Figure 3.
φ
'
nc
vs. I
P
for reconstituted normally consolidated soils as a
function of clay-size fraction and normal effective stress (Stark and Eid
1997)
Based on the literature data a cautious lower bound (LB)
estimate of the relationship between
φ
'
nc
and I
P
for NC clays can
be derived together with a best estimate from the best-fit
regression line through the data, as indicated in Figure 2 and
given below.
Cautious LB estimate:
φ
'
nc
= 39-11
·
log I
P
(deg.) (2)
Best estimate:
φ
'
nc
= 43-10
·
log I
P
(deg.) (3)
The lower bound estimate, which correspond roughly to the 5%
fractile, also approximately match the lower bound of the range
of results reported by Stark and Eid for clay-size fractions above
50% and a stress level of 400kPa. Hence for clay-size fractions
below 50% and stress levels below 400kPa the effective angle
of shearing resistance
φ
'
nc
can be expected to be significantly
greater than estimated from eq. 2 (up to approximately 12deg.
for CF<20% and
σ
'
n
=50kPa, as seen from Figure 3).
2 SOIL DESCRIPTION AND TEST PROCEDURES
A number of triaxial compression tests have been performed by
GEO on various undisturbed overconsolidated clays over the
past decades. Test data have been collected from older tests (>
30 years) and more recent test series as listed in Table 1.
2.1
Soil description
The tested soils range from very low plasticity clay tills to
extremely high plasticity Eocene clays. The recent tests include
a test series in connection to the 1992 Great Belt bridge (GB)
ground investigation, which provides a significant amount of
test data for very low plasticity clay till. While the newly
completed 2011 Fehmarnbelt (Fixed Link) (FB) ground
investigation contribute significantly to the understanding of the
strength behavior of very high to extremely high plasticity
Eocene and Paleocene marine clays from the Røsnes, Ølst and
Holmehus clay formations. The majority of the investigated
clays from the Fehmarnbelt (Fixed Link) ground investigation
have been assessed to be situated within glacial folded strata. A
0
5
10
15
20
25
30
35
40
45
1
10
100
1000
φ
' (deg.)
I
p
(%)
Mean
φ
'
nc
=43-10·log I
P
n=233,R
2
=0.41, SE
y
=3.7
LB
φ
'
nc
=39-11·log I
P
