Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273312074

Gas permeability of a compacted bentonite-sand mixture: Coupled effects of


water content, dry density, and confining pressure

Article  in  Canadian Geotechnical Journal · January 2015


DOI: 10.1139/cgj-2014-0371

CITATIONS READS

28 233

3 authors, including:

JiangFeng Liu Jean Talandier


École Centrale de Lille Andra
39 PUBLICATIONS   1,694 CITATIONS    94 PUBLICATIONS   832 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

quicksand View project

Hydromechanical behaviour of bentonite pellet-powder mixtures View project

All content following this page was uploaded by JiangFeng Liu on 03 July 2018.

The user has requested enhancement of the downloaded file.


1159

ARTICLE
Gas permeability of a compacted bentonite–sand mixture:
coupled effects of water content, dry density, and confining
pressure
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

Jiang-Feng Liu, Frédéric Skoczylas, and Jean Talandier

Abstract: The gas-tightness of compacted bentonite–sand mixtures is important to the total sealing efficiency of geological
repositories. The initial aim of this work was to determine whether the combination of a high confining pressure (Pc) and
incomplete saturation could cause a bentonite–sand mixture to become gas-tight. The results show that the physical character-
istics of the materials (degree of saturation, Sr; porosity, ␾; and dry density, ␳d) are very sensitive to changes in the applied
confining pressures and their own swelling deformation (or shrinkage). The combination of these changes affects the sample’s
effective gas permeability (Keff). For materials prepared at a relative humidity (RH) of 98%, Keff decreased from 10−16 to 10−20 m2
when Pc increased from 1 to 7 MPa. This means that gas-tightness can be obtained for a compacted bentonite–sand mixture when
the materials experience a series of changes (e.g., w, Sr, ␾, and ␳d). In addition, larger irreversible deformation (or hysteresis) was
observed during the loading–unloading cycle for the sample with higher water content. This phenomenon may be attributed to
larger interactions between the macrostructural and microstructural deformations and the decrease of preconsolidation pres-
sure during hydration.

Key words: gas-tightness, bentonite–sand mixture, effective gas permeability, confining stress, irreversible deformation.
For personal use only.

Résumé : L’imperméabilité aux gaz des mélanges compactés de bentonite et de sable est essentielle pour garantir la bonne
étanchéité globale des aires de stockage géologique. Le but initial de la présente étude a été de déterminer si la combinaison
d’une pression de confinement élevée (Pc) et d’une saturation incomplète pouvait rendre le mélange bentonite–sable impermé-
able aux gaz. Les résultats obtenus ont montré que les caractéristiques physiques des matériaux (le degré de saturation, Sr, la
porosité, ␾ et la densité sèche, ␳d) étaient très sensibles aux variations de la pression de confinement appliquée et de la
déformation (dilatation ou contraction) de ces matériaux. La combinaison de ces variations influe sur la perméabilité effective
aux gaz de l’échantillon étudié (Keff). Dans le cas de matériaux fabriqués à une humidité relative (HR) de 98 %, Keff diminue,
passant de 10–16 à 10–20 m2, lorsque Pc augmente et passe de 1 à 7 MPa. Cela signifie que l’étanchéité aux gaz d’un mélange
bentonite–sable compacté peut être obtenue lorsque les matériaux subissent une série de variations (de w, Sr, ␾ et ␳d, p. ex.). En
outre, une plus grande déformation irréversible (ou hystérésis) est observée durant le cycle chargement–déchargement dans le
cas d’un échantillon à la teneur en eau plus élevée. Ce phénomène est attribuable aux plus grandes interactions qui existent
entre les déformations macrostructurales et microstructurales et à la diminution de la pression de préconsolidation durant
l’hydratation. [Traduit par la Rédaction]

Mots-clés : étanchéité aux gaz, mélange bentonite–sable, perméabilité effective aux gaz, contrainte de confinement, déformation
irreversible.

Introduction site in the Eastern Paris Basin (Delay et al. 2008). This site was
In the design of high-level radioactive waste repositories at great constructed in a geological layer of COx argillite at a depth of
depths, bentonite or bentonite-based materials (a bentonite–sand 500 m. This allows in situ tests to be carried out, which have
mixture was used in this study) are generally chosen as buffer shown that some bentonite–sand blocks can be made gas-tight
materials to separate the radioactive waste canisters from the even though they are not completely water-saturated (De La
host rock or as backfill material to plug the access tunnels of Vaissière 2013). The material comprising these blocks is a mixture
disposal pits (Cui et al. 2008; Dueck 2008; Komine 2010; Ye et al. of 70% MX80 bentonite and 30% siliceous sand. A plausible expla-
2012). Saturated bentonite and bentonite-based materials have nation for the gas-tightness may be a coupling effect between the
low permeabilities, good swelling capabilities, and high radio- “mechanical stress and saturation” (see Fig. 1).
nuclide retardation capacities, which are useful for hindering Once these blocks are placed in the galleries, they are progres-
the transport of nuclides into the host rock (Yong et al. 1986; sively hydrated by pore water infiltration from the host rock for-
Jedináková-Křižová 1998; Cho et al. 1999). In France, the French mation (King et al. 2001; Wang et al. 2013). These blocks begin to
National Radioactive Waste Management Agency (ANDRA) has swell as soon as they come into contact with the water (Davy et al.
developed an underground research laboratory (URL) at the Bure 2009; Liu et al. 2014). The swollen bentonite or bentonite-based

Received 29 August 2014. Accepted 8 January 2015.


