Engineering Geology 192 (2015) 33–45

Contents lists available at ScienceDirect

Engineering Geology

journal homepage: www.elsevier.com/locate/enggeo

Engineered barrier of bentonite pellets and compacted blocks: State after

reaching saturation
José Luis García-Siñeriz a, María Victoria Villar b,⁎, María Rey a, Beatriz Palacios a
a
AITEMIN, C/ Margarita Salas 14, Parque Leganés Tecnológico, 28918 Leganés, Madrid, Spain
b
CIEMAT, Avd. Complutense 40, 28040 Madrid, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The EB experiment was a large-scale test performed in the Underground Research Laboratory of Mont Terri
Received 27 May 2014 (Switzerland) for the demonstration of an engineered barrier concept for nuclear waste disposal consisting of
Received in revised form 30 January 2015 the simultaneous use of high-density bentonite blocks and a lower-density bentonite pellets mixture (the gran-
Accepted 1 April 2015
ular buffer material, GBM). For that purpose, a gallery was excavated in the Opalinus clay and a dummy waste
Available online 9 April 2015
canister was placed on a bed made of bentonite blocks and surrounded by the GBM material. The bentonite bar-
Keywords:
rier was artificially hydrated with Pearson water and after 10.5 years of operation at isothermal conditions it was
Engineered barrier considered that the bentonite was completely saturated and the dismantling of the barrier was undertaken. A
Bentonite blocks sampling campaign was done to assess the final state of the bentonite barrier with regards to dry density and
Bentonite pellets water content.
Nuclear waste Upon dismantling, the GBM looked perfectly homogeneous, with every void and gap between the different ele-
Swelling ments (blocks/GBM, GBM/host rock, GBM/canister, etc.) having been sealed. Full saturation had been reached all
Large-scale test through the barrier. Moreover, the dry density of the blocks had decreased to values similar to those of the GBM,
Saturation
and the average water contents for both kinds of materials were similar. Nevertheless, the initial conditions of the
system did have a certain impact on the final distribution of dry density and water content: the bottom of the bar-
rier had a chance to quicker and higher water uptake (due to the heterogeneities in the initial porosity and char-
acteristics of the artificial hydration system), which gave place to immediate swelling that resulted irreversible,
with permanent higher water contents and lower dry densities towards the floor and back of the gallery, partic-
ularly in the GBM. Despite these heterogeneities, the water contents and dry densities of the whole barrier (GBM
and blocks) were much more homogeneous than at the beginning of the test and remained within a relatively
narrow range. The bentonite degree of saturation was homogeneous and very close to 100% all through the
barrier.
The feasibility and performance of this kind of initially heterogeneous barrier was proved in that it had an optimal
sealing capacity and developed acceptable swelling pressures between 1.3 and 2.2 MPa.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
The backfilling and sealing of deposition galleries or holes and access
galleries and shafts is an important part of nuclear waste underground
repository design. Any opening created during the construction of the
repository is a potential preferential pathway for water, gas and radio-
nuclide migration, and has to be effectively sealed. Bentonite or
bentonite-based mixtures have been proposed as backfill and sealing
materials for their low permeability, high swelling capacity and high re-
tention capacity. These materials have been widely characterised in the
form of high-density pre-compacted blocks, and their performance
under repository conditions has been also mainly assessed in large-
⁎ Corresponding author.
E-mail addresses: jlg.sineriz@aitemin.es (J.L. García-Siñeriz), mv.villar@ciemat.es
(M.V. Villar), maria.rey@aitemin.es (M. Rey), beatriz.palacios@aitemin.es (B. Palacios).
scale experiments in which the barrier around the waste packages
consisted of compacted blocks, such as the FEBEX mock-up and in situ
tests (ENRESA, 2006; Villar et al., 2005, 2012) or the Prototype Reposi-
tory (Börgesson et al., 2002).
The use of high-density bentonite pellets (combined or not with
powdered bentonite) was also proposed long ago (Salo and Kukkola,
1989; Volckaert et al., 1996). A mixture of high-density bentonite pel-
lets and bentonite powder was used in the large-scale demonstration
experiment RESEAL, carried out in a vertical shaft under isothermal con-
ditions at the Hades Underground Research Facility in Mol, Belgium
(Volckaert et al., 2000; van Geet et al., 2009). Bentonite pellets were
also proposed to backfill the slot between bentonite blocks and the
host rock in deposition holes, such as in the Prototype Repository in
situ experiment (Svensk Kärnbränslehantering AB, 2002). The advan-
tage of this material is that it is easy to manufacture and instal: the ben-
tonite pellets can be emplaced using auger discharge tubes or even

