Long-term experimental evidences of saturation of compacted bentonite under repository conditions

Long-term experimental evidences of saturation of compacted bentonite under repository conditions

Engineering Geology 149–150 (2012) 57–69 Contents lists available at SciVerse ScienceDirect Engineering Geology journal homepage: www.elsevier.com/l...

2MB Sizes 0 Downloads 1 Views

Engineering Geology 149–150 (2012) 57–69

Contents lists available at SciVerse ScienceDirect

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

Long-term experimental evidences of saturation of compacted bentonite under repository conditions M.V. Villar a,⁎, P.L. Martín a, I. Bárcena b, J.L. García-Siñeriz b, R. Gómez-Espina a, A. Lloret c a b c

CIEMAT, Madrid, Spain AITEMIN, Madrid, Spain UPC, Barcelona, Spain

a r t i c l e

i n f o

Article history: Received 30 April 2012 Received in revised form 7 August 2012 Accepted 9 August 2012 Available online 29 August 2012 Keywords: Bentonite Hydration Engineered barrier Radioactive waste repository Temperature

a b s t r a c t This paper summarises the information gathered in the last 15 years on the saturation of compacted bentonite obtained from different laboratory-scale tests, a large-scale mock-up test, and a real-scale in situ test, that were performed to simulate the conditions of the bentonite barrier in a high-level radioactive waste repository and to better understand the hydration/heating processes. In all the tests the bentonite used was the Spanish FEBEX bentonite, the maximum temperature in the system was 100 °C and the water used was of the granitic type, with low salinity. Some of the tests were running for more than thirteen years. The migration of water vapour in areas affected by the high temperature induced by the radioactive waste decay is very rapid, its extent depending on the actual temperature and bentonite porosity. The water vapour condensates in cooler areas and this causes water content increases in internal zones of the barrier where the liquid water coming from the host rock has not yet arrived. The hydration kinetics is initially quicker when the temperature is high, provided no vapour phase is formed. Nevertheless, the major effect of the thermal gradient on saturation is a delaying of it in the inner parts of the barrier, which can be very persistent and depends on the actual thermal gradient and consequently, on the barrier thickness and boundary conditions. During the transient period in which the barrier is saturating, important changes in the water content and dry density of the bentonite are generated, which induce bentonite density and water content gradients along its thickness. These gradients could eventually disappear once the barrier is fully saturated, depending on the irreversibility of the deformations. The average density of the water in the saturated barrier will be higher than 1 g/cm3, due to the predominance of high-density, interlayer water in the compacted bentonite, and consequently, more water than expected, according to calculations made considering the density of free water, would fit in the bentonite pores. The rate of hydration of the barrier depends on the bentonite and surrounding media hydraulic properties (that is, water availability), waste temperature and buffer thickness and geometry. © 2012 Elsevier B.V. All rights reserved.

1. Introduction A current design for engineered barriers in high-level radioactive waste (HLW) repositories includes bentonite compacted blocks initially unsaturated. The heat released by the waste will induce a thermal gradient through the bentonite barrier while groundwater will tend to flow into it. Most models predict that full saturation of the barrier will be reached before the dissipation of the thermal gradient, which will take place between 100 and 1000 years after deposition, depending on the particular characteristics of the repository. However,

⁎ Corresponding author at: Avd. Complutense 40, 28040 Madrid, Spain. Tel.: +34 913466391; fax: +34 913466542. E-mail addresses: [email protected] (M.V. Villar), [email protected] (P.L. Martín), [email protected] (I. Bárcena), [email protected] (J.L. García-Siñeriz), [email protected] (R. Gómez-Espina), [email protected] (A. Lloret). 0013-7952/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.enggeo.2012.08.004

experimental and modelling results have shown that the barrier saturation can be an extremely slow process greatly affected by the bentonite microstructural modifications that take place upon hydration under constant volume conditions and by other processes such as thermoosmosis and the effect of low hydraulic gradients on hydraulic conductivity (Sánchez et al., 2007, 2012). Moreover, it still remains unclear whether the high temperatures around the canister would hinder the full saturation of the inner part of the barrier or just delay it. This paper summarises the information collected in the last 15 years on the saturation of compacted FEBEX bentonite obtained from different laboratory-scale tests, a large-scale mock-up test, and a real-scale in situ test, that were performed to simulate the conditions of the clay barrier in the repository under controlled boundary conditions. The results obtained help understand the hydration/ heating processes and their consequences on the long-term bentonite performance, what would allow the validation and verification of the near-field thermo-hydro-mechanical (THM) models. The large-scale

58

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

tests were proposed as a complementary step in the task of demonstrating the feasibility of installing a clay engineered barrier surrounding a simulated canister in a gallery excavated in granite, keeping with the ENRESA “Deep Geological Disposal for Granite” reference concept (ENRESA, 1995). In this concept the waste canisters are emplaced horizontally in drifts, surrounded by a barrier of high density compacted bentonite blocks. In the two large-scale tests the thermal effect of the waste is simulated by means of heaters; hydration is natural in the in situ test – which is being performed in a gallery excavated in granite – and controlled in the case of the mock-up. Both in the large-scale and in the laboratory tests, the temperature at the heater surface was fixed at 100 °C, which is the maximum temperature expected on the surface of the waste container in the Spanish concept. Also, in the mock-up and in most of the laboratory tests, the hydration water used was a low salinity granitic type. The material used in all these tests is the Spanish FEBEX bentonite, selected by ENRESA as suitable material for the backfilling and sealing of HLW repositories. To obtain the swelling pressure and hydraulic conductivity necessary for a good barrier performance it must be compacted to dry densities of about 1.6 g/cm 3 (Villar et al., 2006). 2. Material The FEBEX bentonite was extracted from the Cortijo de Archidona deposit (Almería, Spain) and the processing at the factory consisted on disaggregation and gently grinding, drying at 60 °C and sieving by 5 mm. The physico-chemical properties of the FEBEX bentonite, as well as its most relevant thermo-hydro-mechanical and geochemical characteristics obtained during the projects FEBEX I and II are summarised in the final reports of the project (ENRESA, 2000, 2006a), and a comprehensive study related to its hydro-mechanical and microstructural properties is given in Lloret et al. (2003). A summary of the results obtained is given below. The montmorillonite content of the FEBEX bentonite is above 90 wt.% (92 ± 3%). The smectitic phases are actually made up of a smectite–illite mixed layer, with 10–15 wt.% of illite layers. Besides, the bentonite contains variable quantities of quartz (2 ± 1 wt.%), plagioclase (3 ± 1 wt.%), K-feldspar (traces), calcite (1 ± 0.5 wt.%), and cristobalite–trydimite (2 ± 1 wt.%). The cation exchange capacity of the smectite is 102 ± 4 meq/100 g, the main exchangeable cations being calcium (35 ± 2 meq/100 g), magnesium (31 ± 3 meq/100 g) and sodium (27 ± 1 meq/100 g). The predominant soluble ions are chloride, sulphate, bicarbonate and sodium. The liquid limit of the bentonite is 102 ± 4%, the plastic limit 53 ± 3%, the density of the solid particles 2.70 ± 0.04 g/cm 3, and 67 ± 3% of particles are smaller than 2 μm. The hygroscopic water content in equilibrium with the laboratory atmosphere (relative humidity 50 ± 10%, temperature 21 ± 3 °C, total suction about 100 MPa) is 13.7 ± 1.3%. The external specific surface area is 32 ± 3 m 2/g and the total specific surface area is about 725 m 2/g. The saturated hydraulic conductivity of compacted bentonite samples is exponentially related to their dry density. For a dry density of 1.6 g/cm 3 the saturated permeability of the bentonite is about 5·10 −14 m/s at room temperature, either with granitic or deionised water used as percolating fluid. The temperature increase results in an increase in permeability. The swelling pressure of compacted samples is also exponentially related to the bentonite dry density, and when the bentonite at dry density of 1.6 g/cm 3 is saturated with deionised water at room temperature, the swelling pressure has a value of about 6 MPa. Saturation with granitic water gives similar values, whereas temperature causes a decrease of them. The retention curve of the bentonite was determined in samples compacted to different dry densities at different temperatures

(Lloret et al., 2004; Villar and Lloret, 2004; Villar and Gómez-Espina, 2009). The volume of the samples remained constant during the determinations, since they were confined in constant volume cells. Following an approach similar to that presented by Sánchez (2004) to fit the data from these laboratory determinations, the empirical Eq. (1) can be obtained: 0 2   c −α ðT−T 0 Þ w ¼ @ b⋅n ⋅e ⋅41 þ

!