J.-F. Liu. Laboratoire de Méchanique de Lille (LML) and École Centrale de Lille, BP 48, F-59651 Villeneuve d’Ascq Cedex, France; State Key Laboratory for
GeoMechanics and Deep Underground Engineering, China University of Mining & Technology, Xuzhou 221116, China.
F. Skoczylas. Laboratoire de Méchanique de Lille (LML) and École Centrale de Lille, BP 48, F-59651 Villeneuve d’Ascq Cedex, France.
J. Talandier. ANDRA, 1-7 rue Jean Monnet, 92298 Châtenay-Malabry Cedex, France.
Corresponding author: Jiang-Feng Liu (e-mail: jeafliu@hotmail.com).

Can. Geotech. J. 52: 1159–1167 (2015) dx.doi.org/10.1139/cgj-2014-0371 Published at www.nrcresearchpress.com/cgj on 14 January 2015.
1160 Can. Geotech. J. Vol. 52, 2015

Fig. 1. Schematic diagram of the in situ coupling effects between water saturation and radial stress (modified after ANDRA 2005).
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

materials fill up all potential voids (the ones between the benton- with a solid density of 2.74 g/cm3. The siliceous sand was made of
ite blocks or between the blocks and host rock) that appear during 99.0% SiO2 (quartz), with a unit density of 2.65 g/cm3. The particle-
the construction process (Komine and Ogata 1996; Komine 2004; size distributions of the bentonite powder and sand used in the
Villar and Lloret 2008; Wang et al. 2012). This function, known as present tests are presented in Table 1 (Gatabin 2005). The powder
“self-sealing”, is important to the overall safety of the disposal pit was pre-conditioned in an environmental test chamber at 85%
(Komine 2004, 2009). With respect to the confinement conditions, relative humidity (RH) and 25 °C until a constant mass (w = 15.2%)
the bentonite or bentonite-based materials first experience free-
For personal use only.

was reached. More details on the mixture characteristics are given


swelling (prior to filling the voids) before being confined by the
by Gatabin (2005).
host rock. As contact is made with the host rock, a swelling pres-
The samples were obtained by compaction under a 12 MPa axial
sure is developed. In the massive plug, a significant water satura-
tion gradient is present between the core and external surface, pressure in a steel cylindrical mould. The resulting diameter was
itself in contact with the host rock and underground water. As a 37 ± 0.5 mm and the height was 12.5 ± 1 mm. The desired proper-
result, owing to a high water content, the external part of the ties were a water content equal to 15.2% and a dry density equal to
massive barrier swells in contact with a stiff host rock (e.g., COx 1.77 g/cm3. These properties are important because the compacted
argillite), so that it applies pressure to the partially water-saturated mix is intended to target a given swelling pressure of 7 MPa after
core (see Fig. 1). In addition, the convergence of the gallery wall full water saturation (Gatabin 2005). As shown by the results pre-
due to creeping of of host rock (COx argillite) can also compress sented hereafter, these properties were generally achieved to within
the barriers and lead to the decrease of their porosities, further a small variation.
affecting their effective gas permeabilities. The total in situ stress
is 12 MPa and the pore water pressure is 5 MPa, which correspond Experimental method
to an in situ depth of 500 m. As designed by ANDRA, the target This study was initially designed to simply evaluate the sealing
swelling pressure of the clay is 7 MPa, which approaches an equi- properties of an unsaturated but highly compacted bentonite–
librium state after water saturation to maintain a long-term bal- sand mixture. After compaction, an initial gas permeability test
ance between the host rock (COx argillite) and swelling clay. campaign was performed up to 5 MPa confining pressure (with
It has been generally established that anaerobic conditions are an exception of sample S2-2, which went up to 12 MPa), which is
reached in the repository after sealing. Under these conditions, considered a moderate pressure. The samples were then placed at
gas (mainly hydrogen) is produced within the repository tunnels a controlled level of relative humidity (RH: 75%, 85%, 92%, and 98%)
due to humid corrosion, organic matter degradation, and water
until their mass became stable. These samples were subsequently
radiolysis (Ortiz et al. 2002; Birgersson et al. 2008). Over time, gas
taken out to test their gas permeabilities again up to a confining
and radionuclides may leak from the repository. In this context,
pressure of 12 MPa. The entire experimental campaign is shown in
the main objective of this study was to determine whether, for a
high water content, this in situ compression (between the exter- Fig. 2. Loading–unloading cycles allowed certain irreversible prop-
nal and central part of the barrier or due to the convergence of the erties of the material to be visualized. The volumes were measured
host rock) could produce a hydraulic cutoff in the direction of the regularly during these steps using digital calipers that are accurate
gas flow. to within 1 ␮m. Two methods were generally used to determine
the effective gas permeability (Keff) depending on its magnitude:
Material preparation and experimental methodology the steady state, and transient methods (Billiotte et al. 2008). As
Materials and sample preparation indicated by some researchers, the steady state method is conve-
Based on preliminary studies performed by ANDRA (Bosgiraud nient and precise for permeabilities higher than 10−19 m2 (Billiotte
2004), a single bentonite–sand mixture was chosen to be used et al. 2008; Davy et al. 2007). When the permeability is below
throughout this study. The powder material, supplied directly by 10−19 m2, the steady state method usually requires an extremely
the French Atomic Energy Commission (Commissariat à l’énergie long time. In such cases, the pulse-test technique was used (Brace
atomique-CEA), was a mixture of 30% (by mass) of pure siliceous et al. 1968). In this study, the pulse-test method was only used for
sand and a conventional, so-called MX80 bentonite. The MX80 sample S2-3 after stabilization at RH = 98% and confining pressure
bentonite contained large quantities of montmorillonite (75%–90%), Pc = 5⬃12 MPa.