http://dx.doi.org/10.1016/j.enggeo.2015.04.002
0013-7952/© 2015 Elsevier B.V. All rights reserved.
34 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45
pneumatic projection techniques. Thus, the backfilling operation be-
comes an easier and potentially robotised procedure. On the other
hand, the problem of this emplacement technique is that the final values
of the key parameters of the barrier (such as dry density and permeabil-
ity), which depend on the initial characteristics of the backfill material,
the emplacement technique used and the hydro-mechanical interaction
between the backfill and the surrounding rock, have not been yet dem-
onstrated (Mayor et al., 2005). Besides, in the case of horizontal place-
ment a rigid support for the canister is needed, what can be achieved
by a bentonite block bed on which the canister is laid, and this must
be used in combination with the pellets' backfill. Although the hydro-
mechanical characterisation of pellets performed in laboratories has
shown that their saturated permeability and swelling pressure are
mainly controlled by the overall dry density of the sample rather than
the initial grain size distribution, and are thus similar to those of benton-
ite blocks of the same dry density (Imbert and Villar, 2006; Hoffmann,
2005; Alonso et al., 2010), the combined use of pellets and blocks in
the same section of the barrier introduces difficulties in the understand-
ing and modelling of the system performance that have to be addressed.
The Engineered Barrier Emplacement Experiment in Opalinus Clay,
“EB” Experiment, aimed the demonstration of a new concept for the
construction of nuclear waste repositories in competent clay formations
using horizontal deposition drifts (Mayor et al., 2005). The principle of
the new construction method was based on the combined use of a
lower bed made of highly compacted bentonite blocks in which the
waste canister rested and an upper buffer made of granular bentonite
material, consisting of a mixture of different-sized bentonite pellets.
To asses that this backfilling technique was feasible, it had to be first
demonstrated that the final values of the key parameters of the clay
buffer that may be achieved with this method, such as dry density,
Fig. 1. EB drift at Mont Terri URL, longitudinal and cross sections (Mayor et al., 2005).
swelling pressure and permeability, were suitable within the limits con-
sidered in the Performance Assessment of the repository concept, and
this was precisely the objective of the EB experiment. This paper pre-
sents a brief description of the configuration of the test and its opera-
tion, and focuses on the observations made and results obtained
concerning the state of the bentonite barrier upon its dismantling
after reaching full saturation.
2. The EB experiment: configuration and operation
The full-scale Engineered Barrier Emplacement Experiment (EB)
was carried out in a horizontal tunnel specially excavated with a road
header in the Opalinus clay formation at the Mont Terri Underground
Research Laboratory (URL) in Switzerland. The EB tunnel was 15 m
long and had a geometry of a horseshoe section, 2.55 m high and 3 m
wide (Fig. 1). A dummy canister, with the same dimensions and weight
as the Spanish reference canister (ENRESA, 1995), was placed on top of
a pedestal of highly compacted bentonite blocks which lay over a con-
crete bed. The remaining volume of the drift was backfilled with a gran-
ular material made of highly compacted bentonite pellets (GBM).
Finally, the drift was sealed with a mass-type concrete plug, with a con-
crete retaining wall between the plug and the GBM. Due to the short
amount of free water available in this formation, an artificial hydration
system consisting of a series of porous tubes that crossed along the
GBM and the bentonite blocks was installed to accelerate the hydration
process in the bentonite (Fig. 2). The hydration tubes were connected in
such a way that the water went into the material from the front to the
back and from the floor to the roof of the gallery. To enhance the
water homogeneous distribution, the concrete bed, the surface of the
container and the three rings of the bentonite blocks bed were covered
 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45 35

the emplacement and density inhomogeneities were acknowledged

due to the restrictions imposed by the components of the artificial hy-
dration system. According to the laboratory characterisation of the
GBM (Mayor et al., 2005; Hoffmann, 2005), for these dry density values
it was estimated that the saturated hydraulic conductivity should be
lower than 5 · 10− 12 m/s and the swelling pressure about 1.3 MPa.
However, the average dry density of the EB barrier should be higher,
due to the contribution of the compacted blocks, and consequently the
final average hydraulic conductivity and dry density of the barrier
should be related to this higher average dry density.
Since the end of the test installation the evolution of the different
thermo-hydro-mechanical parameters were monitored, both in the
barrier and the rock. Relative humidity and temperature in the rock
and in the bentonite buffer, rock displacement, pore pressure and total
pressure were registered by means of different types of sensors. The in-
stallation of the experiment, including dummy canister, sensors, hydra-
tion system and engineered barrier was carried out between November
2001 and April 2002 (Mayor et al., 2007). The artificial hydration of the
bentonite started in May 2002 and ended in June 2007. There was an
Fig. 2. Appearance of the concrete and bentonite block bed, dummy canister and hydration
system before the installation of the GBM (Mayor et al., 2005). initial hydration phase with an important amount of water injected
(6.7 m3 in two days) that was stopped due to a leakage along the plug
contact with the rock. After letting the bentonite swell for several
with geotextile. Pearson water coming from a deposit was used for hy- weeks, the hydration was resumed and from September 2002 to June
dration. The Pearson water is a predominantly sodium-chloride solution 2007 there were different hydration phases, during which the injection
of 19 g/L salinity and has a composition similar to the Opalinus Clay for- pressure was increased until reaching a maximum of 1.6 bar. A total of
mation pore water (Pearson, 1998). Its chemical composition is indicat- 18.9 m3 of water were injected, although this value does not represent
ed in Table 1. the actual barrier water intake, since leaks were acknowledged. The de-
The GBM material was a mixture of different-sized pellets with dry tailed record of effective water inflow for bentonite hydration is includ-
density higher than 2.1 g/cm3 and was installed using an auger dis- ed in García-Siñeriz et al. (2008). The monitoring of the experiment
charge tube (Fig. 3). According to the data available from the experi- continued even after no more water inflow was recorded and the hy-
ment installation and given the difficulties to accurately measure the dration phase was considered to have finished. Most of the relative hu-
volume occupied by the GBM, the initial average dry density of the midity sensors recorded values of 100% in less than a year after the
GBM would be between 1.3 and 1.36 g/cm3. Some segregation during beginning of hydration (Fig. 4). The evolution of total pressure inside
the buffer was more gradual though, with the sensors reaching steady
values between 1.3 and 2.2 MPa – depending on their location – after
Table 1 approximately 4 years (Fig. 5). The test run under isothermal conditions
Chemical composition of Pearson water (mg/L). (average temperature 16 °C) for 10.5 years.
Cl− SO2−
4 HCO−
3 Mg2+ Ca2+ Na+ K+ Sr+ pH
3. Dismantling and sampling
10,636 1354 26 413 1034 5550 63 45 7.6

Having the evidence that the buffer was fully saturated (as con-
firmed by the high relative humidities measured and in particular by
Gravity fall the stabilisation of the pressures inside the buffer), the main objective
of the dismantling of the EB experiment was to know the real status of
the GBM: degree of saturation, permeability, density, aspect, homoge-
neity, etc. It was also important to check the status of the bentonite
Hydration blocks that support the canister, the rock in contact with the buffer
and the degree of saturation of the concrete in the vicinity of the buffer
tubes (plug and concrete bed). Therefore, the activities of the dismantling
were coordinated with a sampling programme intended to analyse
the parameters of interest on-site as well as in the laboratories of the
different organizations involved (e.g. Stroes-Gascoyne et al., 2013; Villar
et al., 2014).
Canister The dismantling of the test started in October 2012 with the demo-
GBM lition of the concrete plug, which took almost a month, and went on
until February 2013. As the dismantling moved forward, all the experi-
ment components such as hydration pipes, sensors or metallic piping
from boreholes were also removed. The dismantling operation ended
with the removal of the Data Acquisition System. The monitoring of
the test was kept running during the whole dismantling process. A de-
tailed description of the dismantling operation was given in Palacios
et al. (2013).
Fig. 6 shows the appearance of the GBM and the bentonite blocks
upon exposure during dismantling, as well as details of the GBM-
blocks contact. The GBM looked completely homogeneous and every
Fig. 3. GBM emplacement using an auger discharge tube. void in the barrier had been filled. The contact between the blocks and
36 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45