1 1−λ1

s −ηðn−n0 Þ

P 0 ⋅e

⋅e

−α ðT−T 0 Þ

3−λ 1 1

5

  λ  2 A⋅ 1− s ⋅ðSr −Slr Þ P sec

ð1Þ where w is the water content in percentage, n and n0 the porosity and reference porosity, s the suction in MPa, T and T0 the temperature and reference temperature in °C, Sr and Slr the liquid degree of saturation and liquid residual degree of saturation, P0, Psec, λ1 and λ2 parameters to define the retention curve at reference temperature and porosity, and b, c, α and η fitting parameters to take into account the influence of temperature and porosity. The values of parameters are indicated in Table 1. The differences between measured values and the estimated values using Eq. (1) are smaller than 2% in terms of water content. The thermal conductivity (λ, W/m·K) of the compacted bentonite at laboratory temperature is related to the degree of saturation (Sr) through the following expression: λ¼

0:57−1:28 þ 1:28 ðSr −0:65Þ =0:100

ð2Þ

1 þ exp

Some isothermal infiltration tests and heat flow tests at constant overall water content were performed during the FEBEX I project (ENRESA, 2000, 2006a) and they were back analysed using CODEBRIGHT. The experimental data were fitted using a cubic law for the relative permeability and a value of 0.8 for the tortuosity factor. In all the tests described below, the bentonite was used compacted with its hygroscopic water content (14%) at dry densities between 1.6 and 1.7 g/cm3, which is the range expected in the repository. 3. Description of tests 3.1. In situ test The FEBEX in situ test is performed under natural conditions and at full scale in a gallery excavated in the underground laboratory managed by NAGRA at Grimsel, Switzerland (ENRESA, 2000, 2006a). The basic components of the test (Figure 1) were: the gallery, measuring 70 m in length and 2.3 m in diameter, excavated through the Aare granite; the heating system, made up of two heaters placed inside a liner installed concentrically with the gallery and separated one from the other by a distance of 1.0 m, with dimensions and weights analogous to those of the real canisters; the clay barrier, formed by blocks of compacted bentonite; the instrumentation and the monitoring and control system for data acquisition and supervision and control of the test both autonomously and remotely from Madrid. Up to 632 sensors of very diverse types were initially installed to monitor the different thermo-hydro-mechanical processes that occurred in both the clay barrier and the surrounding rock throughout the entire life of the test. The gallery was closed by a concrete plug.

Table 1 Values of parameters in Eq. (1). b

c

P0 (MPa)

λ1

λ2

η

n0

α (1/°C)

T0 (°C)

Psec (MPa)

Sr

Slr

145

1.9

25

0.2

1.1

20

0.4

0.0015

20

1000

1.0

0.01

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

59

Fig. 1. General layout of the in situ test during phase I, including instrumented sections (ENRESA, 2000).

To build the clay barrier, various types of blocks were manufactured from the bentonite in the shape of 12-cm thick circular crown sectors. The blocks were arranged in vertical slices with three concentric rings. In the heater areas the interior ring was in contact with the steel liner, whereas in the non-heater areas a core of bentonite blocks replaced the heaters. The thickness of the bentonite barrier in the heater areas was 65 cm. The blocks were obtained by uniaxial compaction of the FEBEX clay with its hygroscopic water content at pressures of between 40 and 45 MPa, what gave place to dry densities of 1.69–1.70 g/cm 3. The initial dry density of the blocks was selected by taking into account the probable volume of the construction gaps and the need to have a barrier with an average dry density of 1.60 g/cm 3 (ENRESA, 2000). The heating stage of the in situ test, known as operational stage, began on February 27th 1997. The power of the heaters was adjusted so that to keep the temperatures at their surfaces at 100 °C. After five years of uninterrupted heating at constant temperature, the heater closer to the gallery entrance was switched off (February 2002). In the following months the heater and all the bentonite and instruments preceding and surrounding it were extracted. The remaining part was sealed with a new sprayed concrete plug. New sensors were installed in the buffer, and a second operational phase started and is currently still running. 3.2. Mock-up test The mock-up test is an experiment at almost full scale carried out at the CIEMAT facilities (Madrid) under controlled boundary conditions. Its main components are (Figure 2): a stainless steel confining structure that simulates the gallery, through which hydration takes place; a Programmable Logic Control that manages the heater control system, composed of two electric heaters (0.17 m in diameter and

1.625 m long) concentric to the confining structure, that simulate the heat generation of the waste canisters; a hydration system that supplies the granitic water (Table 2) to hydrate the bentonite mass at a constant controlled pressure of 0.5 MPa; a 63-cm thick engineered barrier composed of compacted FEBEX bentonite blocks surrounding the heaters; instrumentation that monitor the boundary conditions and the system behaviour by sensors installed within and outside the buffer material; and a monitoring and control system that records the data by a Data Acquisition System. More than 500 sensors were installed to measure temperature, total pressure, pore pressure, water injection pressure, relative humidity and strains. These sensors were installed along the vertical sections indicated in Fig. 2. In particular, for the relative humidity measurements, the transmitters used were VAISALA HMP234 (capacitive-type sensors). The blocks were manufactured by compacting the bentonite with its hygroscopic water content at a dry density of 1.77 ± 0.20 g/cm 3, applying uniaxial pressures of 40–50 MPa. They were arranged in vertical sections: the sections around the heaters were formed by two concentric rings of blocks, and the other sections were formed by two concentric rings and a core of blocks. The total initial weight of the bentonite blocks inside the mock-up test was 22.5 t, with an average dry density value close to 1.65 g/cm 3, and a total volume of gaps of 6.6%. These gaps were initially flooded with water, which increased the average initial water content of the bentonite to 17.1%, since 634 L were injected in this preliminary phase. The operational stage – simultaneous hydration and heating – started in February 1997 (ENRESA, 2000, 2006a). The power supply of the heaters was automatically adjusted in order to supply a constant temperature of 100 °C at the heater surface. After 5 years of operation, the power supply was fixed at 700 W/heater, hereby maintaining the surface temperature of the heaters close to 100 °C. The temperature of the test room is 20±4 °C. The test is currently running.

60

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

N2 Pressure line

Water tanks weighting

Heaters power line

18 Temperature Temperature (328), pressure Temperature (20), injection sensors in the heaters (80), and relative humidity pressure (2) and (40) sensors in the displacement (19) sensors instrumentation seccions in the confining structure

Φ ext. 340 mm

6000mm

806 mm

A12 A11 A10 A9

A8

A7

A6

A5

A4

A3

A2

A1

AB

B1

B2

B3

B4

B5

B6

B7

B8

B9

B10 B11 B12

1259 mm

1000 mm 881 mm

1625 mm750 mm

EN0203B.CDR

Room temperature

Pressurized water line

Bentonite blocks

Instrumented sections

Confining structure

Heaters

186 Sensor nozzles

Geotextile layer

48 Hydration system nozzles

Fig. 2. General layout of the mock-up test, including instrumented sections.