Published by NRC Research Press


Liu et al. 1161

Table 1. Particle-size distributions of bentonite (5) ⌬P1,2(t) ⫽ ⌬P1 exp(⫺cPf t)


powder and sand obtained by dry sieving.
Material Size (mm) Amount (%) with
Sand (silicon) 1.25 0.49
0.8
0.63
14.9
40.7
(6) c⫽ 冉
Keff A 1
␮g h V1

1
V2 冊
0.5 16.45
0.315 23.16
0.2 2.72 where the parameter c is a function of reservoir volumes, fluid
<0.2 1.57 compressibility, sample accumulation coefficient, and external
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

Bentonite powder 0.2–2.1 98 factors such as room temperature variations; V1 and V2 are the
>2.2 <1 volume of two reservoirs, and Pf is the final gas pressure, which
<0.212 <2 can be calculated from

⌬P1V1
(7) Pf ⫽ P1 ⫹
V1 ⫹ V2
Steady state method
As shown in Fig. 3, a circular cylindrical sample was jacketed
inside a protective Viton membrane and placed into a triaxial cell. The gas was assumed to be ideal and all tests were conducted in
A confining pressure (Pc) was applied and controlled by an oil an air-conditioned room at constant temperature. The pressure
pump. A buffer reservoir was installed between the gas source and difference ⌬P1,2 was measured by a pressure manometer and per-
sample. Gas was injected from the upstream side at a pressure P1 meability Keff was determined using eqs. (5) and (6).
(1.0 MPa) while the other side was maintained at atmospheric More information related to the two methods can also be found
pressure (P0). The gas (argon) flowed through the sample (along in previous publications by the authors: Meziani and Skoczylas
the x-axis) due to the pressure gradient. After a period (⌬t), the gas (1999) (steady state method), Loosveldt et al. (2002) (steady state
flow led to a pressure drop (⌬P) in the buffer reservoir. This pres- method), Davy et al. (2007) (steady state and pulse-test methods),
sure drop (⌬P) was much lower than the injection pressure (P1). Dana and Skoczylas (1999) (pulse-test method), Chen et al. (2012)
Therefore, it was assumed that the gas flow was at a steady flow (pulse-test method), and Duan et al. (2014) (pulse-test method). For
over the period (⌬t). According to Darcy’s law, the gas flow rate the purposes of this study, the confining pressure was between
For personal use only.

(Q mean) is expressed by 7 and 12 MPa.

⫺Keff A ⭸P Results and analysis


(1) Q mean ⫽
␮g ⭸x Characteristics of the samples after compacting and
wetting
where ␮g is the viscosity of argon and A is the cross-sectional area Figures 4a–4d show the main characteristics of the samples in
of sample. The variable Q mean is obtained from the law of conser- the experiment. Significant variations in w, degree of saturation
vation of mass (the mass of gas output from the reservoir buffer is (Sr), porosity (␾), and dry density (␳d) were observed from the
equal to the mass of the gas passing through the sample) initial compacted state, after stabilization at a given RH and also
after drying. Due to variation of the volume at different stages, ␳d
V0 ⌬P was calculated with the different volume, ␳d = md/VTi, where md is
(2) Q mean ⫽ the dry mass (at 105 °C oven-drying) and VTi is the corresponding
Pmean ⌬t volume measured at different stages. The water content after
compaction was on average 15.3% ± 0.6, which compares very well
where Pmean = P1 – ⌬P/2 is the average gas pressure and V0 is the to the target value of 15.2%. The dry densities (after compaction)
volume of buffer reservoir. The variable P1 in eq. (1) is the inner gas of the five samples are almost the same as the target value of
pressure in the sample and is assumed to be a function of coordi- 1.77 g/cm3.
nate x and time t (Dana and Skoczylas 1999) Following the initial permeability measurements, samples S2-3,
S2-4, S2-5, and S2-8 were held at specific relative humidity and

