Version confidentielle non publiée

A laboratory set up to evaluate the hydraulic behavior of inclined

capillary barriers
B. Bussière
NSERC-Polytechnique-UQAT Industrial Chair on Environment and Mine Waste Management, UQAT,
Rouyn-Noranda, Québec, Canada
M. Aubertin
NSERC-Polytechnique-UQAT Industrial Chair on Environment and Mine Waste Management, École
Polytechnique, Montréal, Québec, Canada
R.P. Chapuis
Dept. CGM, École Polytechnique, Montréal, Québec, Canada

ABSTRACT: Capillary barriers have been proposed as an alternative to low saturated hydraulic conductivity
covers to encapsulate wastes. Capillary barriers can be used to reduce water infiltration and/or gas migration,
depending on the climatic conditions. Different factors can affect the efficiency of a capillary barrier such as
material properties and layer thickness. Geometry is another parameter, frequently neglected in the past that
could affect the performance of a capillary barrier. In this study, a physical model called the inclined box has
been designed and constructed to evaluate slope effects on the hydraulic behavior of a capillary barrier. The
test results show that the slope influences water distribution in the layers. These results have been validated
by numerical modeling. The study also shows the ability of the physical set up to measure the diversion ca-
pacity of a capillary barrier. Hence, the authors demonstrate that geometry is one of the key issues to be con-
sidered in order to properly analyse the behaviour of a capillary barrier used for controlling the influx of water
and/or gas migration.

1 INRODUCTION the cover is inclined (e.g. Ross, 1990; Steenhuis et

al., 1991; Stormont, 1996).
Covers with capillary barrier effects (or CCBE) In humid climates, capillary barriers are used
have been recently proposed as an alternative to mainly for reducing the flux of gas and limiting wa-
covers with a low hydraulic conductivity layer to ter percolation through the cover. The high water
limit the exchanges between wastes and the site sur- content (and degree of saturation) in the fine grained
roundings. This technology is based on the capillary soil layer results in a reduced gas diffusion coeffi-
barrier principle; a phenomenon that appears under cient that leads to a lowered gas flux (downward or
partly saturated conditions when a fine-grained ma- upward), since the flux is proportional to the effec-
terial overlies a coarser one (e.g. Rasmuson and tive diffusion coefficient De. The flux can be ex-
Erikson, 1986, Morel-Seytoux, 1992). It is related to pressed according to Fick’s first law (Bear, 1972;
the contrast in hydraulic properties between the two Aubertin et al., 1995):
materials, which restricts vertical water flow at the ∂C (t )
interface. The fine-over-coarse material system can F (t ) = − De (1)
then act as a barrier to the downward movement of ∂z
water due to the difference in the unsaturated hy- where F(t) is the diffusive flux of the element, C(t)
draulic conductivity of the soils at similar suctions. is the element concentration at time t and z is the
A CCBE can be used in both wet and dry cli- depth.
mates. The objective of a capillary barrier in a dry Different factors can influence the ability of a
climate is to reduce water infiltration through the capillary barrier to function properly including mate-
cover. The main advantages of such a cover lies in rial properties, climatic conditions and layer thick-
its design simplicity, its long-term stability and the ness. Many authors have investigated these factors
relatively low cost of construction (Morris and and their influence is well understood for unidimen-
Stormont, 1997). Recent studies have shown that, sional flow conditions (e.g. Rasmuson and Erikson,
with an adequate design and a reasonably dry cli- 1986; Akindunni et al., 1991; Morel-Seytoux, 1992,
mate, the performance of a capillary barrier could be Aubertin et al., 1996). However, capillary barriers
better than that of a traditional cover (Wing and Gee, built for real world conditions are usually inclined
1994; Morris and Stormont, 1997), especially when and can have slopes that vary from gentle to steep.