3.3. Laboratory tests Conducting large-scale tests as the two described above is complicated and extremely costly. For this reason, laboratory tests of different scales are very useful to identify and quantify processes in shorter periods of time. These are performed in cylindrical cells in which the compacted bentonite is subjected simultaneously to heating and hydration, in opposite directions. Several of these thermo-hydraulic (TH) tests were carried out at CIEMAT laboratories, involving different dimensions and operation times, as described below. 3.3.1. 60-cm long cells A series of infiltration tests in large-scale cells (inner length 60 cm, internal diameter 7 cm) was running for different periods of time (Villar et al., 2008a). The cells were made of Teflon to prevent lateral heat conduction, and externally covered with steel semicylindrical pieces to reduce deformation of the cell by bentonite swelling. Six 10-cm height blocks of FEBEX clay compacted with its hygroscopic water content at an initial nominal dry density of 1.65 g/cm 3, were stacked inside each cell. An average compaction pressure of 30 MPa was applied to manufacture the blocks. The actual average initial water content of the bentonite in the seven tests performed was 13.6%, and the average initial dry density was 1.66 g/cm 3. The bottom part of the cells was a flat stainless steel heater set during the tests at a temperature of 100 °C. Over the upper lid of the cells there was a deposit in which water circulated at room temperature (20–30 °C). Hydration with a commercial granitic water (salinity 0.02%, Table 2) took place through the upper surface under an injection pressure of 1.2 MPa. This simulated the water that saturates the barrier in a repository excavated in granitic rock and was the same employed to saturate the mock-up test. The water intake was

measured as a function of time in some of the tests with an electronic equipment placed at the entrance of the cell. In addition, the cells were instrumented with thermocouples inserted in the bentonite at different levels along the column. Seven tests were performed: two of 0.5 year duration, two of 1 year duration, two of 2 years duration and one of 7.6 years duration. At the end of the tests the bentonite columns were extracted and cut into 24 horizontal sections. In each of these sections gravimetric water content and dry density, among many other parameters, were determined. 3.3.2. 40-cm long cells Two hydration tests are being performed in cylindrical cells similar to those described above, whose internal diameter is 7 cm and inner length 40 cm. The bentonite was compacted as for the 60-cm long cells: with its hygroscopic water content at an initial nominal dry density of 1.65 g/cm 3. The hydration procedure and kind of water was the same as for the 60-cm long cells. In one of the tests (GT40) the clay is heated at the bottom surface at a temperature of 100 °C. The other test (I40) is carried out at isothermal conditions (laboratory temperature). The cells are instrumented with capacitivetype sensors placed inside the clay at three different levels separated 10 cm. The transmitters used are similar to those used in the mock-up test (VAISALA HMP237). The water intake and the relative humidity and temperature evolution at different levels inside the clay are being measured as a function of time. The two cells are in operation since January 2002 (Villar and Gómez-Espina, 2009). 3.3.3. 8-cm long cell These tests were performed in a cylindrical cell designed so that to provide facing hydraulic and thermal gradients and isotherms parallel to the sample ends (Villar, 1995). The compacted sample was placed inside the cell, where it was heated at the bottom and saturated with

Table 2 Chemical composition of the granitic water injected in the mock-up and laboratory tests. pH

Cl− (mg/L)

SO42− (mg/L)

HCO3− (mg/L)

Mg (mg/L)

Ca (mg/L)

Na (mg/L)

K (mg/L)

8.3

13.1 ± 0.7

14.4 ± 0.1

144.0 ± 0.7

9.4 ± 0.7

44.8 ± 7.1

10.6 ± 0.7

1.0 ± 0.1

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

100

Relative humidity (%)

90 80

Non-heater area

70

C-04

Heater area F2-14

60 50

C-05

40

F2-11

30

E1-03

20

E1-04 Heater area

10 0 0

1000

2000

3000

4000

5000

Time (days) Fig. 3. Evolution of the relative humidity of the bentonite at 3.5 cm from the heater (sensors E1-03, E1-04, F2-11 and F2-14) and at approximately the same distance from the gallery wall (54 cm) in a section not affected by the heater (sensors C-04 and C-05) of the in situ test (see Figure 1 for sections location).

distilled water injected at a pressure of 1.1 MPa on top. The cell had an external body made of stainless steel and an internal 5-mm thick Teflon jacket to lessen heat transmission along the cell walls. The base of the cell was made of stainless steel and it was placed directly on a heating plate. The temperature of this heating plate was set so that to get a temperature of 100 °C at the bottom surface of the bentonite. The bentonite blocks were obtained by uniaxial compaction of the clay with hygroscopic water content to a dry density of 1.65 g/cm 3. The compaction was done in five layers to get a uniform density along the block, whose nominal height and diameter were 7.8 and 3.8 cm, respectively. The tests were dismantled after different periods of time, from 4 to 108 days. The samples were extracted and cut into five horizontal sections parallel to the sample ends. In each of these sections gravimetric water content and dry density were determined.

4. Results 4.1. In situ test The partial dismantling of the FEBEX in situ test was carried out after five years of continuous heating (Bárcena et al., 2003). A large number of samples were taken for analysis. The dismantling was carried out causing minimum disturbance in the section of the test corresponding to the second heater that was kept in operation at all times. Also the process of data acquisition was maintained during the dismantling. This allowed to follow the evolution of temperature and relative humidity (among other parameters) during cooling, and to know exactly which the conditions of the barrier were at the moment of dismantling. After five years of heating, and according to the sensors measurements, the bentonite blocks the closest to the heater had water contents below the initial ones, although they were recovering after the intense initial drying (Gens et al., 2009). On the contrary, for the same period of time, the sensors located at the same distance from the gallery wall, but in an area not affected by the thermal gradient, recorded much higher relative humidity (Figure 3).

61

The gravimetric water content and dry density of numerous bentonite samples taken during dismantling were determined on site and in the laboratory. This allowed checking the sharp clay water content and dry density gradients inside the barrier. In addition, the suctions computed from the sensors' recordings (temperature and relative humidity, converted to suction through the psychrometric law) just before dismantling were related to gravimetric water content using the water retention curves empirically determined for different bentonite dry densities and temperatures. The water contents thus determined agreed well with the values gravimetrically determined for the same positions after dismantling (Villar et al., 2005a; Villar et al., 2005b). The other half section continued running and after more than thirteen years of operation, the relative humidity (RH) values at the outer and intermediate rings were almost 100% everywhere (AITEMIN, 2009). These high values were reached approximately after 1500 days of operation in the cold areas (section H), and 2000 days in the hot areas (section F2). However, in the internal ring, RH stayed in values around 70% even in the cold sections (section H), and even lower in the hot sections (between 40 and 70% in section F2). A sensor placed in the core of the barrier in a cold section (H) recorded a RH of just 45% after 2300 days of operation (it failed afterwards). Similarly to the water content, the total pressure – which is an indication of swelling pressure – continued increasing, although at a slower rate, in all points inside the buffer, particularly at the contact bentonite/rock. After thirteen years of heating, the experiment temperatures seemed to be stabilised, and in the bottom part of the drift they were above those at the sides and upper part for each section (Table 3). 4.2. Mock-up test After more than thirteen years of test operation, the data showed homogeneity between the two halves of the experiment and between points located at similar radial distances (Martín, 2004; Martín and Barcala, 2005, 2011). Statistical tools demonstrated that the signals observed were representative of the THM behaviour of the system and that no artefacts modified the response of the sensors in the experiment (Cañamón et al., 2004). The thermal regime was homogeneous and symmetric, both with respect to the central section and the longitudinal axis. The fluctuations of temperature close to the structure were related to external variations. Differences among sensors located at the same radial distances were lower than 2 °C. The temperature gradient had its maximum value close to the heater surface (3 °C/cm). The water volume injected is measured on line. The current water intake is very low, with an average value for the last 7 years of 42 cm 3/day over a surface area of 30 m 2. Taking into account the dry mass and initial water content of the bentonite and the water quantity injected during the initial flooding, the evolution of the average water content of the barrier could be computed. After 12 years of operation, 1111 L of water had been injected, corresponding to an overall water content of 22.8% and a degree of saturation of 97% if the water density is taken as 1 g/cm 3. The saturation of the bentonite barrier was apparently controlled by the hydraulic properties of the bentonite and the thermal gradient

Table 3 Temperature (°C) in the bentonite barrier of the in situ test after thirteen years of operation (AITEMIN, 2009).