(3) P(x, t) ⫽ 冑P(0, t) 共1 ⫺ hx 兲 ⫹ P(h, t) hx


2 2
free-swelling conditions (RH = 75%⬃98%). The results in Fig. 5
show the variations in w, Sr, ␾, and ␳d with respect to those mea-
sured after the initial gas permeability test. At first, the variation
in w is very large at higher values of RH (up to 64.5% at RH = 98%),
where h is the sample height, P(0, t) = Pmean, and P(h, t) = P0. Using whereas the change in Sr is limited and independent of the change
eqs. (1)–(3), the gas permeability may be obtained from in RH. This phenomenon is explained by addressing the change in
porosity (e.g., 35.7% at RH = 98%) or dry density (e.g., 18.3% at RH =
␮gQ mean 2hPmean 98%). This means that there exists a competition between an in-
(4) Keff ⫽
A 共P2 ⫺ P2兲
mean 0
crease in water content and an increase in porosity, which leads to
a limited change in the degree of saturation. The combinations of
Pulse-test method these changes are the main factors that lead to the variations in
Gas pressure P1 (2 MPa) was applied initially to both sides of the Keff of the samples, which are analyzed later.
sample from two reservoirs R1 and R2 (see Fig. 3). Then, a slight The swelling behavior of the compacted bentonite–sand mix-
excess pressure (⌬P, 0.2 MPa) was applied to reservoir R2 to form a ture is mainly due to the swelling of bentonite, which contains the
pressure difference between both sides of the sample. Gas flowed swelling clay mineral (Na-montmorillonite). During the hydration
from reservoir R2 to reservoir R1 through the sample due to the process, water molecules are absorbed into the interlayers of
pressure gradient. According to Dana and Skoczylas (1999), the the crystal lattice, causing it to expand. The hydration of the ex-
pressure difference (⌬P1,2) between both ends of the sample may changeable cations largely governs this adsorption and that of the
be fitted by resulting increase in the interlayer spacing. With increases in the

Published by NRC Research Press


1162 Can. Geotech. J. Vol. 52, 2015

Fig. 2. Experimental procedure for test series S2.


Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

RH, this behavior becomes more active and the distance between unloading; see Fig. 6). Point 2 corresponds to the material, which
the layers will continue to increase until it stabilizes. As a result, was kept at a controlled humidity, and is the starting point of the
the sample’s volume will increase during the wetting process. loading–unloading cycle under the confining pressure. Point 3 is
obtained after unloading at a confining pressure of 1 MPa. A proper
Initial gas permeability after compaction comparison uses points 1 and 2, rather than 0 and 2, because the
To provide an accurate comparison of the changes in gas per- sample only experienced free swelling (or shrinkage) during this
For personal use only.

meability, an initial gas permeability measurement campaign just period without any other operation. The permeabilities corre-
after compaction was carried out and this was selected as a refer- sponding to points 1, 2, and 3 are denoted as Kg1, Kg2, and Kg3,
ence (phase T2, see Fig. 2). As shown in Fig. 4d, the porosities respectively.
obtained after confinement (␾T2) are systematically lower than
those after compaction (␾T0) even though the compaction was Sample S2-8 (RH = 75%)
carried out under greater pressures. This is a consequence of the The effective gas permeability of sample S2-8 (RH = 75%) is shown
fact that when the material is removed from the compaction in Fig. 7. The water content of this sample increased slightly (from
mold, it expands immediately, and re-compaction at 5 MPa leads 13.55% to 14.60%; see Fig. 4a) during the water retention test. From
to an additional irreversible reduction in volume. It must also be the perspective of the water content, Kg2 should be smaller than
noted that the effect of the pressure applied during compaction is Kg1. However, Fig. 7 indicates the opposite phenomenon. This
slightly different than that applied for confinement: (i) the com- difference is completely justified in view of the changes in poros-
paction pressure is uniaxial and the confining pressure is isotropic ity (because Sr essentially stayed the same: from 0.70 to 0.71), and
and (ii) the compaction takes less time than for a gas permeability in particular with the comparison between ␾T2 and ␾T3, which
test (usually several hours for a single loading–unloading cycle). reveals a slight increase (approximately 4.5%; see Fig. 5) between
All of the initial Keff values at Pc = 1 MPa lie in the range from points 1 and 2. This porosity increase is a result of wetting of the
10−16 to 10−15 m2. In the case of sample S2-2, which was placed material as well as the influence of various unavoidable manipu-
under a pressure of up to 12 MPa, an irreversible reduction in Keff lations required during the tests. The increase (of one order of
of one order of magnitude was observed (see Fig. 6) when the magnitude) in Keff indicates that the change in porosity plays a
confining pressure was reduced to its initial value of 1 MPa. The dominant role with respect to the change in water content. For
reduction in Keff of two orders of magnitude was observed when this sample, the impact of the volume (porosity) difference is
the confining pressure was increased to between 1 and 12 MPa. present over a wide range of confining pressures, so much so that
At this stage, Keff can already be considered to be very low. the latter was increased to 7 MPa for the initial Keff, which was
Therefore, a confining pressure leading to a decrease in porosity measured at 5 MPa, to ensure recovery. When the values of Keff at
(or an increase in dry density) has a very strong impact on the 5 MPa (during a loading cycle) are compared, the increase in water
material’s Keff. content does not lead to a reduction in Keff. However, loading to
12 MPa induced additional compression in the material, which
Permeability values obtained after equilibrium under was clearly irreversible because the data point Kg3 was much lower
humid conditions than Kg2 (and Kg1). This phenomenon will be discussed further
After the specimens were stabilized under different relative later.
humidities, new gas permeability measurements were performed
(phase T4, see Fig. 2). As stabilization took place under free- Sample S2-5 (RH = 85%)
volume conditions, both the water content and dry density of the The behavior of sample S2-5 (RH = 85%) is shown in Fig. 8.
samples changed. The values reached are those shown in Fig. 4 for Between points 1 and 2, the 85% RH led to a 10.6% increase in the
phase T3. Figures 7–10 show the influence of humidity associated water content and a 0.8% increase in the degree of saturation (see
with the confining pressure on Keff. The confining pressure was Fig. 5). In sample S2-8, the value of Kg2 was considerably greater
applied in a single loading–unloading cycle from 1 to 12 MPa. The than Kg1. This phenomenon can be explained by the difference of
same representation is used in all of the figures: points 0 and 1 are approximately 6.2% between the porosities ␾T2 and ␾T3 (see Fig. 5).
related to the measurements carried out after compaction However, the influence of the confining pressure is more sensitive in
(0 corresponds to the first data point and 1 to the data point after this second case, because at Pc = 5 MPa Kg–RH = 85% (gas permeability