Version confidentielle non publiée

Version confidentielle non publiée

In most situations, the effects of both vertical and flux) was done prior the experiments.
horizontal unsaturated flow have to be considered. Figure 1: Description of the inclined box
The behavior of a capillary barrier in a real situation
can be considerably different from that of an ideal- The layers of soil placed in the box contained 15
ized 1D model (Aubertin and Bussière, 2001). TDR probes to measure volumetric water content.
To investigate the effect of slope on the perform- The TDR probes used were 20 cm «three wire buri-
ance of a capillary barrier used to limit water infil- able probes» from Soilmoisture. The Trase system,
tration and gas diffusion, a new physical apparatus also from Soilmoisture ("Soilmoisture Trase
has been designed and constructed: the inclined box. 6050X1"), was used to recover the data. The authors
The paper first presents a detailed description of the have used this system successfully during other
physical model. The main characteristics of the ma- laboratory experiments to evaluate the performance
terials used in the study are then presented. Exam- of multi-layered cover with capillary barrier effects
ples of test results obtained in the inclined box to (e.g. Aubertin et al., 1995, 1999). Because the
evaluate the impact of the geometry on the hydraulic probes are not completely buried in the box, calibra-
behavior of capillary barriers follow. A comparison tion of the probes for this particular case was done
between experimental and numerical results is also (Bussière, 1999).
presented and briefly discussed. Tensiometers (652x11-B1M3 from Soilmoisture
with an air entry value AEV of 1 bar) were used to
measure soil suction. The porous cups were placed
2 PHYSICAL MODEL DESCRIPTION beside the TDR probes. Each tensiometer was linked
to a pressure transducer (OMEGA PX243-15BG 5V
The apparatus constructed for this study is a rectan- with an integrated temperature adjustment), which
gular box of 2.5m x 1m with a thickness of 0.25 m, gives an output voltage between 1 and 6 volts when
with walls made of Plexiglasc (see Figure 1). Holes excited with a source of 8 volts. The output voltage
were drilled at the bottom every 20 cm to recover corresponds to a pressures ranging between –15 and
the water that percolates through the system and to +15 psi (or between -10.55 and +10.55 m of water).
evaluate the effective length of the capillary barrier The 15 tensiometers were connected to a data-logger
Lcb (or diversion capacity of the cover). These two system to register suction measurements 1 time per
parameters are used in the literature to quantify the minute. The location of all instruments is shown in
performance of capillary barriers to divert water in- Figure 1.
filtration (e.g. Ross, 1990; Steenhuis et al., 1991,
Stormont, 1995). A valve was installed at the lower
part of the slope to recover runoff. The slope angle α 3 MATERIALS AND TESTS PERFORMED
of the system can be adjusted to the desired inclina-
tion by the use of a jack (0°<α<30°). To observe the capillary barrier effects in a small-
A precipitation simulator is located on top. This scale laboratory apparatus such as this one, it was
simulator was made of sprinklers linked to a closed necessary to use materials sensitive enough to the
water reservoir. A constant air pressure can be ap- anticipated suction variations. A particular sand and
plied in the reservoir to regulate the flux (or the pre- gravel were selected, even if such combination
cipitation rate) at the top of the box. Calibration of would be unsuitable for use in the field in a CCBE.
sprinklers (relationship between air pressure and However, the trends observed in the layered system
hydraulic behavior should be similar to those of real
capillary barriers.
The grain size analysis of the sand shows that
about 75 % of particles are smaller than 1 mm and 5
% smaller than 0.1 mm. The gravel has a uniform
distribution with about 95 % of the particles between
1 and 10 mm. The saturated hydraulic conductivity
ksat of the two materials, evaluated in a rigid wall
permeameter, is 5.1x10-5 m/s and 4.7x10-3 m/s for
the sand and the gravel respectively. The materials
were placed in the box and densified to the desired
unit weight. The average porosity was 0.30 for the
sand and 0.42 for the gravel.
Simultaneous measurements of the volumetric
water content and suction in the box allowed us to
evaluate the water retention curves. Figure 3 shows
Version confidentielle non publiée
Version confidentielle non publiée

