In the late 1980s, ONDRAF/NIRAS, the Belgian Waste Management Agency, started the PRACLAY programme. The PRACLAY (Preliminary demonstration test for clay disposal of high-level radioactive waste) project aims at demonstrating the feasibility of the Belgian concept for the disposal of high-level radioactive waste (see figure 1 hereafter). An important item of the disposal concept is the backfilling of the disposal galleries. A material must be selected to fill in the void between the disposal tube and the lining of the gallery. The safety principle of the disposal concept was based on the properties of the clay layer. The main function of the backfilling is consequently not to enhance the performance of the host rock but to minimize any perturbation on the host rock.

This backfill material has therefore three functions:

These basic functions are completed by requirements dealing with handling and emplacement of this material. Considering the horizontal configuration of the disposal galleries, precompacted blocks were considered to be the most appropriate form. For the PRACLAY project, it was planned to install and operate a dummy disposal gallery in the Boom clay similar, as far as possible, to the real ones. Several technical aspects of this in-situ testing being not yet worked out in detail, ONDRAF/NIRAS decided in the early 1990’s to first design and construct a large-scale surface mock-up called OPHELIE.
The mock-up is performed by the EIG EURIDICE (Economic Interest Grouping between NIRAS/ONDRAF and SCK•CEN), in collaboration with his members and CEA (the French Atomic Energy Commission)It must be pointed out that this reference concept is currently under review due to open questions arising mainly from preliminary results of the mock-up OPHELIE and the preparation of the PRACLAY experiment.

The general objective of the mock-up test was to prepare the PRACLAY experiment. More specifically, the objectives were :

Aside from these technical and scientific considerations, the mock-up is also featured in the permanent exhibition of EIG EURIDICE on the HLW disposal. This provides a tool to communicate directly with both the scientific community and the general public about the research work under way.

The main preliminary studies concerned the hydration system design and the selection of the backfill material.
Two configurations were studied :

After selection of an hydration system only at the periphery (emplacement of the backfill blocks easier), the backfill material had to be optimised to fulfil the following requirements:

The composition of the backfill material resulted in a mixture (called M2) of 60 mass% FoCa-clay (swelling properties and low permeability) , 35 mass% sand (mechanical stability of the blocks and reduction of the swelling ability) and 5 mass% graphite (improve the thermal conductivity up to 2.5 W/mK – higher than the thermal conductivity of the satureted FoCa clay- , independent from the degree of saturation of the mixture). The backfill blocks, compacted at a pressure of 61 MPa, have a dry density of 2.09 g/cm³.

Figures 2 to 5 give different views of the mock-up. The mock-up has an internal diameter of 2 m and a length of 5 m. The clay host rock and the gallery lining are replaced by a steel liner designed to resist to the internal pressure, caused by the porewater and the swelling pressures of the backfill material. The backfill blocks are placed in three concentric rings around the central tube. A peripheral placement gap of 38 mm allowed the installation of the hydration tubes on the inner surface of the steel cylinder to radially hydrate the backfill blocks. This gap also allowed the routing of the cables of the sensors embedded in the backfill material near the instrumentation covers made on the main cover closing the mock-up. All surfaces in contact with the backfill material are made from stainless steel (AISI 304 or similar).

There are 36 sections of backfill blocks, with a thickness of 130 mm each, around the disposal tube. The four last sections (close to the bolted cover) are modified due to the presence of a concrete ring. To test the behaviour of instruments to be placed in and against a concrete gallery lining, a concrete ring, consisting of 6 segments, has been installed around the disposal tube. In order to exert pressure on the segments, backfill blocks have been installed outside of the concrete ring. To guarantee the same swelling pressure at saturation around the concrete ring as the one exerted by the section with three rings of backfill blocks, the blocks used for the four last section have a higher clay content (85% instead of 60%). The gap inside the concrete ring and between the main cover and this ring is partly filled with sand.

The disposal tube has an external diameter of 508 mm and a thickness of 25 mm. This tube contains heating elements, which dissipate heat at a power of 450 W/m to simulate the radioactive wastes.

