Environmental Engineering Reference
In-Depth Information
It clearly appears that the collapse due to wetting corresponds to: i) a decrease in
large pores greater than 7.5 µm; and ii) a decrease in medium-sized pores between
0.7 and 7.5 µm, with no effect on the smallest pores, below 0.7 µm. Also, the
dominant large-pore population after collapse is well graded around an average
value of
r
= 1.8 µm. Referring to the SEM photos previously presented, the collapse
apparently affects the larger inter-grain pores and results in a more organized and
homogenous microstructure characterized by a well-graded pore population of
around 1.8 µm. The smaller porosity not affected by soaking corresponds to the
porosity inside the clay aggregations that were observed as being irregularly
scattered within the skeletal granular structure made up of silt grains.
SEM observations on collapsed samples, however, did not show any significant
microstructure changes, even in the sample at 2.2 m. However, some differences
were suspected in the appearance of the clay fraction, with a smoother appearance
of the hydrated clay platelets compared to the drier clay platelets observed before
soaking [DEL 96].
Since it was suspected that the natural water content in the loess profile changed
due to seasonal effects, the effect of initial water content on collapse was
investigated on the 2.2 m sample. The collapse susceptibility was investigated at six
different water contents: 0, 4, 10, 14, 18% (natural water content), and at 23%. The
starting water contents were obtained either by drying soil in the laboratory
atmosphere for a given time (
w
<18%) or by adding water using a wet filter paper
(
w
>18%).
Both the simple and double oedometer methods [JEN 57] were used. In the
double oedometer method, two oedometer tests were conducted in parallel: a
compression test under constant water content up to 800 kPa and a compression test
up to 800 kPa on a sample previously saturated under 3 kPa. Figure 6.12 shows the
results obtained with
w
i
= 4, 14 and 23%. In all tests, a satisfactory agreement is
observed between the two methods, confirming the validity of the double oedometer
method. No fitting between the different curves was necessary.
Clearly, collapse increases with decreased initial water content, as shown in
Figure 6.13, which also reports the data obtained at the three other water contents
(
w
= 0, 10 and 18%). The good agreement between the two methods is observed at
all water contents, and the decrease in collapse with increased water content appears
to be significant and roughly linear. Significant collapse susceptibilities are
observed in drier states, showing the need to properly monitor the changes in natural
water content profiles resulting from ground-atmosphere exchanges [BLI 97],
particularly during the drier and wetter periods.