Outer ring Intermediate ring Inner ring

Section G (1 m from heater)

Section I (heater front)

Section S (1 m into heater zone)

Section F2 (middle of heater)

Section D2 (heater rear end)

Section B2 (gallery end)

29–33 33–37 35–40

35–37 50–58 76–81

– 68 90

– 73 94–100

35–36 51–53 80–88

22 22 22

62

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

100

Relative humidity (%)

imposed by heating. Both liquid water inflow and water vapour outflow followed mainly radial directions. The relative humidity values measured close to the hydration surface suggested that the bentonite was fully saturated in the external ring (30 cm thickness). The evolution of the relative humidity in the inner part of the barrier showed clearly the effects of evaporation (drying) and condensation (wetting) at different zones. Close to the heaters three phases could be distinguished according to the recordings of the inner sensors (Figure 4): (1) a vapour phase generated by heating (sharp RH increase), (2) drying of the clay as the vapour phase moved away to cooler zones (RH decrease), and (3) eventually hydration reached the dried zone and overcame the drying process (slow RH increase). The relative humidity recovery up to values close to the initial ones took about 2 years in the proximity to the heater. The RH evolution curves in all sections showed similar trends to those just described and exemplified in the figure. The only differences were related to the relative radial and longitudinal positions with respect to the heaters (hot and cold zones): in zones without heater (cold sections), the sensors located at 13 cm from the external surface recorded an increase in relative humidity slower and less intense than the sensors located a bit closer to the external surface (10 cm) in sections around the heater (Figure 5). Eventually, the cold zones showed higher RH values, with a RH above 90% at distances as far as 48 cm from the external surface. Under hydration the bentonite developed a swelling pressure recorded by the total pressure sensors located at different radial distances and orientated in the three main directions. The measured pressures were in accordance with swelling pressure values of the bentonite measured in the laboratory, between 10 and 6 MPa. The higher pressure values were located within the outer ring, where the average pressure values and their slow increase indicate full saturation.

90

70

V_A10_3 V_B10_3 V_A4_4

60

V_B4_4 V_B6_4

80

50 40 0

400

800

1200

1600

Time (days) Fig. 5. Initial evolution of RH at 13 cm from the external surface in cold sections (A10 and B10) and at 10 cm in hot sections (A4, B4 and B6) of the mock-up test.

the heater, decreasing from 100 °C at the heater surface to 50 °C at 10 cm from it. In addition, a slight decrease in temperature with saturation (longer tests) was observed, especially in the wettest zones, within the closest 40 cm to the hydration surface. This was due to the higher heat dissipation of the wet clay. Immediately after setting the heater temperature or after 24 h, depending on the particular test, hydration started. The final overall water content of the bentonite columns at the end of the tests (computed from their weight difference before and after the tests) could be related to their duration through an exponential law (Figure 6). The extrapolation of these values towards longer times is not straightforward, since in these tests there was a slight change over time of the overall dry density of the bentonite – due to its swelling and the deformability of Teflon –, and the changes in total porosity can affect the water intake. Nevertheless, the fitting drawn in the figure points to a final water content of about 28%, much higher than needed to reach full saturation if the expected dry densities and a value of water density of 1 g/cm 3 were considered. In addition, at the end of all the tests important water content and dry density gradients were measured along the bentonite columns. After 7.6 years of testing the water content of the bentonite was lower than the initial one in the 5 cm closest to the heater. The water content reduction by the temperature effect was similar for the three shorter durations, which means that desiccation took place rather quickly and affected only the closest 18 cm to the heater, where the water content was lower than the initial one even after 2 years. The final average degree of saturation in the longest test

4.3. Laboratory tests 4.3.1. 60-cm long cells During the operation of the TH tests in cells, the bentonite temperature was measured at different positions along the columns. The temperatures quickly stabilised (in less than 200 h) and afterwards, they were mainly affected by the room temperature changes (Villar et al., 2008a, b). The average temperatures recorded by each thermocouple were stable, and there was a good agreement between the average temperatures measured in the duplicate tests of the same duration. There was a sharp temperature gradient in the vicinity of

Relative humidity (%)

100 90 80 70 60

V_A4_1

50

V_A4_2

40

V_A4_3

30

V_A4_4 4550

4350

4150

3950

3750

3550

3350

3150

2950

2750

2550

2350

2150

1950

1750

1550

1350

950

1150

550

750

350

-50

150

20

Time (days)

Fig. 4. Evolution of the relative humidity of the bentonite in a hot section (A4) of the mock-up test (temporary power failures took place starting at days 1392, 1966 and 3146; sensor V_A4_2 failed at day 2975).

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

30

63

the hydration surface, due to the fact that their saturation and, consequently, their swelling, was higher, which was favoured by the deformation of the cells' Teflon walls.

28

Water content (%)

26 24

w = 28.28 – 2.34 . e

-t

7

-t

–12.21. e

2274

22 20 18 16 14 12 0.1

1

10

100

1000

10000

Time (days) Fig. 6. Water content at the end of the TH tests performed in 60-cm long bentonite columns (t: time, days; w: water content, %).

(7.6 years), computed considering a water density of 1 g/cm 3, was of 92%. The water content increase caused by hydration was linked to a reduction in dry density resulting from swelling, whereas the water content decrease caused by evaporation near the heater led to an increase in dry density resulting from shrinkage. The dry density changes along the bentonite columns at the end of the tests reflected the different swelling of the bentonite, since the more hydrated sections swelled more. This swelling gave place to the sealing of all the 6 blocks that initially constituted the column in the 7.5-year test and to the sealing of just the 2 upper blocks in the 0.5-year test. The dry density decreased from the heater towards the hydration surface following an approximately linear trend. Although the dispersion of the values measured was large, most of them were in the range between 1.6 and 1.7 g/cm 3 near the heater to around 1.4 g/cm 3 close to the hydration surface, the latter being well below the initial value (Villar et al., 2008a, b). The combination of water content and dry density changes along the bentonite columns of the different tests, gave place to the distribution of water degrees of saturation shown in Fig. 7. Due to the decrease in dry density associated to the swelling of the wet areas, the increase in degree of saturation in the upper part of the columns was not as high as could have been expected, and a significant change in degree of saturation along the columns was observed even for the longest test. Moreover, the overall densities were lower as the duration of the tests was longer, particularly near 105

Degree of saturation (%)

95 85 75 65 55

Initial

45 35

6 months

12 months

25

24 months

92 months

15 0

10

20

30

40

50

60

Distance from the heater (cm) Fig. 7. Degree of saturation along the bentonite column measured at the end of tests in 60-cm long cells.