Fig. 4. Evolution of relative humidity inside the buffer in sections B1 and B2 during the first phase of operation (Mayor et al., 2005).

the GBM was easily recognisable, since the blocks presented a coarse-
grained texture, whereas no grains could be identified in the GBM.
The pictures show also the numerous tubes and cables that crossed
the barrier and the blocks, the appearance of the steel container, the
concrete bed and the geotextile layers that separated the three rings
of blocks. Regarding the rest of elements of the experiment, such as hy-
dration pipes, cables and geotextile, all of them were found in perfect
conditions. In general the corrosion effects on metallic parts were
negligible.
Samples of the GBM, the bentonite blocks, the concrete bed and the
concrete plug, the Opalinus clay, and other materials were taken for
analysis in the laboratory. Altogether, almost 500 samples were taken.
Around 210 of them were analysed on-site and the rest were packed
and sent to the different partners' laboratories. Most of the bentonite
samples were taken in the vertical sections indicated in Fig. 7, following
approximate radii (Fig. 8). Sections A1, B1, E, B2 and A2 were instru-
mented. The origin of coordinates for location of every section and sam-
ple was set on the floor, in the centre of the outer face of the concrete
plug.
After testing several sampling methods, and in order to avoid exces-
sive disturbance, irregular pieces of GBM were extracted with a hammer
(a geologist one or an electric percussion hammer with a flat-end bit

Fig. 5. Evolution of total pressure inside the buffer (section E) during operation (Palacios et al., 2013, Appendix III).
 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45 37

Fig. 6. Appearance of the GBM (left up), the bentonite blocks (right up), the GBM-blocks contact (left bottom) and the hydration pipes (right bottom) upon dismantling. The rectangular
marks were caused by the tools used to dig the bentonite.

tool attached). For the extraction of the bentonite blocks, the percussion
hammer was equipped with a pointy bit in order to separate the blocks
and remove them in one whole piece. Before extracting the blocks, the
dimensions of some of them were measured. The removal of the ben-
tonite was done in such a way as to keep the face of work as parallel
to the plug as possible. From the moment the bentonite was reached,
a fast drying process was observed at the excavated bentonite front
(Fig. 9), so after every day of work, the excavation was covered with a
protection plastic. Besides, before starting sampling next day, at least
30 cm of bentonite were removed.
Two main parameters were determined on site: water content and
dry density (Palacios et al., 2013). These determinations were also per-
formed at CIEMAT laboratories in Madrid in bentonite samples that
were quickly packed after sampling and sent. The packaging consisted
of a plastic film and two aluminium-foil bags that were vacuumed be-
fore being sealed.

Fig. 7. Longitudinal cross-section of the EB test at Mont Terri. Bentonite sampling sections (Palacios et al., 2013).
38 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45

100

90
Hoffmann 2005
80
Villar 2012

Percentage passing
70

60

50

40

30

20

10

0
10 1 0.1 0.01
Fig. 8. Cross-section of a bentonite sampling section (section E) with indication of the sam-
Pellet diameter (mm)
ples taken (Palacios et al., 2013).
Fig. 10. Granulometric curve of the GBM used for the EB test as determined by dry sieving
in two laboratories.

4. Material

the GBM preparation. A detailed hydro-mechanical characterisation of

The GBM used in the EB experiment was prepared from FEBEX ben-
the GBM was performed by Hoffmann (2005).
tonite dried and milled in a three-step process to produce a fine grade
The montmorillonite content of the FEBEX bentonite is above
powder with a water content of 3.3%. Later, a commercial plant with
90 wt.%. The smectitic phases are actually made up of a smectite–illite
an in-line highly automated briquetting process produced coarse
mixed layer, with 10–15 wt.% of illite layers. The bentonite contains
(N7 mm) and fine (0.4–2 mm) grained materials with dry densities of
also variable quantities of quartz, plagioclase, K-feldspar, calcite, and
2.11 and 2.13 g/cm3, respectively. These two grain size fractions were
cristobalite–trydimite. The liquid limit of the bentonite is 102 ± 4%,
subsequently combined after several trials to produce a material with
the plastic limit 53 ± 3%, the specific gravity 2.70 ± 0.04, and 67 ± 3%
the granulometric curve shown in Fig. 10, which was used for the in
of particles are smaller than 2 μm. The hygroscopic water content is
situ emplacement (Mayor et al., 2005). On the other hand, the blocks
13.7 ± 1.3%. The total specific surface area obtained using the Keeling
used came from the series that was manufactured for the FEBEX in
hygroscopicity method is 725 m2/g. The cation exchange capacity is
situ test in 1997, and had a dry density of 1.69 g/cm3 and a water con-
102 ± 4 meq/100 g, the main exchangeable cations being calcium, mag-
tent of 14.4% (ENRESA, 2006).
nesium and sodium. The predominant soluble ions are chloride, sul-
The physico-chemical properties of the FEBEX bentonite, as well as
phate, bicarbonate and sodium.
its most relevant thermo-hydro-mechanical and geochemical charac-
The saturated hydraulic conductivity to deionised water (kw, m/s) of
teristics obtained during the projects FEBEX I and II were summarised
samples of FEBEX bentonite compacted at different dry densities of less
in the final report of the project (ENRESA, 2006). The FEBEX bentonite
than 1.47 g/cm3 and determined at room temperature, is related to dry
was extracted from the Cortijo de Archidona deposit (Almería, Spain).
density (ρd, g/cm3) according to the following equation:
The processing at the factory consisted of disaggregation and gently
grinding, drying at 60 °C and sieving by 5 mm, and this was the material
used for the laboratory characterisation, the block manufacturing and log kw ¼ −6:00ρd – 4:09: ð1Þ

