3002
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
3 RESULTS
Figure 1 shows the hydraulic conductivity trend over time of the
mixture permeated with the salt solutions of K
2
SO
4
at different
concentrations together with the k trend of the same mixture
permeated with water. With reference to the performance in
water, the mixture shows the decrease of k with time typical of
the well selected CB mixtures: k values of the order of 10
-8
cm/s
can be reached just after two months of curing. The reduction of
k is significant during the first year of curing.
As far as the hydraulic conductivity with the salt solutions is
concerned, each sample shows an initial decrease in the k value
(no effluent SO
4
2-
concentration were detected in this period),
followed by an increasing hydraulic conductivity. Finally an
almost constant trend of hydraulic conductivity with curing time
occurs with the effluent SO
4
2-
concentration that was measured
to be equal to the inlet one. From Figure 2 it is also evident that
the rate of reduction or increase of k with time depends on the
permeant concentration: the higher the concentration, the faster
the reduction and the successive increase in the k value with
time, as well as the shorter the curing time at which the
inversion of the k trend occurs. The reduction in the hydraulic
conductivity exibited by the mixture during the first month of
curing is equal or even higher than that of the mixture
permeated with water. Significant (increasing) concentrations of
SO
4
2-
were measured at the effluent only when the k values
increase with time; this transient phase lasted when the k values
starts to be constant with time (Fratalocchi et al. 2010;
Brianzoni, 2012).
In order to explain the results of Figure 1 it is necessary to
consider the reactions occurring between sulphates and
cementitious materials. Sulphate attack initially develops from
the reation between SO
4
2-
and both calcium hydroxite, Ca(OH)
2
,
and (partially) calcium hydro-silicate, C-S-H, with the
consequent precipitation of gypsum (CaSO
4
.
2H
2
O) and release
of OH
-
. The subsequent (damaging) reactions take place
between the sulphate and the hydrated calcium alluminate or
calcium alluminate monosulphate hydrate (Bensted 1995;
Gollop and Taylor, 1992, 1995) to produce ettringite with a
consequent expansion.
All the aforesaid interaction mechanisms are progressive
through the sample with curing time: the initial gypsum
precipitation into the pores (reaction confirmed by the high pH
values measued at the effluent, pH = 12-12.4) causes clogging
of pores and contributes to the reduction of the hydraulic
conductivity with time. This contribution tends to reduce and
become negligible as the ettringite starts to form along the
reaction front; such an expansive reaction is able to
progressively invert the hydraulic conductivity trend with time
up to a rapid increase. The expansive reaction of ettringite was
confirmed by the increase in volume on the samples equal to 7-
10% (Fratalocchi and Pasqualini 2007) and by a net of diffuse
fissures observed on the samples at the end of the tests. The
fissures give rise to preferential pathways that are responsible of
the overall constant value of k (2-3 x 10
-6
cm/s) at the end of the
interaction mechanisms.
Samples of the same mixtures were permeated also with three
aqueous solutions of sulphuric acid at different pH (Table 1).
Two samples of different thickness (sample A and B) were
permeated with the same H
2
SO
4
solution (pH = 2.0) in order to
evaluate the interaction effect taking into consideration different
curing time. Figure 2 shows the hydraulic conductivity trends
over time of all the samples. It is evident that the hydraulic
conductivity trends versus time are similar to those measured on
the samples permeated with the salt solutions. The main
chemical reactions are indeed the same, with the addition of the
dissolution of the cement hydration products (mainly calcium
hydroxide and C-S-H) by the acidic solutions. SO
4
2-
being
equal, the increase in hydraulic conductivity tends to be faster
when the mixture is permeated with the acidic solution. This is
evident if we compare the k trend of the samples permeated
with the salt solution at concentration of SO
4
2-
of 2756 mg/l and
the acidic solution of 1176 mg/l: both samples show a fast
decrease followed by an increase of hydraulic conductivity but
with the acidic solution the increase in k starts after about 250
days of curing whereas the increase of k occurs much more later
(about 520 days) when the CB mixture is permeated with the
saline solution, notwithstanding the lower concentration of
SO
4
2-
.
Considering the chemical reactions occurring between the
different solutions and the mixture, it is necessary to point out
that the overall hydraulic conductivity trend measured on the
samples in Figures 1 and 2 depend on the sample thickness and
on the flow rate through them. Therefore, in order to define how
long the mixture is able to keep a good performance, the curing
time cannot be considered as a reference parameter. To this
purpose, the pore volume of flow, PV, is an appropriate
parameter. Different criteria can be adopted to establish a
satisfactory hydraulic performance for the CB mixture: for
example, hydraulic conductivity lower than a maximum
allowable value, or k lower than the k value measured with
water, etc.; among them, the number of pore volumes of flow
until k is decreasing, PV*, can be appropriate for the following
reasons:
- at brief curing, both for K
2
SO
4
and H
2
SO
4
, whatever
concentration, the reduction of k with time is equal or lower
than that with water;
- in the long term, if there is no inversion of the k trend with
time, low k values can be reached;
- it is not necessary to establish a target k value (that would be
related to a particular curing time);
- an increasing k trend with time does not imply a bad
performance at least immediately; therefore, the criterion is on
the safe side.
Therefore, the number of pore volumes of flow at which the k
value stops decreasing (named “critical pore volumes”, PV*)
was assumed as the reference value for the mixture good
performance. This value (calculated assuming a porosity of the
mixture equal to 0.6) was related to the concentration of SO
4
2-
for the different salt and acidic solutions, as shown in Figure 3.
Referring to the samples permeated with the acidic solution at
pH = 2.0, only the thin one (sample B) showed a stop in
decreasing of k after about 49 PV (500 days of curing) whereas
the other one (sample A) still shows a decreasing k with time
after 850 days of curing (at this time only 4 PV permeated the
sample). Therefore, data of sample A could not be considered in
Figure 3.
Figure 3 shows that PV* decreases as the sulphate ion
concentration increases; the relation is well represented by a bi-
logarithmic correlation. In particular, for concentrations of the
order of g/l or more, that is, in aggressive conditions, the critical
number of pore volumes assumes the same trend for both the
solutions (saline and acidic). For concentrations of SO
4
2-
lower
than 1 mg/l, data are currently available only for the H
2
SO
4
solution; the critical PV* should be lower for the H
2
SO
4
solution than that of the salt solution considering its combined
deleterious effect due to ettringite formation and dissolution of
cement hydration products. Data are necessary to confirm this
hypothesis.
Even assuming the criterion of good performance as the PV
value until the k trend over time be not increasing (instead of
decreasing), the data in Figure 3 would be practically the same
except for the data with the lowest concentration of SO
4
2-
which
would have a PV* slightly higher (equal to about 57). It is
important to point out that a little increase in PV* means a
significant longer lasting for the mixture when the k value is
low.
The results in Figure 3 can be useful from the practical point of