Professional Documents
Culture Documents
Key words: expansive clays, thermo-hydraulic cell tests, numerical modelling, non-
standard ow models
1 Introduction
The long-term behaviour of low permeability (expansive) clays under low hy-
draulic gradient is not well-know, especially when the hydration takes place
under non-isothermal and conned conditions, which are the expected con-
ditions in high-level radioactive waste repositories (HLW). A trend to a very
low rate of hydration, compared to the expected or predicted one, has been
The clay behaviour during the inltration tests is very complex due to the
THM processes and couplings that take place during the simultaneous heat-
ing and hydration. A full description of the THM formulation adopted in
the analysis is presented in Olivella et al. (1994). The formulation incorpo-
rates the main thermal, hydraulic and mechanical phenomena. The problem
is approached using a multi-phase, multi-species formulation that expresses
mathematically the main THM phenomena in terms of: i) balance equations,
ii) constitutive equations and iii) equilibrium restrictions.
q = K (P g) (1)
where Pl and Pg are liquid (l) and gas (g) phase pressures, respectively,
is the phase density ( = l, g) and g is the gravity vector. The permeability
tensor, K depends on uid viscosity, degree of saturation (relative perme-
ability law) and pore structure (intrinsic permeability). Figure 1a presents the
variation of intrinsic permeability with porosity adopted in this work.
The retention curve establishes the link between the degree of saturation
and suction. A modied van Genuchten law has been adopted (Fig. 1b):
1
1 o d
s o s
Sl= 1+ fd : fd = 1 (2)
Po Pd
where s is the suction, Po is the air entry value and o is a model parameter.
The function fd is included to obtain more suitable values at high s.
It has been assumed that the diusion ows are ruled by Ficks law (i.e.
water vapour diusion in gas phase and air diusion in water) and that the
heat conduction is governed by Fouriers law. To describe the mechanical
behaviour the Barcelona Basic Model (BBM) has been adopted (Alonso et
al. 1990). All the laws and parameters used in the analysis are presented in
detail in Snchez and Gens (2005).
The system of equations is solved in a coupled way. One unknown (or
state variable) is associated to each balance equation. Then, from the state
variables, the dependent variables are calculated using the constitutive equa-
tions or the equilibrium restrictions. In this way the nite element program
CODE_BRIGHT Olivella et al. (1996) has been developed. This code has been
used to perform the numerical analysis presented herein.
3 Inltration Tests
The inltration tests are being performed by CIEMAT in cylindrical cells,
whose internal diameter is 7 cm and inner length 40 cm (Fig. 2). They are
made of Teon to prevent as much as possible lateral heat conduction, and
externally covered with steel semi-cylindrical pieces to avoid the deformation
of the cell by bentonite swelling. The material tested has been unsaturated
FEBEX bentonite, which is the clay used in the FEBEX Project to manufac-
ture the blocks of the FEBEX large scale tests.
The FEBEX clay was compacted with its hygroscopic water content
(around 14%) at an initial nominal dry density of 1.65 g/cm3 . Granitic water
is injected through the upper lid of the cell at a pressure of 1.2 MPa. In one
of the tests (GT40) the clay is being heated through the bottom surface at a
temperature of 100 C. The other test (I40) is being carried out at isothermal
conditions. The cells are instrumented with sensors of relative humidity and
temperature placed inside the clay at three dierent levels separated 10 cm.
The water intake and the relative humidity and temperature evolution at
dierent levels inside the clay are being measured online.
4 Numerical Modelling
The modelling has been focused on the analysis of the temperature and rela-
tive humidity. In relation to the water intake, only one analysis is presented,
due to problems occurred in its measurement. The study of these tests has
the additional interest to check whether the, already explained, problems in
the hydration rate observed in the FEBEX mock-up and ITT tests are also
present in these cells. The initial and boundary conditions of the model have
been imposed in order to be as close as possible to the experiments. An initial
suction of 120 MPa has been adopted, which corresponds approximately to
the relative humidity of 42% registered by the sensors inside the bentonite
just before hydration. An initially uniform temperature of 23 C has been as-
sumed. Initial hydrostatic stresses of 0.15 MPa have been adopted. Regarding
the boundary conditions a temperature of 100 C is imposed in the GT40 cell
at the contact between heater and bentonite (the bottom of the cell), while
a constant water pressure of 1.2 MPa is imposed at the other end of the cell
(upper part). The thermal boundary condition along the sample has been
adopted in order to adjust the temperature eld, in that sense an external
temperature of 23 C has been xed with a radiation coecient of 1 (one).
Three main cases have been considered. Firstly, a base case has been
adopted using the basic formulation presented in Sect. 2. The constitutive laws
adopted in the base case are the same one used in the analysis of the mock-up
test (Snchez and Gens 2005). The second and third cases studied correspond
to the application of non-standard ow models developed to explain the slow
hydration observed in test GT40.