The swelling pressure (Ps, MPa) of FEBEX samples flooded with

deionised water up to saturation at room temperature and constant vol-
ume conditions can be related to dry density (ρd, g/cm3) through the
following equation:

ln P s ¼ 6:77ρd –9:07: ð2Þ

This correlation was obtained from tests performed with bentonite

Fig. 9. Drying at the GBM front after 5 days of being exposed.
compacted with its hygroscopic water content, but it is considered
that the initial water content has little impact on swelling pressure
(Villar and Lloret, 2008).
Although this characterisation was performed in compacted granu-
lar material with a grain size of b 5 mm, it was proven that the saturated
hydro-mechanical properties of pellets mixtures are similar to those of
granulates (Imbert and Villar, 2006).
 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45
5. Methodology of postmortem tests
5.1. On-site determinations
The samples for the on-site determinations were taken from sec-
tions A1–25, B1, E, B2 and A2. After collecting every GBM sample, it
was cut and trimmed into three subsamples of between 6 and 12 cm3
in volume. The water content was obtained from all the subsamples
and the dry density from two of them.
The gravimetric water content (w) is defined as the ratio between
the mass of water and the mass of dry solid expressed as a percentage.
The mass of water was determined as the difference between the
mass of the sample and its mass after oven drying at 110 °C for 48 h
(mass of dry solid). Dry density (ρd) is defined as the ratio between
the mass of the dry sample and the volume occupied by it prior to dry-
ing. The volume of the specimens was determined by immersing them
in a recipient containing mercury and by weighing the displaced mercu-
ry. The same samples whose volumes had been determined were used
for the water content determination.
5.2. Laboratory determinations
The samples for the laboratory determinations were taken from sec-
tions CMT1, CMT2 and CMT3, and also duplicates of some of the samples
taken in sections E and B2. Until their analysis, the packed samples sent
from Mont Terri were kept at CIEMAT facilities in a storage room in
which the temperature was between 7 and 16 °C and the relative hu-
midity between 70 and 90%. The samples were taken one at a time out
of the storage room and unpacked in the laboratory.
The dry density and water content were determined in small sam-
ples to complement the determinations performed on site. These sam-
ples were prepared by trimming regular specimens of the right size,
with volumes of between 6 and 13 cm3. Two specimens were trimmed
and analysed from each GBM sample. The subsamples from each block
were taken at three different distances from the container, and for
each distance at least two specimens were used. The methodology
followed was the same as for the on-site determinations. Additionally,
in some cases larger samples were used just for water content
determination.
5.3. Evaluation of the methodology
Since most of the samples analysed had very high water contents,
much higher than hygroscopic, their manipulation could imply certain
water content change. The process of trimming took some minutes,
and during this time some drying of the sample could have taken
place, because the samples remained exposed to drier room conditions
than those in the barrier. In particular, the relative humidity and tem-
perature at the gallery of the Mont Terri URL where the determinations
were done were 48% and 17 °C, and the CIEMAT's laboratory conditions
during the determinations were relative humidity of 35 ± 6% and tem-
perature of 22 ± 1 °C.
In order to determine how this drying could affect the values of
water content obtained, some samples were allowed to air dry both in
the laboratory and in the gallery, and their weight change was periodi-
cally checked before putting the samples in the oven to determine their
dry mass. Taking into account that the trimming process took less than
20 min, it was concluded that the decrease in water content during ma-
nipulation could have been between 0.7 and 1.5%, which would imply a
decrease in the degrees of saturation computed, of 2% in the worst cases
(Palacios et al., 2013; Villar, 2013).
Additionally, a comparison of the results obtained on-site and in the
laboratory was made. Twin samples were taken in sections E and B2;
one of them was analysed on-site and the other one was sent to the lab-
oratories in Madrid. A total of 13 samples from section B2 and 17 from
section E were cross-checked. For section B2 the water contents
39
obtained in the laboratory were on average 0.7% higher than those ob-
tained on site, and the dry densities 0.2% higher. For section E the
water contents obtained in the laboratory were on average the same
as those obtained on-site and the dry densities 1.4% higher. The differ-
ences were very small, particularly regarding water content, conse-
quently they are not considered significant. This proved that the
packaging methods were good enough to keep the physical conditions
of the samples upon extraction. In the following, the determinations
performed on-site and in the laboratory are jointly considered.
6. Results
6.1. Dry density and water content
The samples for the water content and density determinations were
taken following approximately sampling radii in the GBM, as shown in
Fig. 8. As the dismantling operation started it was observed that the bot-
tom part of the barrier, particularly the zones closer to the concrete bed,
looked wetter than the rest of the GBM, and it was also decided to ana-
lyse the samples taken from the part of the barrier closer to the floor and
not belonging to any particular radius. About 300 bentonite samples
were analysed.
The results obtained in the different sampling sections were plotted
in contour maps using a grid-based mapping software that interpolates
irregularly spaced XYZ data into a regularly spaced grid. Kriging was
used as gridding method. Fig. 11 shows this representation for section
B2, which was one of the most thoroughly sampled. Each point in the
figures is the average of two or three measurements. The higher water
content and lower dry density of the GBM at the bottom of the barrier
becomes clear in this kind of representation, whereas the degree of sat-
uration did not display any particular spatial trend. The same kind of
technique was used to obtain longitudinal sections along the barrier.
Fig. 12 shows two of these sections for the water content, one corre-
sponding to the low part of the barrier (Y coordinate: 44–54 cm) and
the other one to the top (Y coordinate: 190–210). The former shows
the lower water content in the blocks in relation to the adjacent GBM,
whereas the latter displays the trend to find slightly higher water con-
tents in the GBM towards the back of the gallery. The dry density follows
the inverse behaviour.