the laboratory measurements and the fitted curves system. The main hydraulic properties of the materi-
estimated with the well-known van Genuchten als are given in Table 1.
(1980) model using the RETC code (van Genuchten Figure 3. Permeability functions of the materials studied
et al., 1991):
mv
In this study, ten tests were performed with dif-
θ −θr  1  (2) ferent angles, precipitation rates and stratigraphy.
θe = = n  Due to space constraints, the results of only two of
θ s − θ r 1 + (α vψ )  v

them will be presented in the paper (see Table 2).

Here θe is the normalized volumetric water content, Details on the other tests can be found in Bussière
θr is the volumetric water content, θs is the saturated (1999).
volumetric water content and αv, nv and mv are fit-
ting parameters. As can be seen in Figure 2, the hy- Table 1. Main hydraulic properties of material tested
draulic contrast between the two materials is very Material ψa θr θs αv nv
(cm of water) (cm-1)
significant; the air entry value ψa of the sand is
Sand 35,0 0 0,30 0,03 3,72
about 35 cm of water while that of the gravel is less Gravel < 5,0 0 0,40 149,6 1,20
than 5 cm of water.
0.45
Gravel
Table 2. Description of the test performed
0.4
Sand Test Slope P Duration of Number of
van Genuchten (1980)
0.35 angle (cm/s) precipitation layers
Volumetric water content

0.3 S1 0° 5.9x10-3 1h50min 2

0.25 S2 18,4° 5.9x10-3 2h00min 2
0.2

0.15 For these tests, the precipitation rate P applied on

0.1
top of the box is approximately equal to ksat; this ap-
0.05
proach (P=ksat) was used by others in laboratory ex-
periments on inclined multi-layered systems (Miya-
0
0 20 40 60 80 100 120 zaki, 1988; Frind et al., 1976). The duration of the
Suction (cm of water) precipitation was set at 110 min for test S1 and 120
Figure 2. Water retention curves of the two materials. minutes for test S2. The duration was long enough to
reach a steady state regime in the box. Each pre-
The permeability functions k(ψ), which represent cipitation period was followed by a free drainage pe-
the evolution of the hydraulic conductivity k with riod.
suction, were calculated from the water retention The two tests were done on bi-layered capillary
curves from the closed-form analytical approach of barrier. The thickness of the bottom gravel layer was
the model proposed by Mualem (1976) with the van 0.3 m and the sand layer was 0.5 m thick. These
Genuchten (1980) equation: thicknesses were selected based on material proper-

[ ]
ties, to induce the type of response typical of a
krel = θ eλ 1 − 1 − θ e1 mv ( )
mv
(3) CCBE. Thus, the coarse grain layer has a thickness
greater than the ψa of the gravel as proposed by
where krel is the relative hydraulic conductivity Aachib (1997), and the sand layer has a thickness
(k(ψ)/ksat) and λ is a pore interaction index. The greater than his ψa to have a maximum saturated
permeability functions of the two materials are rep- thickness available to impede gas diffusion. The
resented in Figure 3. One can see that the hydraulic slope angle studied was between 0 and 18° (in this
conductivity of the gravel drops rapidly below that paper, only the extreme values of 0 and 18° are pre-
of the sand. This helps create the capillary barrier ef- sented; see Bussière, 1999 for details). 18° corre-
fects necessary for the appropriate response of the spond to the maximum value typically observed in
1.00E+00 the field for capillary barriers in a wet climate.
1.00E-01
Sand
1.00E-02
Gravel
Hydraulic conductivity (cm/s)