4.3.2. 40-cm long cells Two tests similar to those just described but performed with 40-cm long bentonite columns have been running for 10 years. In one of them hydration takes place at room isothermal conditions, while in the other one it takes place under thermal gradient. In the case of the test performed under thermal gradient (GT40), the temperature was firstly set on top and bottom of the column. The temperature stabilisation phase in this test took three days. During this time, an increase in relative humidity was registered by the sensor placed at 10 cm from the heater (RH3), due to the quick migration of water in the vapour phase from the bentonite close to the heater towards cooler zones. To a lesser extent, the sensors placed at 20 (RH2) and 30 cm (RH1) from the heater also recorded a relative humidity increase due to the same process. Nevertheless, the RH values of the two lower sensors had not stabilised after the temperature equalisation, what indicates that water vapour migration continued to take place. After this initial heating, hydration started. The relative humidity evolution from the beginning of hydration in the test GT40 is shown in Fig. 8a. The values plotted are those measured during infiltration, which begun after stabilisation of the temperature registered by the sensors. For this reason, an initial difference in the relative humidity measured at different levels was observed. This trend reversed from the beginning of hydration, as shown by the clear increase in the relative humidity registered by the two upper sensors (RH1 and RH2) and by the desiccation that started to affect the zone in which sensor RH3 was placed. This means that at least the 10 cm closest to the heater reduced its relative humidity after 300 h of heating. This relative humidity decrease went down to values around 35%, which increased only to values of 42% after 10 years of hydration. Sensor RH1 recorded values around 93–94% since approximately 5 years after the start of operation and occasionally showed signs of liquid water condensation inside (relative humidity values above 100%). With respect to the temperatures inside the clay, they remained constant since the beginning of the experiment, being just influenced by the distance to the heater and by the seasonal and daily changes in the laboratory temperature. The thermal gradient was not constant along the column, being steeper near the heater. In the case of the isothermal test (I40), the cooling system on top was first set and the data acquisition began. After 18 h, the hydration system was connected. The evolution of relative humidity in the test performed under isothermal conditions is shown in Fig. 8b. The sensor placed at 10 cm from the hydration surface (RH1) showed a steady increase in relative humidity, which was noticeable after 250 h of hydration. The sensor placed at 20 cm from the hydration surface (RH2) started to register an increase in relative humidity after 1200 h of hydration, and the sensor placed towards the bottom (RH3), after 2500 h of hydration. After 5000 h of hydration, the relative humidity in the upper parts of the bentonite column (sensors RH1 and RH2) was increasing in a soft way, probably because the relative humidity in these hydrated zones was already very high and consequently the suction of the bentonite was low and could not be considered a significant driving force. Meanwhile sensor RH3 was recording the sharpest increase. In turn, the high water content of the upper part of the bentonite provided enough water supply to the bottom. After 10 years of operation, the relative humidity values recorded by the three sensors remained quite constant, although if their evolution is closely looked at, the relative humidities recorded by sensors RH2 and RH3 did not stop increasing, while sensor RH1 seemed to have stabilised. The average relative humidity along the bentonite column after 10 years of testing was higher for the sample tested at room

64

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

a

b

100

100

RH1

90 80

RH2

70 60 50

Relative humidity (%)

Relative humidity (%)

RH1

RH3

40

90 RH2

80

RH3

70 60 50 40 30

30 10

100

1000

10000

10

100000

100

1000

10000

100000

Time (hours)

Time (hours)

Fig. 8. Evolution of relative humidity in the tests performed under thermal gradient (a) and isothermal conditions (b) during infiltration of 40-cm long bentonite columns (sensor RH1 placed at 30 cm from the bottom, sensor RH2 at 20 cm and sensor RH3 at 10 cm).

temperature, because the hot zones of the sample tested under thermal gradient remained desiccated for long time. 4.3.3. 8-cm long cell This small-scale cell allowed the performance of several short tests in which a high final average water content was reached. The average water content of the bentonite increased for the first approximately 30 days and then remained quasi-stable. Fig. 9 shows the distribution of water content and dry density inside the bentonite blocks at the end of several tests. In most of the tests there was a water content gradient, with higher water contents towards the hydration surface and lower water contents towards the heater, where the shortest tests displayed a water content lower than the initial. This was caused by the migration of water in the vapour phase. This water content gradient attenuated as the duration of the tests was longer, and thus, the tests that lasted more than 50 days displayed a uniform water content distribution along the bentonite block, which means that water advection (Darcy's flow) eventually overcame the vapour convection (due to temperature gradients). The dry density of the bentonite was also affected by the hydration process. Thus, due to swelling the sections closest to the hydration surface had a dry density lower than the initial one, whereas the sections closest to the heater had a density higher than initial, which was a consequence of the shrinkage caused by the water loss. This dry density gradient attenuated as the water content homogenised. The overall final degrees of saturation in this series of tests were relatively low, probably due to the higher average temperature inside

the bentonite caused by the small size of the cell (despite the fact that the heater temperature was also 100 °C), which made the vapour content in the gas phase higher than in other series of tests. 5. Discussion The comparison between the mock-up and the in situ experiment is straightforward in the sense that the geometry and the barrier thickness were approximately the same in both cases (the barrier thickness around the heaters was 65 cm in the in situ test and 63 cm in the mock-up). However, the different boundary conditions resulted in clear differences, as exemplified by the average temperatures inside the barrier, which were higher in the in situ test (Figure 10). There is a geometrical reason for this difference, since the ratio between the external radius (heater + bentonite barrier radii) and the heater radius – which conditions the heat transport – is 4.76 in the case of the mock-up and 2.47 in the case of the in situ test. An additional factor for this difference could be the isolating effect of the granite around the barrier in the in situ test, through which heat is transmitted by conduction, whereas in the mock-up the heat loss by radiation through the metallic confining structure is high. As a consequence, when comparing the evolution of the relative humidity near the heater in the in situ and the mock-up tests (Figure 11), it becomes clear that the desiccation experienced by the bentonite in the in situ test was much more intense, with values of relative humidity as low as 10% being reached, whereas the lowest values recorded in the mock-up were of 30%. The lowest initial dry

25

1.80

23 1.75

Dry density (g/cm3)

Water content (%)

21 19 17 15 13 11 9

4 days

26 days

7

34 days

108 days

1.70 1.65 1.60

5 days 41 days

1.55

5

26 days 108 days

1.50 0

2

4

6

8

Distance from hydration surface (cm)

0

2

4

6

8

Distance from hydration surface (cm)

Fig. 9. Distribution of water content and dry density along the bentonite blocks in the 8-cm long cell tests (the horizontal bars indicate the average initial value).

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

100

100

Relative humidity (%)

mock-up

90

in situ 80

Temperature (°C)

65

40-cm cell

70 60 50

90 80 70 60 50 40 30 20 0

1000

2000

40 E1_2

30

4000

5000

E1_5

E1_6

E1_9

A6_2

A7_2

Fig. 12. Relative humidity measured by sensors located at 38 cm from the heater in the mock-up (sections A6 and A7) and at 33 cm from the heater in the in situ test (section E1).