The results of the isothermal cell (I40) measurements and modelling are shown
in Fig. 3. The evolution of relative humidity in dierent positions and also the
water intake is well captured by the model. Figure 4 presents the results of
the thermo-hydraulic cell (GT40). The thermal eld is quite well reproduced
considering that the variations of temperature due to the changes in the lab-
oratory temperature are not taken into account in the simulations. In terms
of relative humidity, the model captures qualitatively the main trend of be-
haviour, that is, an increasing saturation in the zones close to the hydration
boundary and, in zones close to the heater, an initial wetting, due to the con-
densation of the water vapour coming form the bottom, and then a drying.
However, the model predicts a quick hydration after the drying, while the
experiment shows practically constant values of relative humidity in the three
measured positions. This is a trend similar to the one observed in the mock-
up test, with a kind of locking of the hydration of the clay. To explain this
unexpected behaviour, two non-standard ow models and the corresponding
numerical analyses are introduced in the following sections.
Fig. 3. Isothermal cell (I40): a) evolution of the relative humidity, b) water intake
(d is the distance from the heater)
Experimental evidences show that under low hydraulic gradients (J), Darcys
simple relationship does not rule the liquid ow in some soils.
The strong clay-water interactions in clayed soils are suggested to explain
this non-Darcian ow behaviour. In this work the law adopted considers two
characteristic gradients (Fig. 5a): the threshold hydraulic gradient (Jo ) and
the critical hydraulic gradient (Jc ). Jo is the hydraulic gradient below which
no ow occurs. Jc is the hydraulic gradient below which ow occurs but it is
not Darcian. If the hydraulic gradient is higher than Jc , Darcys law applies
(Dixon et al. 1992). The average hydraulic gradient applied to obtain the
permeability of FEBEX bentonite shown in Fig. 1a was 15200. To clarify
the eect of hydraulic gradient on the value of hydraulic conductivity, the
hydraulic conductivity of FEBEX clay samples has been measured under low
hydraulic gradients and injection pressures
The hydraulic gradients applied range from 200 to 7200. No measurable
outows have been obtained for certain hydraulic gradients depending on the
dry density of the bentonite. These values are shown in Fig. 5b and could
be regarded as threshold hydraulic gradients, since no ow has been obtained
below these gradients. The dispersion found when hydraulic gradients lower
than 1000 are applied could indicate that the critical gradient for this ben-
tonite would be around this value (Villar and Gmez-Espina 2006).
Figure 6a shows the results obtained using a threshold gradient model. The
base case results are also presented in dot lines. The following considerations
have been taken into account: i) a threshold gradient equal to 50 (this is a
common value found in the literature), ii) a critical gradient close to 1500;
and iii) a power law for the range of hydraulic gradient with non-Darcians
ow. This model can reproduce very well the trend to a very low hydration
observed in test GT40. On the other hand, the model underestimates the
hydration in the isothermal cell.
The hydraulic gradient is the main force inuencing the movement of water in
soils. It is, however, not the only one. Figure 7 presents all the dierent kinds
of ow (except the electrical component) that can occur in a porous media
and the corresponding gradient responsible for the movements. Generally, the
non-diagonal coecients are relatively small and negligible compared to the
diagonal terms and the coupled processes can be ignored. However there are
certain problems in which, due to their particular conditions, the coupled
processes may play a more inuential role. The thermal conditions imposed
on the thermo-hydraulic cell correspond to a practically constant thermal
gradient during the test. In contrast, the hydraulic gradient is very high at
the beginning of the test, but diminishes with the hydration of the barrier.
The liquid ows associated with these two gradients have opposite directions.
5 Conclusions
The analyses of two inltration tests have been presented in this work. First
the results of a base case using a standard THM model have been pre-
sented. The results obtained with this model are not satisfactory, especially
when the comparisons with the thermo-hydraulic cell are analysed. Thresh-
old hydraulic gradient and thermo-osmotic eects have been included in the
formulation, these being two plausible phenomena that could explain the un-
expected trends observed in this kind of experiments. The results obtained are
very interesting since a physical interpretation for the unexpected behaviour
has been done. Each of these phenomena does not exclude the others and it
is possible that an explanation for the whole behaviour of the system would
require the combinations of some of them. These analyses show the need of
improving the classical THM formulation when complex processes, as those
developed in the thermo-hydraulic cells, are studied.
References
Alonso E, Gens A, Josa A (1990) A constitutive model for partially saturated soils,
Gotechnique 40:405430
FEBEX Project (2000) Full-scale engineered barriers experiment for a deep geolog-
ical repository for high level radioactive waste in crystalline host rock (Final
project report EUR 19612 EN EC Brussels)
Dixon D, Gray M, Hnatiw (1992) Critical gradients and pressures in dense swelling
clays, Can Geotech J 29:11131119
Olivella S, Carrera J, Gens A, Alonso EE (1994) Non-isothermal multiphase ow of
brine and gas through saline media, Transp in porous media 15:271293