1.00E-03 4 RESULTS
1.00E-04

1.00E-05 The experiments in the inclined box had two main

1.00E-06 objectives: the first was to investigate the influence
1.00E-07 of the slope on the pressure and moisture distribu-
1.00E-08 tion in the system, and the second was to estimate
1.00E-09 the water diversion capacity of inclined capillary
1.00E-10
0 25 50 75 100 125 150 175 200
Version confidentielle
Suction (cm of water) non publiée
Version confidentielle non publiée

barriers. In both cases, a comparison between physi- the at the top and reached a maximum value of about
cal modeling and numerical modeling was done. 0.2 m of water after 5 hours of drainage. Again, the
The program used for the numerical calculations measured volumetric water content is consistent
is the finite element software SEEP/W developed by with the pressure profiles; θ is lower where the suc-
GEOSLOPE International (1996). This program is tion ψ is higher. It is also interesting to note in Fig-
able to simulate various situations in two dimensions ure 5 that the top part of the sand layer (elevation 0.7
including variably saturated flow, for both steady m) starts to drain but not the bottom (elevation 0.4
state and transient conditions. SEEP/W has been m). This is due to a local suction value that is higher
used successfully by the authors in previous studies (at the top of the layer) than the ψa of the sand (see
(e.g. Aubertin et al., 1995, 1996, 1999; Bussière, Figure 2). In the gravel layer, suction is less than
1999; Chapuis et al., 2001). The boundary condi- the static equilibrium line. This evolution of the
tions that were applied to the numerical model in- pressure profile observed during drainage is typical
cluded a unit flux on top during the saturation period of the one observed in 1D capillary barriers where
and Dirichlet conditions (pressure head equal to ele- the bottom coarse material layer allows a reduction
vation) at nodes located at the bottom valves. Be- of the pressure transmitted to the overlying fine
cause of the contrast between the material properties grained material when properly matched materials
and of the steep hydraulic functions, the constructed are used. The water distribution in the two layers is
model had many discrete elements consisting of also typical of the one observed in capillary barriers:
2600 elements and 8005 nodes. The hydraulic func- the coarse grained material tends to drain rapidly to
tions incorporated in the numerical model are the its residual water content (with a very low unsatu-
ones presented in Figures 2 and 3. The initial pres- rated hydraulic conductivity) while the fine grained
sure head at each node, required for transient analy- material may remain nearly saturated (e.g. Ak-
ses, was obtained from steady state analysis of the indunni et al., 1991; Aubertin et al., 1995,1996,
same model when an infiltration flux of 5.9x10-3 1999; Bussière, 1999). The predicted hydraulic be-
cm/s was applied to the top. This would represent havior of the capillary barrier from the computer
the equilibrium conditions reached in the physical modeling is fairly similar to that measured in the in-
model after saturation of the materials and before the clined box (see Figure 4 and 5). This confirms the
drainage period occurs. validity of the physical set up to represent unsatu-
0.9