From these representations, the software allows one to compute the
median values for each vertical sampling section, blocks and GBM in-
cluded. These values have been plotted in Fig. 13 as “GBM + blocks”.
The average of these median values for all the sampling sections
would be 35.4 ± 1.2% for the water content, 1.36 ± 0.02 g/cm3 for the
dry density and 97 ± 1% for the degree of saturation. Fig. 13, shows
also the average values for each sampling section computed from the in-
dividual measurements, separately for the GBM and for the blocks. The
trend for the water content and dry density change along the barrier ob-
served in Fig. 12 is also clear in this kind of representation, particularly
for the GBM.
All the results obtained for the GBM in terms of water content and
dry density are plotted as a function of the Y coordinate of the samples
in Figs. 14 and 15. The origin for this coordinate is the middle point of
the floor of the gallery (Fig. 8). Each point is the average of three mea-
surements for those determinations performed on-site and of two mea-
surements for the determinations performed in the laboratory. The
trend for the water content to increase and the dry density to decrease
towards the floor is very clear and can even be fit to a potential expres-
sion. The samples having higher water contents (higher than 45%) and
lower dry densities are those located at the lower part of the gallery,
close to the concrete bed. The average of all the GBM water content
measurements is 36.9 ± 4.3%, with the mode in 35.6%, and of the dry
density 1.33 ± 0.07 g/cm3, with the mode in 1.36 g/cm3. Note that
these average values are higher for the water content and lower for
the dry density than the average values found for the whole barrier,
blocks and GBM together (see above), whereas the mode values for
40 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45
Fig. 11. Water content, dry density and degree of saturation isoline graphs of section B2
(see Fig. 7 for location).
the GBM are very close to the average values for the whole barrier. Re-
garding the degrees of saturation, no trends along the GBM were
observed, most of the values being between 96 and 100% (average of
97 ± 4%).
The blocks were subsampled in more detail in the laboratory, in
three different levels: up, middle and bottom, according to the position
of the block in the barrier, i.e. to the distance to the canister. The results
obtained are plotted in Fig. 16 and summarised in Table 2. In sections
CMT1 and CMT2 the water content tended to increase and the dry
densities to decrease towards the concrete bed, that is, away from the
container. This was also the predominant trend in the sections sampled
on-site (A1, B1, E and B2). However, in section CMT3 these trends were
not observed. In any case, all the values are within a range of between 34
and 37% water content and between 1.33 and 1.39 g/cm3 dry density.
These ranges have been marked in Fig. 15, and are slightly above the av-
erage of the GBM for the dry density and below for the water content.
However, the mode for the GBM values is comprised between these
ranges, i.e. the values measured in the blocks are the same as those
more frequently found in the GBM. In terms of degree of saturation of
the blocks there were no clear spatial trends. The values measured
were between 95 and 101%.
6.2. Size of blocks
According to the measurements performed, the density of the ben-
tonite blocks decreased during the operation of the EB test from an ini-
tial value of 1.7 g/cm3 in the installation phase to values from 1.33 to
1.39 g/cm3. This implies necessarily an expansion of the blocks during
operation. The dimensions of some blocks were measured on-site as
dismantling proceed, just before extracting the blocks from the barrier,
in order to compare their dimensions to the original ones and evaluate
their deformation. There were three rings of bentonite blocks (#01, #02
and #03), those at the bottom in contact with the concrete bed, in the
middle part, and in the upper part, in contact with the dummy canister
(Fig. 1). The distance from the centre of the canister to the floor was also
measured. Fig. 17 shows a cross section of the shape of the blocks, with
indication of the dimension measured on site, the ‘c’ parameter, which
had an initial value of 21.4 cm. The dimensions of some of the blocks
sent to CIEMAT laboratories were also taken upon unpacking.
Overall 48 measurements were taken on site. In the type 3 blocks, i.e.
those in contact with the canister, an average decrease of 1.5% in the ‘c’
dimension was measured. However, in blocks of types 2 and 1 this value
increased to 4.4 and 4.7%, respectively, indicating that a significant ver-
tical expansion (Y axis) of the blocks had taken place. In fact, the exten-
someter measurements during the operational phase indicated an uplift
of the canister frontal end. The expansion along the Y axis was also de-
tected during the dismantling by the evolution of the distance between
the centre of the canister and the floor, which showed a rising of the
canister of approximately 3 cm from its original position. However,
this expansion took place only from Z = 461 backwards. From this
point the dismantling operation was modified, and instead of removing
simultaneously the GBM and the bentonite blocks located in the same
vertical section, the bentonite block bed was left on place whilst the
GBM front was taken out. In these sections in which the GBM was
taken away long before the blocks were extracted, more time was
allowed for them to swell once the overload of the GBM was removed.
Consequently, apart from the likely expansions developed in the ben-
tonite blocks during the operational phase, they had an additional ex-
pansion during the dismantling operation.
The measurements taken in the laboratory showed a clear expansion
of all the blocks. The dimension ‘c’ increased an average of 7% in sections
CMT2 and CMT3, indicating the uplift of the canister and confirming the
observations made on site. When comparing these measurements with
those taken on-site during dismantling, it is observed that the measure-
ments taken in the laboratory were larger, which could mean that the
expansion of the blocks continued after excavation. The difference be-
tween both measurements was particularly large for the blocks in sec-
tions CMT2 and CMT3. The time between excavation and sampling at
the laboratory was longer for the blocks in section CMT3.
The other dimensions taken in the laboratory increased also be-
tween 6 and 14% in section CMT1, which was the one closer to the con-
crete plug. In this section the thickness of the blocks increased up to 56%,
whilst in the two other sections the average thickness increase was of
4%, which indicates the longitudinal expansion of the bentonite bed in
the Z direction, much higher in the barrier front. The expansion of the
 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45 41