Published by NRC Research Press


Liu et al. 1163

Fig. 3. Schematic diagram of experimental set-up (triaxial cell and gas panel): (a) steady state method; (b) pulse-test method.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16
For personal use only.

measured after stabilization at RH = 85%) is lower than Kg–ini (gas pronounced for a confining pressure of 1 MPa; the reason for this
permeability measured at initial state), which is different than the hysteresis will be explained in the next case.
previous case (sample S2-8 RH = 75%). Generally, the volume
changes due to wetting lead to corresponding changes in the ma- Sample S2-4 (RH = 92%)
terial strength and stiffness. Because sample S2-5 (at RH = 85%) Figure 9 shows the behavior of sample S2-4 (RH = 92%). There
absorbed more water and swelled more, it was more compressible was a 27.3% increase in the water content and a 4.5% increase in
than sample S2-8 (at RH = 75%), which led to a greater sensitivity to the degree of saturation between points 1 and 2. Although the
the confining pressure for this case. relative differences in permeability between these two points were
At a confining pressure of 12 MPa, the permeability decreased smaller than in the previous two cases, the logic that a higher
by (nearly) three orders of magnitude. This decrease appears to be
water content leads to a lower gas permeability does not apply.
attributed mainly to compression resulting from the confinement
This result is attributed to the observed 13.3% change in porosity
pressure, which causes the pores (or cavities) in which the gas
flows to start collapsing. The return to 1 MPa (point 3) shows a (see Fig. 5). However, the confining pressure had an immediate
hysteresis of one order of magnitude of the permeability. If this is effect. As soon as the pressure reached 2 MPa, the permeability of
compared with the previous sample, comparable phenomena are point 1 was attained. At 5 MPa, the permeability at point B was
found for points 1, 2, and 3, with a greater sensitivity to the con- more than one order of magnitude smaller than at point A. It is
fining pressure for the case of the sample with a slightly greater difficult to distinguish between the factors governing this phe-
porosity (more compressible). Additionally, the hysteresis is not nomenon, which may result from the combination of a greater

Published by NRC Research Press


1164 Can. Geotech. J. Vol. 52, 2015

Fig. 4. (a) Water content, (b) degree of saturation, (c) dry density, and (d) porosity of test series S2 during experimental process. T0, after
compaction; T1, before gas permeability test; T2, after gas permeability test: initial state; T3, stable at a given RH, S2-8: 75%, S2-5: 85%,
S2-4: 92%, and S2-3: 98%; T4, after gas permeability test: RH; T5, 60 °C oven-drying; T6, after gas permeability test: 60 °C oven-drying.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16
For personal use only.

Fig. 5. Changes in dry density, porosity, degree of saturation, and Fig. 6. Effective gas permeability, K (referred to as Keff in the text),
water content due to increasing RH levels (T3) using values of test series S2 after initial compaction.
measured after initial gas permeability tests as a reference (T2).

10−20 m2. At this stage, although it was not fully saturated (before
testing), the material became virtually impermeable to gas.
Evidence of deformability greater than seen in the preceding
cases was found in the hysteresis when the confining pressure
level of saturation and a higher deformability. In practice, the returned to 1 MPa, at which point the permeability was character-
confining pressure has a very strong influence on the permeabil- ized by a difference of more than two orders of magnitude (points
ity because at 7 MPa, the order of magnitude of the permeability 2 and 3). Many researchers have indicated that the swelling defor-
was only 10−18 m2 and at 12 MPa, the permeability dropped to mation is caused by both macrostructural and microstructural

Published by NRC Research Press


Liu et al. 1165

Fig. 7. Effective gas permeability of sample S2-8 after initial Fig. 9. Effective gas permeability of sample S2-4 after initial
compaction (Kini) and after stabilization at RH = 75% (KRH = 75%) as a compaction and after stabilization at RH = 92% as a function of
function of confinement (upon loading and unloading). confinement (upon loading and unloading).
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

Fig. 8. Effective gas permeability of sample S2-5 after initial Fig. 10. Effective gas permeability of sample S2-3 after initial
compaction and after stabilization at RH = 85% as a function of compaction and after stabilization at RH = 98% as a function of
confinement (upon loading and unloading). confinement (upon loading and unloading).
For personal use only.