4.1 Pressure and volumetric water content 0.8

0.7
The results are presented in term of pressure (suc- 0.6
Elevation (m)

tion) and volumetric water content profiles. Meas- 0.5

ured values with TDR probes and tensiometers are 0.4
Measured values

shown with predicted values obtained from numeri- 0.3

cal modeling. 0.2

The objectives of test S1 (see Table 2) were: i) to 0.1

0.0
validate the physical model with the well-known -0.8 -0.6 -0.4 -0.2 0.0
unidimensional test for a bi-layered system and ii) to Pressure (m of water)
t=1 hr t=2 hrs t=3 hrs t=4 hrs t=5 hrs t=24 hrs Static equlibrium
give base-line results for the comparison with the in-
clined case. The pressure and volumetric water con- rated water flow through horizontal capillary barri-
tent profiles are presented in Figures 4 and 5. ers.
Figure 4. Evolution of the suction profiles (linked symbols =
As it can be seen in Figure 4, after 1 hour of pre- predicted and unlinked symbols = measured) for different
cipitation (P = 5.9x10-5 m/s) the pore pressure in the drainage periods in the inclined box with layers placed hori-
two materials was near zero. This is consistent with zontally
the volumetric water content observed in the two
materials (see Figure 5); the sand is in a nearly satu- Figure 6 and 7 show respectively the suction and
rated state (θ = 0.3) while the volumetric water volumetric water content profiles at 45 cm of the top
content of the gravel layer is about 0.12, which cor- of the slope (at instrument locations; see Figure 1)
responds on the water retention curve to a low suc- when the capillary barrier is inclined at 18º. By
tion (see Figure 2). After shutting down the precipi- comparing Figures 4 and 6, one can see the impact
tation simulator suction in the top part of the sand of the slope angle on the pressure profile. The meas-
layer rapidly increased (see results after 2 hrs in ured values at the bottom of the sand layer are
Figure 4) to a value near 0.3 m of water. The pres- higher in the inclined system (elevation 1 m in Fig-
sure then stabilized at about 0.4 m of water after 5 ure 6) than in the horizontal case. For example, after
hours. As expected, due to a lower elevation, suction 5 hours (with about 3 hours of drainage), the suction
in the bottom part of the sand layer was lower than observed in the horizontal capillary barrier was

Version confidentielle non publiée

Version confidentielle non publiée

about 0.22 m of water compared to about 0.37 m of 4.2 Effective length of the capillary barrier
water for the inclined scenario. Such higher suctions As previously mentioned, capillary barriers can be
(or negative pressure) have a direct impact on the used in relatively dry climates to divert water and to
water distribution in the sand layer. The TDR probe reduce deep infiltration. The performance of this
placed at 10 cm of the interface near the highest part
of the slope gives values that are lower than the one 1.4

measured for the horizontal case (see Figures 5 and 1.2

7). The readings in the bottom gravel layer are also
1.0
different for the inclined case, but lower than the

Elevation (m)
ones for static equilibrium. Suction is higher for the 0.8
inclined system near the top of the slope during the 0.6
drainage period, which leads to lower volumetric
water content. The profiles (θ and ψ) at the bottom 0.4 Measured values
of the slope (not shown in the paper) were similar to 0.2
that of the horizontal case. This means that the water
0.0
content and suction in an inclined capillary barrier is 0.0 0.1 0.2 0.3 0.4 0.5
also a function of the location on the slope; the up- Volumetric water content
per part contains less water than the lower part due Initial conditions
t=2 hrs
t=1 hr
t=3 hrs
to a higher suction. Again, the results from numeri- t=22 hrs

cal modeling are pretty consistent with the ones type of capillary barrier is increased when the con-
measured in the box with TDR probes and tensi- tact plane between the two materials is tilted. The
ometers (see Figures 6 and 7). moisture that builds up above the contact flows
0.9 along the interface. At a certain point downdip of the
Measured values
0.8 top of the slope, the fine-grained soil can become
0.7
wet enough to allow significant infiltration into the
0.6
coarse layer underneath; this point called the Down
Elevation (m)

0.5
0.4
Dip Limit or DDL point (e.g. Ross, 1990). The hori-
0.3 zontal projection of the distance between the top of
0.2 the slope and the DDL point is an important charac-
0.1 teristic of a capillary barrier and is referred in this
0.0 paper as the effective length of the capillary barrier
0.0 0.1 0.2 0.3 0.4 0.5
Volumetric water content (Lcb). The new laboratory set up permits physical
t=1 hr t=2 hrs t=3 hrs t=4 hrs t=5 hrs t=24 hrs measurements of this length.
Figure 5. Evolution of the volumetric water content profiles
(linked symbols = predicted and unliked symbols = measured) Figure 7. Evolution of the volumetric water content profiles
for different drainage periods in the inclined box with layers (predicted and measured) near the top of the slope for different
placed horizontally drainage periods in the inclined box with layers inclined at
1.4 3H:1V
During the precipitation (P = 5.9x10-5 m/s for the
1.2
test S2), water was recovered through the holes at
1.0 the bottom of the box. The results presented in Fig-
Elevation (m)