Fig. 12. Water content along the bentonite barrier in two longitudinal sections (bottom and top).

blocks along the Z axis was also proven by the evolution of the distance
between the front face of the canister and the front face of the concrete
bed. They were aligned when the EB dismantling started and at the end
of the dismantling operation there was a separation of about 4 cm be-
tween both.
7. Discussion
The performance of an engineered barrier for a nuclear waste repos-
itory composed of bentonite blocks (dry density 1.7 g/cm3, water
content 14%) and pellets (Granular Buffer Material, GBM, average
installed dry density around 1.3 g/cm3 and water content 3%) was
checked in a large-scale experiment performed in a tunnel excavated
in Opalinus clay. After five years of artificial hydration and a total of
10.5 years of observation at isothermal conditions the test was disman-
tled. The measurements of the relative humidity and total pressure sen-
sors placed in the barrier indicated that a steady state had been reached
and that the bentonite was fully saturated. The GBM looked homoge-
neous and no gaps or voids remained in the barrier system. Numerous
samples were taken for their analysis on-site and in the laboratory.

44 1.45
GBM GBM
1.43
42 blocks blocks
1.41
GBM+blocks GBM+blocks
Dry density (g/cm3)

1.39
40
Water content (%)

1.37

38 1.35
1.33
36
1.31
1.29
34
1.27
32 1.25
300 400 500 600 700 800 300 400 500 600 700 800

Distance to plug front (cm) Distance to plug front (cm)

Fig. 13. Average values of water content and dry density for the different sampling sections along the barrier (the “blocks” and “GBM” values are the average computed from all the mea-
surements in each material, the “GBM + blocks” values are the median vales computed for each section with the mapping software).
42 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45

52 towards the back of the gallery. The disposition of the hydration tubes,
A1-25 which were connected in such a way that the water went into the ma-
50 terial from the front to the back of the gallery and from the floor to
CMT1
the roof, could be an explanation for this observation. The higher
48 B1 water contents in these zones were accompanied by lower dry densi-
CMT2 ties, a consequence of swelling. This swelling seems to have been irre-
46
E versible, since the density difference among different parts of the
Water content (%)

44 B2 barrier remained after 10 years of operation.

The blocks had water contents in the order of the mode for the GBM,
A2
42 and their dry density decreased overall from an initial value of 1.7 g/cm3
CMT3 to values close to 1.4 g/cm3, also in the order of the mode of values found
40 in the GBM. Although the blocks were not analysed as thoroughly as the
GBM, no trends of change in their average values of dry density and
38 water content along the barrier (Z coordinate) were observed, even
mode GBM
though the vertical sections closer to the concrete plug displayed a
36 more heterogeneous distribution, with higher water contents and
lower dry densities towards the floor, i.e. towards the concrete bed.
34 The explanation could be the same as for the vertical GBM density and
water content distribution. The increase in the dimensions of the blocks
32
confirms that they swelled during the test and also during dismantling,
30 when the pressures were released. This is supported by the decreasing
0 50 100 150 200 250 pressure trend of the total pressure cells located in section E recorded
as the dismantling works in the front of the barrier were moving for-
y-coordinate (cm) ward (Fig. 18). In particular, the change in blocks vertical dimensions in-
dicated the uplift of the canister, which would have taken place mainly
Fig. 14. Water content of GBM samples from different sampling sections as a function of
during the final phases of dismantling and towards the back of the tun-
coordinate Y. The shadowed area shows the range of the measurements in blocks.
nel. On the contrary, the blocks closer to the concrete plug swelled
mainly in the longitudinal direction. The fact that the space between
The water contents of the GBM ranged between 31 and 52% and the the blocks and the retaining wall was filled by lower-density GBM
dry densities between 1.48 and 1.12 g/cm3, with a clear trend for the would have favoured the initial, irreversible swelling of the blocks in
water content to increase towards the lower part of the barrier. Two fac- the Z direction.
tors could have played a role in this distribution (Palacios et al., 2014). On average the degree of saturation of the GBM was 97 ± 3%, and of
Firstly, during the installation of the GBM and due to its broad distribu- the blocks 96 ± 3%. No vertical or longitudinal spatial trends were found
tion of grain sizes (Fig. 10), segregation could have occurred, which for this parameter. If the impact of the drying of the samples during
would cause an initial density gradient in the barrier. Secondly, the ef- their handling before being analysed were taken into account to correct
fect of gravity on the water distribution seems to have been relevant. the measurements, the actual water content of the samples before ex-
The accumulation of water in the lower part of the barrier took place traction could have been between 0.7 and 1.5% higher than the values
probably at the beginning of the experiment, favoured by the higher ini- measured, and the degrees of saturation could have increased up to a
tial porosity of the barrier bottom. The analyses showed also a slight in- 2%. In any case, the whole barrier can be considered as having reached
crease of the water content of the bentonite from the concrete plug full saturation and probably a steady hydraulic and mechanical state,
which is supported by the fact that the pressure sensors had not record-
ed changes for six years and the values measured were in the upper
1.50 range of the swelling pressures expected for the FEBEX bentonite
compacted to the range of dry densities found in the barrier (Fig. 19).
1.45 Additionally, it is considered that the average pore water density in
the barrier was close to 1.0 g/cm3, due to the fact that the dry density
1.40 of the bentonite was lower than 1.4 g/cm3.
In a complementary investigation, and in order to assess the hydrau-
Dry density (g/cm3)

lic performance of the barrier, the hydraulic conductivity of some of the

1.35 GBM samples retrieved was determined (Villar et al., 2014). The values
mode GBM measured were in the range from 8 · 10−12 to 2 · 10−13 m/s, mainly re-
1.30 lated to the dry density of the samples. These values were determined
using Pearson water as fluid. The comparison of these values to those
1.25 expected for untreated FEBEX bentonite of the same dry density perme-
A1-25 CMT1 ated with deionised water (Eq. (1)), shows that the values for the GBM
were slightly above the theoretical ones, but mostly inside the expected
1.20 B1 CMT2
range of variation. This can be attributed to the different salinity of the
E B2 water used in the tests, since a higher salinity would increase permeabil-
1.15 ity. Consequently, it is considered that the permeability of the GBM did
A2 CMT3 not change during operation and remained in the same order as that
1.10 of the untreated FEBEX bentonite compacted to the same dry densities.
0 50 100 150 200 250
y-coordinate (cm) 8. Summary and conclusions

Fig. 15. Dry density of GBM samples from different sampling sections as a function of co- The EB experiment was a large-scale test performed in the Under-
ordinate Y. The shadowed area shows the range of the measurements in blocks. ground Research Laboratory of Mont Terri (Switzerland) for the
 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45 43

40 1.41

39 1.40

1.39
38
w.c. CMT1
1.38
37 w.c. CMT2

Dry density (g/cm3)

Water content (%)
1.37
w.c. CMT3
36
1.36 d.d. CMT1

35 d.d. CMT2
1.35
d.d. CMT3
34
1.34

33 1.33

32 1.32
50 70 90 110 130

Distance from container axis (cm)

Fig. 16. Water contents (w.c.) and dry densities (d.d.) measured in blocks from three sampling sections (each point is the average of two measurements).