To obtain in the mock-up thermal conditions similar to the in-situ case i.e. about 120°C at the outer side of the backfill barrier, an external heating has been applied on the jacket. This external heating with two self-regulating cable independently controlled by a temperature controller allows to increase the overall temperature level in the mock-up but also to obtain a more uniform temperature along the jacket and to deal with some uncertainties such as the external temperature variations. To reduce the power needed for the external heating, a mineral wool isolation layer of 60 mm was installed on the jacket. To limit the axial heat flow, 300 mm of isolation was also placed on the covers of the mock-up.

Figure 6 : View of the mock-up after assembly of the set-up.
The mineral wool isolation around the mock-up is finished by a metal sheet.

Instruments are placed inside the backfill (see figure 5) and on the steel structure, mainly to monitor the thermo-hydro-mechanical behaviour of the backfill material. The temperature field is monitored by thermocouples, most of them arranged in radial and longitudinal configurations in the backfill. Additional thermocouples are installed on the heating elements and the external side of the jacket. Piezometers and humidity sensors monitor the hydration of the backfill blocks. Pressure and level sensors on the external hydration system complement these measurements. To monitor the swelling pressure exerted by the backfill material, total stress sensors are installed inside the blocks. Moreover strain gauges are installed on the disposal tube and the jacket. Their deformation allows to indirectly calculate the pressure exerted by the backfill material on the steel structure. As above-mentioned, load cells and pressure cells are also installed inside and between the concrete segments.

The assembly works started in March 1997. The hydration of the backfill material began in December 1997 (figure 8 shows the volume of injected water) . First the placement voids around blocks were filled with water and afterwards the water pressure was gradually increased to reach 1 MPa after two weeks. The water used for hydration of the backfill material is demineralised water with NaHCO3 added at a concentration of 1.17 kg/m³ to approximate the composition of the natural Boom Clay water. Six months after the hydration started i.e. in June 1998, the heating elements were switched on.

During the first month of heating, the hydration system was disconnected from the mock-up to test the thermal-hydraulic interaction. Consequently the pressure increased quickly in the backfill due to the expansion of the water. After about two months, a first maximum of temperature was reached, with a temperature at the central tube of about 115°C. In November 2000 the external heating was switched on to increase the overall temperature level in the backfill material. The temperature level was gradually increased during the experiment. At the occasion of one of these increases, the hydration system was disconnected again and after about 1 week a pressure increase of about 3 MPa was measured. In June 2000, after a last increase of the temperature level, the temperatures to be reached in the backfill (to be representative of the in-situ conditions) was obtained i.e. about 117°C on the outside of the backfill to about 137°C near the central tube (see fig. 8). This temperature profile will be maintained until the end of the heating phase.
The heating elements (external and internal elements) were switched off mid-August 2002 and the mock-up was dismantled (October 2002) after a cooling period estimated at 3 to 4 weeks. At the beginning of 2003, the post dismantling analyses began. They are now finished.

Figures 9 and 10 show respectively the evolution in temperature and total pressure (water and swelling pressures) recorded in the backfill material.

The temperature gradient across the backfill is about 20°C. The thermal conductivity derived from this thermal gradient is about 4.5 W/mK. This value seems to be to high for a porous material. Other additional heat transfer processes occurring in the backfill material, such as convection and/or evaporation/condensation process, could explain the lower than expected thermal gradient.

It must be noted that according to the total pressure measurements inside the backfill, the swelling is rather low. Moreover for about 2 years, the total pressure measured by these sensors is decreasing up to the value of the water pressure.