0.8 ure 8 show the cumulative percentage water recov-

ery for each valve. From these, one can estimate that
0.6
the breakthrough was between valve #5 and valve #7
0.4 Measured values which corresponds to an effective length of the cap-
0.2 illary barrier between 0.75 and 0.95 m (Bussière,
1999). It is also possible to evaluate Lcb by using
0.0
-0.8 -0.6 -0.4 -0.2 0.0
numerical modeling. The technique consists of
Pressure (m of water) evaluating the pressure at the interface and identi-
Initial conditions
t=2 hrs
t=1 hr
t=3 hrs
fying the location where the pressure becomes larger
t=22 hrs Hydrostatic equilibrium than the water entry pressure ψ0* as defined by
Figure 6. Evolution of the suction profiles (predicted and Steenhuis et al. (1991). At this point, the water
measured) near the top of the slope for different drainage peri- should start to infiltrate the coarse material. ψ0* is
ods in the inclined box with layers inclined at 3H:1V approximately equal to pressure ψr corresponding to
the residual water content θr (hysteresis effects be-
tween drainage and wetting functions are neglected
here). With this technique, Lcb was estimated to ap-

Version confidentielle non publiée

Version confidentielle non publiée 100%

% Cumulative water recovery

90%
80%
70%
60%
50% Slope 3:1
40%
30%
20%
10%
0%

Runoff
1

11

13
# valve

proximately 0.8 m, which is close to the value Akindunni, F.F., Gillham, R.W. & Nicholson, R.V. 1991.
measured in the inclined box. Numerical simulations to investigate moisture-retention
characteristics in the design of oxygen-limiting covers for
reactive mine tailings. Canadian Geotechnical Journal, 28
:446-451.
5 CONCLUSION Aubertin, M., Chapuis, R.P., Aachib, M., Bussière, B., Ricard,
J.-F. & Tremblay, L. 1995. Évaluation en laboratoire de
Laboratory testing with the inclined box confirmed barrières sèches construites à partir de résidus miniers,
that the slope has a definite influence on the hydrau- MEND Report 2.22.2a.
Aubertin, M., Bussière, B., Aachib, M. & Chapuis, R.P. 1996.
lic behavior of a capillary barrier. For a sloping cap- Une modélisation numérique des écoulements non saturés
illary barrier, moisture content will vary along the dans des couvertures multicouches en sols. Hydrogéologie,
slope with suction (or the negative pressure). Results 1 : 3-13
of this test show higher suctions and lower water Aubertin, M., Bussière, B., Joanes, A.-M., Monzon, M. Ga-
content in the upper part of the slope than in the gnon, D., Barbera, J.-M., Bédard, C., Chapuis, R.P. & Ber-
nier, L. 1999. Projet sur les barrières sèches construites à
lower part or horizontal CCBE. This hydraulic be- partir de résidus miniers, Phase II: essais en place, MEND
havior can influence the performance of the CCBE. Report 2.22.2c.
In fact, if the slope effect is not taken into account Aubertin, M. & Bussière, B. 2001. Water flow through cover
when designing a CCBE used to impede gas migra- soils using modeling and experimental methods: Discus-
tion, suctions in the moisture retaining layer at the sion. ASCE J. Geotechnical and Geoenvironmental Engi-
top of the slope could become greater than the mate- neering (to be published).
Bear, J. 1972. Dynamics of Fluids in Porous Media. Dover
rial ψa causing it to desaturate. This would affect the Publications Inc., New York.
CCBE’s ability to perform as gas barrier. Bussière, B. 1999. Étude du comportement hydrique de cou-
vertures avec effets de barrière capillaire inclinées à l'aide
Figure 8. Cumulative water recovery through the different de modélisations physiques et numériques. Ph.D. Thesis,
holes at the bottom of the inclined box for test S2 Mineral Engineering Department, École Polytechnique de
The new physical model has also helped us better Montréal.
Chapuis, R.P., Chenaf, D., Bussière, B., Aubertin, M. & Crespo,
understand the diversion capacity of inclined capil- R. 2001. A user's assessment of numerical codes for saturated
lary barriers. The effect of the slope angle on the and unsaturated seepage conditions. Canadian Geotechnical
hydraulic behaviour of a capillary barrier is one of Journal (to be published).
the key issues that control the behaviour of covers Frind, E.O., Gillham, R.W. & Pickens, J.F. 1976. Application
used for controlling the influx of water. The results of unsaturated flow properties in the design of geologic en-
vironments for radioactive waste storage facilities. Pro-
show the capability of the physical set up to estimate ceedings of the 1st International Conference on Finite Ele-
the effective length of an inclined capillary barrier ments in Water Ressources, Princeton, N.J. Gray, W.G.,
submitted to a high precipitation rate. Results from Pinder, G.F. & Brebbia, C.A. (eds), 3.133- 3.163.
the physical modeling, confirmed by numerical GEOSLOPE International 1996. SEEP/W User's Guide.
modeling, show the importance of the CCBE’s ge- GOLDER & ASSOCIATES, 1999. Rapport final de suivi du
ometry on its hydraulic behavior and its ability to ef- comportement du recouvrement multicouche Les Terrains
Aurifères. Report MEND 2.22.4b.
fectively perform as a gas or water infiltration bar- Miyazaki, T. 1988. Water flow in unsaturated soil in layered
rier. slopes. Journal of Hydrology, 102 :201-214.
Morel-Seytoux, H.J. 1992. The capillary barrier effect at the
interface of two soil layers with some contrast in properties.
6 REFERENCES HYDROWAR Report 92.4, Hydrology Days Publications,
57 Shelby Lane, Atherton, CA 94027-3926.
Morris, C.E. & Stormont, J.C. 1997. Capillary barriers and
Aachib, M. 1997. Étude en laboratoire de la performance des Subtitle D covers: estimating equivalency. Journal of Envi-
barrières de recouvrement constituées de rejets miniers ronmental Engineering, 123: 3-10.
pour limiter le DMA. Ph.D. Thesis, Mineral Engineering
Department, École Polytechnique de Montréal.
Version confidentielle non publiée
Version confidentielle non publiée