20 0.0

0.2

0.4

0.6

0.8

1.0

Normalised distance to external surface Fig. 10. Temperatures measured along radii around the heater in the mock-up and the in situ tests and along the bentonite column of a 40-cm long cell test after 5 years operation.

density of the blocks of the in situ test (1.70 g/cm 3 vs. 1.77 g/cm 3 in the mock-up) could have also contributed to the easiest outwards movement of the vapour phase. On the other hand, the comparison of the relative humidity in the middle part of the barrier for the two tests shows less discrepancy, at least in the long term (Figure 12). In those sections not affected by the heater, the temperatures inside the mock-up after 5 years of operation were about 3 °C higher than in the in situ test, what is probably due to the higher external temperatures of the experiment room of the mock-up than of the granite host rock of the in situ test. Overall, the temperature distribution is controlled by thermal conduction, due to the slow-rate transport processes involved in the saturation of the buffer material. Both the two large-scale tests and the infiltration test GT40 showed that the permeability to water vapour of the dry bentonite is very high, since a quick redistribution of water took place when the heaters were switched on and the thermal gradient was established. On the other hand, the initial hydration of compacted bentonite took place quicker under thermal gradient than at laboratory temperature. This was clearly observed when comparing the hot and cold sections of the mock-up (Figure 5). The comparison of the results of tests GT40 and I40 points to the same conclusion (Figure 8): the initial increase in humidity registered by the upper sensor (RH1) was higher in test GT40 than in test I40, whereas the sensor placed in the middle of the column (RH2) started to perceive the humidity increase much earlier 100

Relative humidity (%)

3000

Time (days)

80 60 40 20 0 0

1000

2000

3000

4000

5000

Time (days) E1_3

E1_4

A4_1

A6_1

A7_1

F2_11

F2_14

Fig. 11. Relative humidity measured by sensors located at 5 cm from the heater in the mock-up (sections A4, A6 and A7) and the in situ test (section E1 and F2).

in the case of infiltration under thermal gradient (test GT40) than in the case of infiltration at laboratory temperature (test I40). In the case of test GT40, the humidity initially recorded by sensor RH2 could also come in the form of water vapour from the lower part of the column. Otherwise, the increase in hydraulic conductivity with temperature would account for this initial quicker hydration of the test under thermal gradient. However this behaviour was reversed as saturation continued and eventually, both in the mock-up and in the 40-cm long cells, the water intake was higher for the bentonite hydrated at lower temperature, because the hot zones of the sample hydrated under thermal gradient remained desiccated for long time. The analysis of the bentonite water content in the variety of tests reported in this paper can be performed in two ways. For those already completed and dismantled, the actual gravimetric water contents were determined by oven drying (partial dismantling of in situ test, tests in 60-cm and 8-cm long cells). For those not yet finished and instrumented with capacitive sensors (large-scale tests and tests in 40-cm long cells), the relative humidity and temperature values measured can be converted to suction through the psychrometric law, and related to water content through the empirical Eq. (1), that relates suction and water content for the FEBEX bentonite at different dry densities and temperatures, taking into account the suction computed, the temperature measured, and an estimated porosity. This procedure was successfully applied upon dismantling of the first part of the in situ test at Grimsel, where the water content values estimated from the sensor's readings following this procedure matched very well the values actually measured by oven drying (Villar et al., 2005a; ENRESA, 2006b). The water contents thus inferred for three different times of the mock-up evolution are shown in Fig. 13. To estimate the porosity in different zones of the mock-up barrier, the results obtained upon the partial dismantling of the in situ test and the observations made in tests in which the bentonite was simultaneously heated and saturated (tests in cells) were considered. Thus, a range of dry densities from 1.55 g/cm 3 at 10 cm from the external surface to 1.73 g/cm3 at 5 cm from the heater was estimated for the 2-year operation time, and between 1.53 and 1.74 g/cm 3 at the same positions for the 5- and 10-year operation times. A comparison between the values for the mock-up and the in situ tests around the heater after 5 years of operation gives the results plotted in Fig. 14, in which the water contents inferred for test GT40 after 5 years of operation have also been included. In agreement with the lower relative humidities and higher temperatures measured in the in situ test, the water contents estimated for it – particularly near the heater – are lower than for the mock-up. On the contrary, the water contents in the middle part of the barrier are similar for both tests, what could be explained by the higher proportion of water vapour coming from the internal zones and condensating in the middle part of the barrier in the in situ test.

66

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

30

Estimated water content (%)

28

1999: 2 years

26

2002: 5 years 24

2010: 13 years

22 20 18 16 14 12 10 0

10

20

30

40

50

60

Distance to external surface (cm) Fig. 13. Water contents inferred for the bentonite barrier of the mock-up test in vertical sections around the heaters at different operation times.

Also, in sections not affected by the heaters (cold sections), the estimated water contents in the mock-up are higher than in the in situ test, despite the higher average density of the mock-up barrier. The same geometrical reasons invoked for the temperature differences between both tests could partly account for the higher hydration of the mock-up test. The highest water availability in the mock-up could also contribute to the higher degree of hydration for similar experiment run times, although the modelling results for the in situ test showed that the bentonite permeability would condition the barrier saturation, irrespective of the host rock properties (ENRESA, 1998; Gens et al., 2002). Despite the fact that the temperatures inside the bentonite in the GT40 test were much lower than in the large-scale tests (Figure 10), the water contents after 5 years of operation were not very different, what could be partly explained by the different geometry of both kinds of tests: water flow is radial in the mock-up and in situ tests whereas it is parallel in cells. Accordingly, the ratio between the hydration surface and the total volume of the bentonite barrier is lower in the laboratory column tests. The laboratory tests in 40-cm long cells provided also information about the effect of the thermal gradient on the hydration kinetics. Following the same approach described to estimate the bentonite water content in the mock-up from the capacitive sensors recordings, the

values plotted in Fig. 15 were obtained for the 40-cm long cells after 10 years of operation. In the bentonite column hydrated under thermal gradient, the 10 cm closest to the heater would still be drier than at the beginning of the test. Indeed, under isothermal conditions the degree of hydration for the same run time was higher. Furthermore, in the thermo-hydraulic test performed in the 40-cm cell (GT40) the bentonite was subjected to a higher thermal gradient than in the 60-cm long cells (Figure 16). The water contents measured upon dismantling in the 60-cm cells are plotted in Fig. 17 along with the water contents estimated from the sensors measurements and the water retention curve (Eq. (1)) in the 40-cm long cell which is still running. The water contents thus estimated for an experiment run time of 97 months are lower than those measured after 92 months in the 60-cm long cell, what highlights the influence of the thermal gradient on the hydration kinetics. Both the laboratory tests in large cells and the large-scale tests did not run for long enough for full saturation to be reached, not even the 40-cm long column hydrated at room temperature for more than 10 years. However, in the small-scale tests the water content of the bentonite became homogeneously high even under high thermal gradient. This points to the thickness of the barrier being decisive to determine the time needed for full saturation, which is probably a consequence of its very low permeability. In this context, the importance of the tests in small cells reported here is that, due to the relatively small size of the bentonite block (7.8 cm long), it was possible to attain a high overall water content (for test durations around 30–40 days) and keep the tests going on for long time afterwards (up to 108 days). In this way it was possible to observe that the water content gradient that developed initially by the water advective movement (driven by suction gradients) was eventually levelled, once the degree of saturation was high enough (Figure 9). Water diffusion (due to vapour concentration differences) would be the main mechanism involved in this homogenisation process. Fig. 18 summarises the results in terms of bentonite average water content as a function of time for tests in which hydration took place under thermal gradient. For the mock-up the average water content was computed taking into account the initial mass of bentonite, its initial water content and the water injection measurements. The high initial water content was due to the flooding of all the barrier gaps at the beginning of the test. For the 60-cm and 8-cm long cells the water contents were gravimetrically determined at the end of the tests. For test GT40, the water contents were computed averaging the values shown in Fig. 14 (5 years), Fig. 15 (10 years) and Fig. 17

30 28

mock-up 28

Estimated water content (%)

Estimated water content (%)

33

in situ 40-cm cell

23

18

13

26 24 22 20 18

thermal gradient

16 14

isothermal 20°C

12 8 0.0

0.2

0.4

0.6

0.8

1.0

Normalised distance from external surface

10 0

10

20

30

40

Distance from the heater (cm) Fig. 14. Water contents around the heaters estimated from the in-place suction measurements and the water retention curve for the mock-up and the in situ tests and along the bentonite column of a 40-cm long cell after 5 years operation.

Fig. 15. Water content estimated from the sensors recordings in the 40-cm long cells after 10 years of operation.