Sample S2-3 (RH = 98%)


Sample S2-3 was exposed to the highest level of relative humid-
deformation (Gens and Alonso 1992; Alonso et al. 1999; Sánchez ity: RH = 98%. The results of the tests with this sample are shown
et al. 2005). The former depends on the proximity of the applied in Fig. 10. At equilibrium, the increase in water content was 64.5%,
pressure to the apparent preconsolidation pressure, and the latter whereas the degree of saturation decreased by 0.9% (see Fig. 5).
depends on the applied pressure and the suction change (Lloret Despite this high water content, Fig. 10 shows that permeabilities
et al. 2003). There is also an interaction (a coupling) between the Kg1 and Kg2 were almost the same, with Kg2 being slightly higher
two structural levels (Gens and Alonso 1992). The microstructural than Kg1. At the same time, the material’s porosity increased by
expansion due to hydration gives rise to a reorganization of the 35.7%. The swelling is thus sufficient (at Pc = 1 MPa) to counterbal-
macrostructure, which becomes more porous, causing irrevers- ance the effect of saturation.
ible deformation to occur (Villar Galicia 2002; Lloret et al. 2003). A similar phenomenon was again seen where the sample was
The magnitude of the interaction (coupling) is higher for the sam- more sensitive to the confining pressure than the previous cases.
At Pc = 5 MPa, the permeability (point A) decreased by more than
ple with larger macrostructural porosity (e.g., sample S2-4), which
two orders of magnitude with respect to the initial conditions
leads to larger macrostructural volume variations that cannot be
(point B). Between 7 and 8 MPa, the material could be considered
recovered upon any subsequent loading–unloading. In addition, to be gas-tight, and unloading to 1 MPa produced a permeability
there is a decrease in the preconsolidation pressure during the three orders of magnitude weaker than at the beginning of the
hydration process. This will increase the size of the plastic cycle. Thus, there is an extremely strong coupling between the
domain, which also contributes to the high deformability and two structural levels in the material, which contributes to this
irreversibility phenomena. hysteretic phenomenon. In this case, the disruption of the mac-

Published by NRC Research Press


1166 Can. Geotech. J. Vol. 52, 2015

Fig. 11. Effective gas permeability of test series S2: after 60 °C Conclusions
oven-drying.
The initial aim of this study was to determine whether in the
case of a compacted mixture of MX80 bentonite and fine siliceous
sand, coupling between high hydrostatic stress and incomplete
saturation could cause this mixture to become gas-tight. The re-
sults show that wetting under free-swelling condition leads to
swelling of the compacted bentonite–sand mixture, and the de-
gree of swelling depends mainly on the ambient RH, as all the
samples had the same initial dry density. In the laboratory, vari-
ous confining pressures were applied to investigate these effects.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

It was determined that the gas-tightness (Keff at 10−20 m2) could be


obtained when the sample was first saturated at RH = 98% and
then subjected to a confining pressure of 7 MPa, which was close
to the swelling pressure of the compacted bentonite–sand mixture.
Additionally, irreversible deformation (or hysteresis) was observed
during the loading–unloading cycle. This phenomenon is due
mainly to interactions between the macrostructural and micro-
structural deformations. Swelling of the microstructure causes an
irreversible macrostructural deformation that cannot be recov-
ered upon subsequent loading–unloading. This irreversible defor-
mation is higher for the sample with a higher water content
because the coupling between the two structural levels is larger at
higher RH. With the increase of RH, the sample absorbed more
water, swelled more, and became more compressible. Meanwhile,
the preconsolidation pressure decreased during the wetting pro-
cess. Therefore, the sample’s Keff is more sensitive to the confining
rostructure is more severe due to larger swelling deformation that
pressure at higher RH.
developed during hydration, consisting of larger irreversible mac- In situ, the bentonite or bentonite-based materials swell (due to
rostructural deformation that cannot be recovered upon subse- wetting) and fill the initial voids (those between the blocks, and
For personal use only.

quent loading–unloading (Lloret et al. 2003). This irreversible the voids between the swelling clay and host rock). As a result, the
macrostructural deformation also contributes to the increase of physical characteristics (w, Sr, ␾, and ␳d) of the materials are
the degree of saturation (e.g., Sr = 0.92 at step T4; see Fig. 4b) when changed. These changes affect Keff directly. Increases (or decreases) in
compared with the corresponding value prior to the test (0.73 at Sr and ␾ have an opposite effect on Keff and are the predominant
step T3). factors that govern the changes in Keff. After filling the initial
voids, the swelling is confined by the host rock. As a result, a
Permeabilities obtained in the “dry” state swelling pressure is developed due to the “swelling” property of
To conduct this last phase of permeability measurements, the the material and the confined condition. Due to the existence of a
samples were partially dried in a drying oven at 60 °C and tested significant water saturation gradient in the barrier, the external
for gas permeability (phase T6; see Fig. 2). As shown in Fig. 4d, the surface of the massive barrier swells in contact with COx argillite
␾T5 porosities (60 °C oven-drying) were from 2.41% to 8.84% smaller and compresses the partially saturated core. This compression
than the ␾T0 initial porosities. Thus, there is a simultaneous loss of leads to a decrease in the porosity, ␾, causing an increase in Sr. As
mass due to drying and shrinkage due to desiccation. These two a result, gas-tightness may be obtained due to their coupling ef-
effects had opposite influences on Keff, as shown in Fig. 11. The fects.
drying effect dominates, with an increase in permeability of two To summarize, for expansive materials, wetting (or drying) un-
to three orders of magnitude with respect to the initial value der free-swelling conditions leads to changes in ␾ and w (Sr may
obtained after compaction. There is a real “confining pressure not be changed). The loading causes variations in ␾ and Sr (w may
effect” because as the pressure is increased to 12 MPa, a decrease not be changed). Among these parameters, ␾ and Sr are the main
in the permeability is observed, although the effect is not greater factors affecting Keff. Therefore, special attention should be given
than an order of magnitude. The material becomes more rigid to these variations when researching the gas-tightness of the ex-
because of the increase in the dry density and a priori, the confin- pansive materials.
ing pressure only led to very limited compression of the porous
space. There is a stronger hysteresis for the case of the initially Acknowledgements
more saturated material (due to lower dry densities) when the Research leading to these results has received funding from the
samples were returned to the initial confining pressure of 1 MPa. European Atomic Energy Community’s Seventh Framework Pro-
In addition, microcracking was observed after drying in the oven. gramme (FP7/2007-2011) under Grant Agreement No. 230357, the
That is, there are three factors that can influence the effective gas FORGE project. Support from the State Key Laboratory for
permeability at this stage: microcracking, changes in water con- GeoMechanics and Deep Underground Engineering, China Uni-
tent, and porosity. versity of Mining & Technology (SKLGDUEK1201) is also greatly
After the samples were saturated in a desiccator with RH = 100% acknowledged.
they were again oven-dried at 60 °C, and finally dried at 105 °C to
verify their initial water content values. The m60 °C masses were
References
Alonso, E.E., Vaunat, J., and Gens, A. 1999. Modelling the mechanical behaviour
from 0.25% to 0.42% higher than the m105 °C masses. These differ- of expansive clays. Engineering Geology, 54(1–2): 173–183. doi:10.1016/S0013-
ences are less than that of the pure MX80 bentonite, which is 7952(99)00079-4.
within the range of 2.44%⬃3.46% (Liu and Skoczylas 2014). These ANDRA. 2005. Dossier 2005 Argile: Tome architecture and management of a
results mean that oven-drying at 60 °C can remove most of the geological repository. Technical report.
Billiotte, J., Yang, D., and Su, K. 2008. Experimental study on gas permeability of
water in the pores of the bentonite–sand mixture (70/30), which is mudstones. Physics and Chemistry of the Earth, Parts A/B/C, 33: S231–S236.
not the case for the pure MX80 bentonite. doi:10.1016/j.pce.2008.10.040.