Mualem, Y. 1976. A new model for predicting the hydraulic

conductivity of unsaturated porous media. Water Resources
Research, 12 : 513-522.
Rasmusson, A. & Erikson, J.C. 1986. Capillary barriers in cov-
ers for mine tailings dumps. Report 3307, The National
Swedish Environmental Protection Board.
Ross, B. 1990. The diversion capacity of capillary barriers.
Water Resources Research, 26 : 2625-2629.
Steenhuis, T.S., Parlange, J-Y. & Kung, K-J.S. 1991. Comment
on “The diversion capacity of capillary barriers” by Benja-
min Ross. Water Ressources Research, 27 : 2155-2156.
Stormont, J.C. 1995. The effect of constant anisotropy on cap-
illary barrier performance. Water Resources Research, 31:
783-785.
Stormont, J.C. 1996. The effectiveness of two capillary barriers
on a 10% slope. Geotechnical and Geological Engineering,
14 : 243-267.
van Genuchten, M.Th. 1980. A closed-form equation for pre-
dicting the hydraulic conductivity of unsaturated soils. Soil
Science Society of America Journal, 44: 892-898.
van Genuchten, M.Th., Leij, F.J. & Yates, S.R. 1991. The
RETC code for quantifying the hydraulic functions of un-
saturated soils. Environmental Protection Agency,
EPA/600/2-91/065.
Wing, R. & Gee, G. 1994. Quest for the perfect cap. Civil En-
gineering, October, 38-41.

Version confidentielle non publiée