Table 2 groundwater, was used. The test was instrumented, and among other
Average values measured in blocks from different sections. parameters, the relative humidity and total pressure at different points
Section Dry density (g/cm3) Water content (%) Degree of saturation (%) inside the barrier were monitored. Most of the sensors recorded relative
CMT1 1.36 ± 0.02 35.4 ± 1.6 97 ± 2
humidity values of 100% only one year after the beginning of the test,
CMT2 1.38 ± 0.01 34.7 ± 1.1 98 ± 2 whereas it took four years for the total pressure sensors to record stable
CMT3 1.36 ± 0.01 35.5 ± 0.5 98 ± 1 values that kept approximately constant until the end of the test. Conse-
quently, after 10.5 years of operation it was considered that the benton-
ite was completely saturated and the dismantling of the barrier was
demonstration of an engineered barrier concept for nuclear waste dis- undertaken. The GBM looked perfectly homogeneous, with every void
posal consisting of the simultaneous use of high-density bentonite and gap between the different elements (blocks/GBM, GBM/host rock,
blocks (dry density 1.7 g/cm3, water content 14%) and a lower density GBM/canister, etc.) having been sealed. During dismantling a large
bentonite pellets mixture (the granular buffer material, GBM, with an number of samples of the different elements were taken, particularly
average installed dry density around 1.3 g/cm3 and a water content of of bentonite. Namely a thorough bentonite sampling was done to assess
3%). For that, a gallery was excavated in the Opalinus clay and a the final state of the barrier with regards to dry density and water con-
dummy waste canister was placed on a bed made of bentonite blocks tent. The analyses were performed both on-site and in the laboratory in
and surrounded by the GBM material. The bentonite barrier was artifi- Madrid, the comparison of both sets of results showing a very good
cially hydrated for five years under the isothermal conditions prevailing agreement.
in the gallery, and then it was allowed to evolve during five and a half Full saturation had been reached all through the barrier. Moreover,
years more. Pearson water, which simulates the clay formation the dry density of the blocks had significantly decreased and that of
the GBM had increased, and the water contents for both kinds of mate-
rials were similar. Nevertheless, the initial conditions of the system did
have a certain impact on the final distribution of dry density and water
content: the lower part of the barrier saturated earlier (probably be-
cause of initial porosity heterogeneities and the hydration system
configuration) and had a chance to a higher water uptake and immedi-
ate swelling that resulted irreversible, giving rise to permanent higher
water contents and lower dry densities towards the floor and back of
the gallery, particularly in the GBM. Despite these heterogeneities, the
water contents and dry densities of the whole barrier (GBM and blocks)
were much more homogeneous than at the beginning of the test
and remained within a relatively narrow range. Since the hydro-
mechanical properties of the bentonite, particularly swelling capacity
and permeability, depend mainly on dry density, it is expected that
the values for these properties be quite homogeneous through the bar-
rier. In fact, the pressure developed by the bentonite as measured by the
sensors remained during operation within an optimal range of between
Fig. 17. Cross section of the compacted bentonite blocks used (Mayor et al., 2005). 1.3 and 2.2 MPa and the hydraulic conductivity measured in the
44 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45

Fig. 18. Pressure evolution in section E during dismantling. The ascending line shows the progress of dismantling along the Z axis, with indication of the moments at which each section
was dismantled. The dotted vertical line indicates the moment from which the GBM was removed before the blocks (Palacios et al., 2013, Appendix III).

laboratory in samples retrieved was in the range from 8 · 10−12 to Acknowledgements

2 · 10−13 m/s.
In conclusion, the dismantling operation provided the information
requested: it was possible to inspect and analyse the GBM in order to
determine its properties. Despite the initial heterogeneity due to the
emplacement, its aspect and homogeneity were very good. The feasibil-
ity and performance of this kind of initially heterogeneous barrier was
proven also in that it had an optimal sealing capacity, swelling and per-
meability. Other hydro-mechanical and geochemical determinations
are being performed.
3.0
The first phase of the EB experiment – years 2000 to 2003 – was co-
financed by the European Commission (contract n° FIKW-CT-2000-
00017) and ENRESA (Spain). Between 2003 and 2009 the project oper-
ation continued under the support of the Mont Terri Consortium, being
ENRESA, BRG (Germany), ANDRA (France) and NAGRA (Switzerland)
the principal contractors, project 32.015: EB, phases 10 to 14. From
2010, the experiment was part of the PEBS project, receiving funding
from the European Atomic Energy Community's Seventh Framework
Programme (FP7/2007–2011) under Grant Agreement no249681. The
dismantling work was carried out by Beatriz Palacios, Cristina de la
Rosa, Antonio Martín de Vidales, David Castro, Miguel Ángel Casero
and José Luis Díaz from AITEMIN. The laboratory work was performed
by Juan Aroz and Ramón Campos at CIEMAT.

References
Swelling pressure (MPa)