Published by NRC Research Press


Liu et al. 1167

Birgersson, M., Åkesson, M., and Hökmark, H. 2008. Gas intrusion in saturated ites. Engineering Geology, 71(3–4): 265–279. doi:10.1016/S0013-7952(03)
bentonite – a thermo dynamic approach. Physics and Chemistry of the Earth, 00140-6.
Parts A/B/C, 33: S248–S251. doi:10.1016/j.pce.2008.10.039. Komine, H. 2009. Self-sealing capability of some bentonite buffers in artificial
Bosgiraud, J. 2004. Scope of work. design, fabrication, assembly, handling and seawater. In Proceedings of the 17th International Conference on Soil
packaging of buffer rings. ANDRA. Tech. Rep. C CC ASTE 04-0520/B. Mechanics and Geotechnical Engineering (ICSMGE 2009), Alexandria, Egypt.
Brace, W.F., Walsh, J.B., and Frangos, W.T. 1968. Permeability of granite under pp. 2495–2498.
high pressure. Journal of Geophysical Research, 73(6): 2225–2236. doi:10.1029/ Komine, H. 2010. Predicting hydraulic conductivity of sand–bentonite mixture
JB073i006p02225. backfill before and after swelling deformation for underground disposal
of radioactive wastes. Engineering Geology, 114(3–4): 123–134. doi:10.1016/j.
Chen, W., Liu, J., Brue, F., Skoczylas, F., Davy, C.A., Bourbon, X., and Talandier, J.
enggeo.2010.04.009.
2012. Water retention and gas relative permeability of two industrial con-
Komine, H., and Ogata, N. 1996. Prediction for swelling characteristics of com-
cretes. Cement and Concrete Research, 42(7): 1001–1013. doi:10.1016/j. pacted bentonite. Canadian Geotechnical Journal, 33(1): 11–22. doi:10.1139/
cemconres.2012.04.003. t96-021.
Cho, W.J., Lee, J.O., and Chun, K.S. 1999. The temperature effects on hydraulic Liu, J.F., and Skoczylas, F. 2014. Etude expérimentale sur Bentonite issue du
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by University of Mining Technology on 07/04/16