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

27

60

60 cm, 6 months 60 cm, 12 months

55

mock-up 60-cm cells 40-cm cell 8-cm cell

25

60 cm, 24 months 40 cm, 97 months

50 45

Water content (%)

Temperature (°C)

67

40 35 30

23

21

19

25 17

20 0

0.2

0.4

0.6

0.8

1

Normalised distance from the heater 15 Fig. 16. Steady temperatures measured along the 60-cm and 40-cm long cells.

1

10

100

1000

10000

Time (days) (8 years). The differences found in the bentonite water content evolution for different tests could be explained by several factors, which would have a determining influence on the hydration rate:

Fig. 18. Average bentonite water content computed from the water intake in the mock-up test, measured at the end of tests in 60- and 8-cm long cells, and as calculated from Figs. 14, 15 and 17 for the GT40 long cell.

• The different geometry of the tests, which was bidimensional and axisymmetrical in the mock-up test, whereas it was unidimensional in the cells. Besides, in the mock-up there were “hot” and “cold” zones and the water content computed was an average of both, while in the cells hydration took place under a uniform thermal gradient. • There were also significant differences in dry density, which was higher in the mock-up barrier (average of 1.65 g/cm 3), whereas in the cells decreased over time due to the deformation of the Teflon cell, the final average dry density in the longest test in the 60-cm long cells being 1.57 g/cm 3. Thus, the permeability of the bentonite, which decreases exponentially with dry density, would be a major factor conditioning the hydration rate. • The temperature inside the bentonite, which was higher in the mock-up than in the tests in cells, and in test GT40 than in the 60-cm long cells, despite the fact that the temperature of the heater

was the same in all cases. This was due to the different geometry and boundary conditions of each test.

40 cm, 97 months

40

60 cm, 92 months

Water content (%)

35

60 cm, 24 months 60 cm, 12 months

30

60 cm, 6 months

25 20 15

Initial

10 5 0

0.2

0.4

0.6

0.8

1

Normalised distance from the heater Fig. 17. Water content along the bentonite column measured at the end of tests in 60-cm long cells and estimated for a 40-cm long cell instrumented with capacitive sensors (bentonite initially compacted at 1.65 g/cm3).

According to the water intake measurements in the mock-up, the degree of saturation after 12 years of operation would be of 97% (if the water density is taken as 1 g/cm 3), what can be considered quite high. Nevertheless, a high water content gradient still persists across the barrier (Figure 13), which would point to advection still being a water movement mechanism and to full saturation being far from being attained. Indeed, there is increasing evidence from the fields of neutron diffraction, Monte Carlo computer simulations and quasi-elastic neutron scattering that the density of water attached to clay minerals may be greater than 1.0 g/cm 3, with values of water density in filosilicates of up to 1.38 g/cm 3, higher in smectites with divalent cations in the interlayer (such as FEBEX) than with monovalent ones (Jacinto et al., 2012). This fact becomes especially evident in highly compacted expansive clays close to water saturation, in which degrees of saturation much higher than 100% can be computed if a water density value of 1.0 g/cm 3 is considered (Villar, 2002; Marcial, 2003; Lloret and Villar, 2007). In fact, if it is taken into account that values of about 1.2 g/cm 3 were given for the average water density in the FEBEX bentonite compacted to dry density of 1.65 g/cm 3 (Villar, 2002), the actual overall degree of saturation of the mock-up barrier after 12 years of operation would be lower, around 81%. Additionally, the simulation of the mock-up experiment performed by Sánchez et al. (2012) showed that, in order to obtain a more satisfactory reproduction of the long-term experimental results, it was necessary to use an enhanced double-structure model that explicitly considered the two dominant pore levels that actually exist in the FEBEX bentonite and it was able to account for the evolution of the material fabric. Furthermore, in a large-scale in situ experiment that examined isothermal water inflow from the surrounding granitic rock into highly compacted, unsaturated buffer material, Dixon et al. (2002) observed that, after 6.5 years operation, the water uptake had been much lower than initially expected. The simulation of this experiment performed taking into account that the expansion of the microstructure of the bentonite, as the material saturated, would

68

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69

tend to reduce the void spaces in the macrostructure and consequently, was likely to reduce the material's hydraulic conductivity, allowed modelling with much greater accuracy both the pattern and the rate of water uptake (Thomas et al., 2003).

6. Conclusions Experimental observations concerning the thermo-hydraulic behaviour of a bentonite barrier obtained in large-scale and laboratory tests performed to simulate the conditions of a HLW repository were reported. In all the tests the bentonite used was the Spanish FEBEX bentonite, the maximum temperature in the system was 100 °C and the water used was of the granitic type, with low salinity. Some of the tests were running for more than thirteen years. The following conclusions can be drawn from the analysis of these results: The migration of water in the vapour phase in areas affected by high temperature was very quick. Indeed the desiccation caused thereby was stronger the higher the temperature. The porosity of the barrier had probably an effect on the vapour migration, which was quicker the higher the porosity (that is, the lower the bentonite dry density). Water vapour condensated in cooler areas and this caused water content increases in internal zones of the barrier where the liquid water coming from the host rock had not yet arrived. The hydration kinetics was initially quicker in the areas where the bentonite was affected by the heater temperature, provided the temperature in the bentonite was not as high as to form a significant vapour phase. This was due to the increase of permeability with temperature. Thus, the external rings of the barrier around the canisters, where the temperature was below 50 °C, became saturated earlier and more intensely than the external part of the barrier away from the canisters and not affected by temperature. However, the major effect of the thermal gradient on saturation was its delay in the inner parts of the barrier, which could be very persistent and depended on the actual thermal gradient and consequently on the barrier thickness and boundary conditions. In addition, the thickness of the barrier conditioned critically the time needed for full saturation, what was probably a consequence of its very low permeability. During the transient period in which the barrier was saturating, important water content and dry density gradients were generated in the bentonite. The tests in small cells showed that these gradients could eventually disappear once the barrier was close to saturation, at least in terms of water content. Due to the high dry density of the barrier, once it becomes saturated the average density of the water in it will be higher than 1 g/cm 3, since the density of water adsorbed in smectites can be much higher, and, as the bentonite becomes saturated under confined conditions, the adsorbed water predominates over the free water, due to the swelling of the clay particles and the changes in the bentonite microstructure. In this way, if the calculations of the water degree of saturation are made considering the density of adsorbed water instead of that of free water, more water would fit in the bentonite pores. Consequently, for predicting the long-term behaviour of the barrier, the numerical models must take into account the effect of the change in bentonite microstructure upon hydration on water properties and bentonite permeability, otherwise the time needed for full saturation would be underestimated. Furthermore, the water movement, both in liquid and vapour phase, results in geochemical modifications, particularly of the pore water composition (Villar et al., 2008a; Fernández and Villar, 2010), that can also influence the barrier behaviour. Summarising, the rate of hydration of the barrier depends on the bentonite and surrounding media hydraulic properties (which condition water availability), waste temperature and buffer thickness and geometry. The experimental evidences presented do not allow to ascertain if full saturation could be reached for any combination of

barrier thickness, thermal gradient and material permeability (among other parameters).