2.0 Alonso, E.E., Hoffmann, C., Romero, E., 2010. Pellet mixtures in isolation barriers. J. Rock
Mech. Geotech. Eng. 2 (1), 12–31.
Börgesson, L., Gunnarsson, D., Johannesson, L.E., Sandén, T., 2002. Installation of buffer,
canisters, backfill and instruments in Section 1. SKB Progress Report IPR-02-23.
Swedish Nuclear Fuel and Waste Management Company (SKB), Stockholm, Sweden.
ENRESA, 1995. Almacenamiento Geológico Profundo de Residuos Radiactivos de Alta
Actividad (AGP). Diseños Conceptuales Genéricos. Publicación Técnica ENRESA 11/
1.0 95 (Madrid).
ENRESA, 2006. FEBEX Full-scale Engineered Barriers Experiment, updated final report
1994–2004. Publicación Técnica ENRESA 05-0/2006, Madrid (590 pp.).
García-Siñeriz, J.L., Rey, M., Mayor, J.C., 2008. The engineered barrier experiment at Mont
Terri Rock Laboratory. Sci. Technol. Ser. 334, 65–75 (Andra).
Hoffmann, C., 2005. Caracterización hidromecánica de mezclas de pellets de bentonita.
Estudio experimental y constitutivo. (PhD Thesis), Universitat Politècnica de
0.0 Catalunya, Barcelona (387 pp.).
Imbert, Ch., Villar, M.V., 2006. Hydro-mechanical response of a bentonite pellets/powder
1 1.1 1.2 1.3 1.4 1.5 mixture upon infiltration. Appl. Clay Sci. 32, 197–209.
3 Mayor, J.C., García-Siñeriz, J.L., Alonso, E.E., Alheid, H.-J., Blümbling, P., 2005. Engineered
Final dry density (g/cm )
barrier emplacement experiment in Opalinus Clay for the disposal of radioactive
waste in underground repositories. Final Report. Publicación Técnica ENRESA 02/05
Fig. 19. Empirical correlation between swelling pressure and dry density for the FEBEX (Madrid, 101 pp.).
bentonite (solid line, Eq. (2); the dotted lines indicate the 25% range of variation) and Mayor, J.C., García-Siñeriz, J.L., Alonso, E.E., Alheid, H.-J., Blümbling, P., 2007. Engineered
total pressure measured by the sensors before dismantling against the dry density of the barrier emplacement experiment in Opalinus Clay for the disposal of radioactive
samples retrieved in section E (squares). waste in underground repositories. In: Bossart, P., Nussbaum, C. (Eds.), Mont Terri
 J.L. García-Siñeriz et al. / Engineering Geology 192 (2015) 33–45
Project — Heater Experiment, Engineered Barriers Emplacement and Ventilation
Tests. Rep. Swiss Geol. Surv. 1, pp. 115–180.
Palacios, B., Rey, M., García-Siñeriz, J.L., 2013. Engineered barrier emplacement experi-
ment in Opalinus Clay: “EB” experiment. As-built of dismantling operation. PEBS
Deliverable-n°: D2.1–4. Contract Grant Agreement Number FP7 249681. AITEMIN
(95 pp.).
Palacios, B., García-Siñeriz, J.L., Mayor, J.C., 2014. Outcome of the dismantling of “EB” ex-
periment. In: Schäfers, A., Fahland, S. (Eds.), Proceedings International Conference
on the Performance of Engineered Barriers, 6–7 February 2014. BGR, Hannover,
pp. 45–52.
Pearson, F., 1998. Artificial waters for use in laboratory and field experiments with
Opalinus Clay. Paul Scherrer Institut TM (44-98-08).
Salo, J.-P., Kukkola, T., 1989. Bentonite pellets, an alternative buffer material for spent fuel
canister deposition holes. NEA/CEC Workshop “Sealing of Radioactive Waste Reposi-
tories”. (Braunschweig, Germany), OECD, Paris.
Stroes-Gascoyne, S., Frutschi, M., Hamon, C., Bagnoud, A., Leupin, O., Bernier-Latmani, R.,
2013. Microbiological analysis of samples from the Engineered Barrier Experiment
at Mont Terri Rock Laboratory. Final Report. TN 2013-56 (23 pp.).
Svensk Kärnbränslehantering AB, 2002. Äspö Hard Rock Laboratory. Annual Report 2001.
SKB Technical Report TR-02-10 (Stockholm. 184 pp.).
Van Geet, M., Bastiaens, W., Volckaert, G., Weetjens, E., Sillen, X., Maes, N., Imbert, C., Billaud,
P., Touzé, G., Filippi, M., Plas, F., Villar, M.V., García-Gutiérrez, M., Mingarro, M., Gens, A.,
Vallejan, B., 2009. RESEAL II. A large-scale in situ demonstration test for repository
sealing in an argillaceous host rock — Phase II. Final report.: EUR 24161, Nuclear Science
and Technology, European Commission, Luxembourg (288 pp.).
45
Villar, M.V., 2013. EB experiment. Contribution of CIEMAT to EB dismantling report. Phys-
ical state of the bentonite. EC Contract 249681 PEBS. Informe Técnico CIEMAT/DMA/
2G210/04/2013 (Madrid, 21 pp.).
Villar, M.V., Lloret, A., 2008. Influence of dry density and water content on the swelling of
a compacted bentonite. Appl. Clay Sci. 39, 38–49.
Villar, M.V., Garcia-Sineriz, J.L., Barcena, I., Lloret, A., 2005. State of the bentonite barrier
after five years operation of an in situ test simulating a high level radioactive waste
repository. Eng. Geol. 80, 175–198.
Villar, M.V., Martín, P.L., Bárcena, I., García-Siñeriz, J.L., Gómez-Espina, R., Lloret, A., 2012.
Long-term experimental evidences of saturation of compacted bentonite under re-
pository conditions. Eng. Geol. 149–150, 57–69.
Villar, M.V., Campos, R., Gutiérrez-Nebot, L., 2014. EB experiment. Laboratory post-
mortem analyses report. EC Contract 249681 PEBS. PEBS Deliverable 2.1–7. Technical
Report CIEMAT/DMA/2G210/01/2014 (Madrid, 34 pp.).
Volckaert, G., Bernier, F., Alonso, E., Gens, A., Samper, J., Villar, M.V., Martín, P.L., Cuevas, J.,
Campos, R., Thomas, H.R., Imbert, C., Zingarelli, V., 1996. Thermal–hydraulic–mechan-
ical and geochemical behaviour of the clay barrier in radioactive waste repositories
(model development and validation). Nuclear science and technology. EUR 16744.
Commission of the European Communities, Luxembourg (722 pp.).
Volckaert, G., Dereeper, B., Put, M., Ortiz, L., Gens, A., Vaunat, J., Villar, M.V., Martín, P.L.,
Imbert, C., Lassabatère, T., Mouche, E., Cany, F., 2000. A large-scale in situ demonstra-
tion test for repository sealing in an argillaceous host rock. Reseal project — Phase I.
EUR 19612. Commission of the European Communities, Luxembourg (273 pp.).