44
Geotechnical News • June 2016
GROUNDWATER
Laboratory tests have shown that
the hydraulic conductivity,
K
, of
compacted clay is influenced by the
moulding water content,
w
c
, and
degree of saturation,
S
rc
, after compac-
tion. Specimens compacted wet of
optimum are usually more impervious
than those compacted dry of optimum
with the same energy (Lambe 1954;
Mitchell et al. 1965). Two types of
porosity coexist in compacted clay
(e.g., Li and Zhang 2009; Romero
2013; Della Vecchia et al. 2015).
The primary porosity corresponds to
the fine, micron-scale, structure of
solid particles within the clods. The
secondary porosity is due to poorly
interlocked clods and lifts after com-
paction. The
K
value of compacted
clay, once saturated, depends mostly
upon secondary porosity, as shown
by dye-stained seepage (Elsbury et al.
1990). Compaction wet of optimum
corresponds roughly to
S
rc
values close
to 90%, which means that 10% of the
void space is filled with trapped air
which is difficult to remove by com-
paction. The resulting low air perme-
ability was used as a field control of
compaction conditions (Langfelder et
al. 1968).
For clay compaction the dry density,
ρ
d
, must be equal to or higher than
a given percentage of the optimum
value,
ρ
opt
, of the standard or modi-
fied Proctor test. The two optima are
well correlated to each other (Chapuis
2002). In the 1980s, the minimum
moulding water content,
w
c,min
,
was the
optimum water content,
w
opt
,
whereas
after 1990 for Quebec it became the
plastic limit,
w
P
. Other specifications
(after 1990) may include a minimum
value of
S
rc
of 90% and also some
maximum,
w
c
,
value,
w
c,
max,
to permit
normal equipment traffic. Other
criteria are the liner total thickness,
the thickness of each lift (15–30 cm),
the length of the steel feet on a sheep
foot roller used to knead a clay lift,
and sometimes, a maximum clod size
(Benson and Daniel 1990).
In Quebec, most liners before 2000
have been built with local clays, which
are not fissured in natural deposits,
except in their upper crust. These
natural clays lose their sensitivity, if
any, after drying, which allows their
use in liners. All clays are mostly rock
flour (Foscal-Mella 1976; Locat et al.
1984). The
K
value of natural (in-situ)
non-fissured clays is usually in the
10
-10
–10
-9
m/s range (e.g., Tavenas
et al. 1983; Duhaime et al. 2013;
Duhaime and Chapuis 2014), thus
lower than the
K
of compacted clays
in the 10
-10
–10
-7
m/s range (Chapuis
1999; Chiasson 2005).
Sufficient compaction of each clay
layer is verified by compaction
control tests, using neutron probes or
other methods. In addition, to retain
its quality, the liner must not dry or
freeze before being used. This is a
major issue in Canada. Otherwise, the
K
value could increase by two orders
of magnitude, and other properties
may be affected (Chamberlain and
Gow 1979; Kim and Daniel 1992;
Benson et al. 1995; Eigenbrod 1996;
Chapuis 2013; Xue et al. 2014). As
soon as the clay liner is constructed,
in Quebec, it is covered with 20–30
cm of sand or sand-and-gravel that is
kept moist to avoid detrimental effects
of desiccation (Albrecht and Benson
2001; Yesiller et al. 2000). The use of
geotextile may also help to decrease
cracking (Safari et al. 2014).
Other field controls include permeabil-
ity tests in specially installed moni-
toring wells, or long-term infiltration
tests (Day and Daniel 1985; Chapuis
1990a, 1999; Guyonnet et al. 2003),
which are time-consuming and can be
carried out only at a few places. When
the liner construction season is short,
as in Canada, only a few tests can be
performed. In addition, experience in
Quebec indicates (unpublished results)
that the few
K
values, “measured” at a
few places, poorly predict the actually
measured and monitored total leakage
rate.
Many case histories, from several
countries, were published with little
information other than reporting a too
high leakage rate. However, technical
reports by engineers and technicians
fully document compaction conditions
and
K
values as determined by labora-
tory or field tests. The liner perfor-
mance may then be predicted from
statistics based on these tests, but alas
the total leakage is usually not mea-
sured. This seriously limits our ability
to confront prediction and theory. In a
few published cases, the total leakage
rate was simply said to be 10 to 1000
times higher than predicted (Auvi-
net and Hiriart 1980; Daniel 1984;
Picornell and Guerra 1992; Chapuis
2002). A few authors have tried to
explain this difference by large-scale
effects. Their opinion is that full-
scale tests are more likely to contain
preferential flow paths, and thus yield
large-scale
K
values higher than the
K
of smaller-scale tests (Shackelford and
Javed 1991; Cazaux and Didier 2002).
However, the statistics for large sets of
compaction control data reveal lognor-
mal distributions: these may then be
used to predict the full-scale
K
, for the
total leakage rate, which eradicates the
need to invoke scale effects (Chapuis
2013).
The project as built and
repaired in the 1980s
The two rectangular lagoons were
built, tested and successfully repaired
in the 1980s. Their bases had areas
of 130 m x 50 m and 130 m x 40 m
respectively. Their sides had a 1V/3H
slope. Each liner was 75 cm thick,
built in five 15-cm thick lifts. The
liners were constructed in July, during
a warm, dry, summer. The completed
clay liner was covered with 20 cm of
gravelly sand on the bottom and 30
cm of crushed stone (0- 20 mm) on
the slopes. The equipment for waste-
water treatment was installed. Four
pipes, which carried influents and
effluents, passed through each liner. A
photograph of a representative lagoon,
half-full of wastewater, is shown in
Figure 1.
The clay had a mean plastic limit
w
P
of 26%, a natural water content,