Acknowledgements This work was financed by ENRESA (Spanish Agency for Nuclear Waste Management) and the European Commission through contracts “Modelling and validation of the thermal–hydraulic–mechanical and geochemical behaviour of the clay barrier” (EC Contract FI2W-CT910102), the FEBEX Projects (EC Contracts FI4W-CT95-006 and FIKWCT-2000-00016), the NF-PRO Project (EC Contract FI6W-CT2003-02389) and the PEBS Project (European Atomic Energy Community’s Seventh Framework Programme (FP7/2007-2011) Grant Agreement no249681). The authors also acknowledge the financial support of the Spanish “Ministerio de Ciencia e Innovación” through the research project BIA2008-06614 (BIGRISK). References AITEMIN, 2009. Sensors Data Report (In situ Experiment). FEBEXe Project Full Report n.2. (Madrid, 120 pp.). Bárcena, I., Fuentes-Cantillana, J.L., García-Siñeriz, J.L., 2003. Dismantling of the heater 1 at the FEBEX “in situ” test. Description of operations. Publicación Técnica ENRESA 09/2003, Madrid (134 pp.). Cañamón, I., Elorza, F.J., Mangin, A., Martín, P.L., Rodríguez, R., 2004. Wavelets and statistical techniques for data analysis in a mock-up high-level waste storage experiment. International Journal of Wavelets, Multiresolution and Information Processing 2 (4), 351–370. Dixon, D., Chandler, N., Graham, J., Gray, M.N., 2002. Two large-scale sealing tests conducted at atomic energy of Canada's underground research laboratory: the buffer-container experiment and the isothermal test. Canadian Geotechnical Journal 39, 503–518. ENRESA, 1995. Almacenamiento geológico profundo de residuos radiactivos de alta actividad (AGP). Diseños conceptuales genéricos. Publicación Técnica ENRESA 11/95, Madrid (105 pp.). ENRESA, 1998. FEBEX. Preoperational thermo-hydro-mechanical (THM) modelling of the “in situ” test. Publicación Técnica ENRESA 09/98, Madrid (99 pp.). ENRESA, 2000. FEBEX Project. Full-scale engineered barriers experiment for a deep geological repository for high level radioactive waste in crystalline host rock. Final Report. Publicación Técnica ENRESA 1/2000, Madrid (354 pp.). ENRESA, 2006a. FEBEX Full-scale Engineered Barriers Experiment, Updated Final Report 1994–2004. Publicación Técnica ENRESA 05-0/2006, Madrid (590 pp.). ENRESA, 2006b. FEBEX Project Final report. Post-mortem bentonite analysis. Publicación Técnica ENRESA 05-1/2006, Madrid (183 pp.). Fernández, A.M., Villar, M.V., 2010. Geochemical behaviour of a bentonite barrier in the laboratory after up to 8 years of heating and hydration. Applied Geochemsitry 25, 809–824. Gens, A., Guimaraes, L. Do N., Garcia-Molina, A., Alonso, E.E., 2002. Factors controlling rock–clay buffer interaction in a radioactive waste repository. Engineering Geology 64 (2–3), 297–308. Gens, A., Sánchez, M., Guimaraes, L. Do N., Alonso, E.E., Lloret, A., Olivella, S., Villar, M.V., Huertas, F., 2009. A full-scale in situ heating test for high-level nuclear waste disposal: observations, analysis and interpretation. Geotechnique 59 (4), 377–399. Jacinto, A., Villar, M.V., Ledesma, A., 2012. Influence of water density on the waterretention curve of expansive clays. Geotechnique 62 (8), 657–667. Lloret, A., Villar, M.V., 2007. Advances on the knowledge of the thermo-hydromechanical behaviour of heavily compacted FEBEX bentonite. Physics and Chemistry of the Earth Parts A/B/C 32 (8–14), 701–715. Lloret, A., Villar, M.V., Sánchez, M., Gens, A., Pintado, X., Alonso, E.E., 2003. Mechanical behaviour of heavily compacted bentonite under high suction changes. Geotechnique 53 (1), 27–40. Lloret, A., Romero, E., Villar, M.V., 2004. FEBEX II Project Final report on thermo-hydromechanical laboratory tests. Publicación Técnica ENRESA 10/04, Madrid (180 pp.). Marcial, D., 2003. Comportement hydromécanique et microstructural des matériaux de barrière ouvragée. Ph.D. Thesis. École Nationale des Ponts et Chausées, Paris, 316 pp. Martín, P.L., 2004. Desarrollo y diseño de ensayos termo-hidro-mecánicos integrados para la validación del concepto español de almacenamiento de residuos nucleares de alta actividad. PhD Thesis. Universidad Politécnica de Madrid, Madrid, 408 pp. Martín, P.L., Barcala, J.M., 2005. Large scale buffer material test: mock-up test at CIEMAT. Engineering Geology 81 (3), 298–316. Martín, P.L., Barcala, J.M., 2011. PEBS Report D2.2-4: FEBEX mock-up Annual Report. CIEMAT, Madrid (120 pp.). Sánchez, M., 2004. Thermo-hydro-mechanical coupled analysis in low permeability media. PhD Thesis. Universitat Politècnica de Catalunya (UPC), Barcelona, 281 pp. Sánchez, M., Villar, M.V., Lloret, A., Gens, A., 2007. Analysis of the expansive clay hydration under low hydraulic gradient. Experimental Unsaturated Soil Mechanics. Springer Proceedings in Physics, Vol. 112. Springer, Berlin, pp. 309–318.

M.V. Villar et al. / Engineering Geology 149–150 (2012) 57–69 Sánchez, M., Gens, A., Olivella, S., 2012. THM analysis of a large-scale heating test incorporating material fabric changes. International Journal of Numerical and Analytical Methods in Geomechanics 36, 391–421. Thomas, H.R., Cleall, P.J., Chandler, N., Dixon, D., Mitchell, H.P., 2003. Water infiltration into a large-scale in-situ experiment in an underground research laboratory. Geotechnique 53, 207–224. Villar, M.V., 1995. Thermo-hydro-mechanical characterization of the Spanish reference clay material for engineered barrier for granite and clay HLW repository. Laboratory and Small Mock-up Testing. Publicación Técnica ENRESA 03/95, Madrid (251 pp.). Villar, M.V., 2002. Thermo-hydro-mechanical characterisation of a bentonite from Cabo de Gata. A Study Applied to the Use of Bentonite as Sealing Material in High Level Radioactive Waste Repositories. Publicación Técnica ENRESA 01/2002, Madrid (258 pp.). Villar, M.V., Gómez-Espina, R., 2009. Report on thermo-hydro-mechanical laboratory tests performed by CIEMAT on FEBEX bentonite 2004–2008. Informes Técnicos CIEMAT 1178, Madrid (67 pp.). Villar, M.V., Lloret, A., 2004. Influence of temperature on the hydro-mechanical behaviour of a compacted bentonite. Applied Clay Science 26 (1–4), 337–350.

69

Villar, M.V., García-Siñeriz, J.L., Bárcena, I., Lloret, A., 2005a. State of the bentonite barrier after five years operation of an in situ test simulating a high level radioactive waste repository. Engineering Geology 80 (3–4), 175–198. Villar, M.V., Martín, P.L., Barcala, J.M., 2005b. Modification of physical, mechanical and hydraulic properties of bentonite by thermo-hydraulic gradients. Engineering Geology 81, 284–297. Villar, M.V., Pérez del Villar, L., Martín, P.L., Pelayo, M., Fernández, A.M., Garralón, A., Cuevas, J., Leguey, S., Caballero, E., Huertas, F.J., Jiménez de Cisneros, C., Linares, J., Reyes, E., Delgado, A., Fernández-Soler, J.M., Astudillo, J., 2006. The study of Spanish clays for their use as sealing materials in nuclear waste repositories: 20 years of progress. Journal of Iberian Geology 32 (1), 17–36. Villar, M.V., Fernández, A.M., Martín, P.L., Barcala, J.M., Gómez-Espina, R., Rivas, P., 2008a. Effect of Heating/Hydration on Compacted Bentonite: Tests in 60-cm Long Cells. Colección Documentos CIEMAT, Madrid . (72 pp.). Villar, M.V., Sánchez, M., Gens, A., 2008b. Behaviour of a bentonite barrier in the laboratory: experimental results up to 8 years and numerical simulation. Physics and Chemistry of the Earth 33, S476–S485.