Since December 1998 water leaks outside the mock-up through the protective sheath of some wires of strain gauges installed on the central tube are observed. These leaks gave us the opportunity to analyse the backfill water. The analyses indicated a high content of chlorides (up to 1 kg/m³). After checking the unusual sampling conditions, it was found that this high concentration was most likely due to the backfill material itself. Unexpected values were also obtained for the NO3-, Si, DOC (dissolved organic carbon) concentration and the pH. A mass transport process could be at work in the mock-up concentrating salts towards the central tube. Indeed, independent laboratory experiments at room temperature performed by SCK•CEN have confirmed that soluble salts such as chloride are transported and concentrated by a water front migrating through the unsaturated material during the hydration phase. This advective transport maybe combined to other heat-coupled transport processes such as thermo-osmosis, thermo-diffusion, or advection-evaporation cycles, causes a salt enrichment of Cl- and other soluble salts near the central tube.

Another important observation done in routine operations when purging the accumulator of the hydration system was the presence of dissolved sulphides (analyses showed concentrations of 10-4 M in sulphides but no sulfate and thio-sulphate) in the mock-up water.

One objective of the mock-up OPHELIE was also to test the instruments working in harsh conditions of temperature and pressure in the backfill material. Failures were observed in some type of sensors. The problems encountered with the sensors deal mostly with watertightness, probably linked with corrosion, of the sensor body and the connection of the cables.

• the swelling properties of the backfill material were verified. No technical voids existed anymore (figure 11) ;
• the presence of sand and graphite in the mixture allowed to visualise the swelling process (figure 11);
• the joints between blocks are still visible (figure 11) for the M2 mixture. The water content of these joints is higher than the water content inside a block;
• the water saturation grade increases from the center to outside (figures 12 and 14) but the material was not saturated;
• perfect contact between central tube and backfill material;
• a strong cohesion of the blocks. It was necessary to use a electrical saw instead of a core boring machine for the taking of some samples;
• the presence of corindon (Al2O3) which came from pollution of the backfill material (pollution initialy present in the fresh blocks and coming from the factory);
• for the M14 mixture, the joints between the blocks were difficult to distinguish;
• All relative humidity sensors were severely damaged by galvanic corrosion (copper – stainless steel), and the power supply of these sensors probably made this corrosion even worse.
• corrosion of the hydration tubes : due to a failure of a relative humidity sensor feeded with DC current, one stainless steel tube was suffert from galvanic corrosion (anodic oxidation). Chromium was detected by SEM – EDS. Chromium (III) could be responsible for the green color observed around and on the tube (figure 13). On some other hydration tubes, a grey coating was observed, which might be due to pyrite recrystallisation
• very few traces of corrosion on the central tube were detected excepted where it was in contact with the sand
• Glotzl cells have totally failed spreading their oil in the fluid inside the mock-up.

• the corrosion of the metallic parts (by investigating geochemical –pore water chemistry- and microbiological phenomena, as they could affect the corrosion resistance of metallic components, and by direct analysis of the metal surfaces);
• the THM properties of the backfill material (swelling, permeability, water retention, thermal conductivity, and by investigating the mineralogical changes of the material…);
• the sensor performance (with recalibration or analysis of failure mode where applicable).

• an increase of the saturated hydraulic conductivity from values around 9 10-12 – 5 10-13 m/s (depending on the dry density) for the fresh material up to values around 3 10-12 – 9 10-12 m/s for the exposed one.
• an increase of the thermal conductivity. The mean value of the thermal conductivity coefficient of the fresh material is around 2,5 W/mK (water content of 7,7 %). For the exposed one, this coefficient raises slightly up to values around 2,7 – 3,1 W/mK for blocks from the middle and external rings and stays around 2,5 W/mK for the internal rings.
• very few mineralogical, physico-chemical and chemical changes were detected (MEB, E-SEM, TG, DSC, DRX, FT-IR).
• It seems that the fresh material was contaminated with bacteria (SRB, MFB – 200 bact./g). These bacteria, before their destruction by temperature, have contaminated the hydration circuit (accumulator - SRB, MFB, TSRB – 140.000 bact./ml ) by the thermo-convective circulation which took place in the mock-up.

For technological reasons not directly inherent in the experiment, this concept is no more considered as the reference concept for HLW. These technological problems were :

The problem of the FoCa clay availability and the difficulty to theoretically understand and to model the hydration process were also taken into account in the decision to change the concept.