conductivity of compacted bentonite. Applied Clay Science, 14(1–3): 47–58. démantèlement d’un ouvrage souterrain à Mont Terri – Suisse. ANDRA.
doi:10.1016/S0169-1317(98)00047-7. Liu, J.F., Davy, C.A., Talandier, J., and Skoczylas, F. 2014. Effect of gas pressure on
Cui, Y.J., Tang, A.M., Loiseau, C., and Delage, P. 2008. Determining the unsatu- the sealing efficiency of compacted bentonite-sand plugs. Journal of
rated hydraulic conductivity of a compacted sand–bentonite mixture under Contaminant Hydrology, 170: 10–27. doi:10.1016/j.jconhyd.2014.09.006.
constant-volume and free-swell conditions. Physics and Chemistry of the PMID:25305640.
Earth, Parts A/B/C, 33: S462–S471. doi:10.1016/j.pce.2008.10.017. Lloret, A., Villar, M.V., Sanchez, M., Gens, A., Pintado, X., and Alonso, E.E. (2003).
Dana, E., and Skoczylas, F. 1999. Gas relative permeability and pore structure of Mechanical behaviour of heavily compacted bentonite under high suction
sandstones. International Journal of Rock Mechanics and Mining Sciences, changes. Géotechnique, 53(1): 27–40. doi:10.1680/geot.2003.53.1.27.
36(5): 613–625. doi:10.1016/S0148-9062(99)00037-6. Loosveldt, H., Lafhaj, Z., and Skoczylas, F. 2002. Experimental study of gas and
Davy, C.A., Skoczylas, F., Barnichon, J.-D., and Lebon, P. 2007. Permeability of liquid permeability of a mortar. Cement and Concrete Research, 32(9): 1357–
macro-cracked argillite under confinement: gas and water testing. Physics 1363. doi:10.1016/S0008-8846(02)00793-7.
and Chemistry of the Earth, Parts A/B/C, 32(8–14): 667–680. doi:10.1016/j.pce. Meziani, H., and Skoczylas, F. 1999. An experimental study of the mechanical
2006.02.055. behaviour of a mortar and of its permeability under deviatoric loading. Ma-
Davy, C.A., Skoczylas, F., Lebon, P., and Dubois, T. 2009. Gas migration proper- terials and Structures, 32(6): 403–409. doi:10.1007/BF02482711.
ties through a bentonite/argillite interface. Applied Clay Science, 42(3–4): Ortiz, L., Volckaert, G., and Mallants, D. 2002. Gas generation and migration in
639–648. doi:10.1016/j.clay.2008.05.005. Boom Clay, a potential host rock formation for nuclear waste storage. Engi-
neering Geology, 64(2–3): 287–296. doi:10.1016/S0013-7952(01)00107-7.
De La Vaissière, R. 2013. Hydration versus gas percolation in bentonite. In-situ
experiment PGZ2. Experimental borehole results. FORGE Report D3.17- Sánchez, M., Gens, A., do Nascimento Guimarães, L., and Olivella, S. 2005. A
D3.18. double structure generalized plasticity model for expansive materials. Inter-
national Journal for Numerical and Analytical Methods in Geomechanics,
Delay, J., Lesavre, A., and Wileveau, Y. 2008. The French underground research
29(8): 751–787. doi:10.1002/nag.434.
laboratory in Bure as a precursor for deep geological repositories. Reviews in
Villar Galicia, M.V. 2002. Thermo-hydro-mechanical characterisation of a ben-
Engineering Geology, 19: 97–111. doi:10.1130/2008.4119(10).
tonite from Cabo de Gata: a study applied to the use of bentonite as sealing
For personal use only.

Duan, Z., Davy, C.A., Agostini, F., Jeannin, L., Troadec, D., and Skoczylas, F. 2014.
material in high level radioactive waste repositories. Publicaciónes técnicas
Gas recovery potential of sandstones from tight gas reservoirs. International
(Empresa Nacional de Residuos Radiactivos). Vol. 4, pp. 15–258.
Journal of Rock Mechanics and Mining Sciences, 65: 75–85. doi:10.1016/j.
Villar, M., and Lloret, A. 2008. Influence of dry density and water content on the
ijrmms.2013.11.011.
swelling of a compacted bentonite. Applied Clay Science, 39(1–2): 38–49.
Dueck, A. 2008. Laboratory results from hydro-mechanical tests on a water doi:10.1016/j.clay.2007.04.007.
unsaturated bentonite. Engineering Geology, 97(1–2): 15–24. doi:10.1016/j. Wang, Q., Tang, A.M., Cui, Y.-J., Delage, P., and Gatmiri, B. 2012. Experimental
enggeo.2007.11.001. study on the swelling behaviour of bentonite/claystone mixture. Engineering
Gatabin, C. 2005. Selection and THM characterization of the buffer material. Geology, 124: 59–66. doi:10.1016/j.enggeo.2011.10.003.
ANDRA. Tech. Rep. RT DPC/SCCME 05-704-B. Wang, Q., Minh Tang, A., Cui, Y.-J., Delage, P., Barnichon, J.D., and Ye, W.M. 2013.
Gens, A., and Alonso, E.E. 1992. A framework for the behaviour of unsaturated The effects of technological voids on the hydro-mechanical behaviour of
expansive clays. Canadian Geotechnical Journal, 29(6): 1013–1032. doi:10.1139/ compacted bentonite–sand mixture. Soils and Foundations, 53(2): 232–245.
t92-120. doi:10.1016/j.sandf.2013.02.004.
Jedináková-Křižová,V. 1998. Migration of radionuclides in the enviroment. Jour- Ye, W.M., Wan, M., Chen, B., Chen, Y.G., Cui, Y.-J., and Wang, J. 2012. Tempera-
nal of Radioanalytical and Nuclear Chemistry, 229(1–2): 13–18. doi:10.1007/ ture effects on the unsaturated permeability of the densely compacted
BF02389439. GMZ01 bentonite under confined conditions. Engineering Geology, 126: 1–7.
King, F., Ahonen, L., Taxén, C., Vuorinen, U., and Werme, L. 2001. Copper corro- doi:10.1016/j.enggeo.2011.10.011.
sion under expected conditions in a deep geologic repository. Vol. 1. Svensk Yong, R.N., Boonsinsuk, P., and Wong, G. 1986. Formulation of backfill material
Kärnbränslehantering AB. for a nuclear fuel waste disposal vault. Canadian Geotechnical Journal, 23(2):
Komine, H. 2004. Simplified evaluation for swelling characteristics of benton- 216–228. doi:10.1139/t86-031.

Published by NRC Research Press

View publication stats

You might also like