Experiments are being performed in the Asse salt mine in Germany to study the sealing capacity of backfill and host rock in a deep geologic salt repository and to provide a basis for further refinement and validation of numerical codes required for the prediction of long-term repository performance. The experiments investigate different disposal concepts for heat generating radioactive waste which were developed for a repository in a salt dome. The paper presents a description of the projects and summarizes the results from several years of experiment performance.

The work described in this paper is related to two emplacement concepts for heat generating radioactive waste (Bechthold et al., 1997).

• The drift emplacement concept was developed to dispose of spent-fuel assemblies. The assemblies are containerized in self-shielding 1.6-m-diameter and 5.9-m-long casks weighing 65 tonnes (so-called Pollux casks). The casks are emplaced on the floor of horizontal drifts about 800 m deep in the repository. The empty space in the drifts is backfilled immediately. Initially, this concept was developed for the direct disposal of spent fuel. However, it can also be used to dispose of high level waste in self-shielding casks.
• The borehole emplacement concept provides the deposition of high level waste canisters in 300-m-deep vertical boreholes beneath repository drifts. The annulus between waste canisters and borehole wall is backfilled. In order to isolate the waste canisters from the drifts, the remaining volume above the canister stack is also backfilled. After all boreholes are filled, the drifts are backfilled, too. Finally, the access drifts and shafts are backfilled and sealed.

In these concepts the functions of the backfill are to

• support the host rock around the drifts and boreholes,
• stabilize the convergence process,
• conduct the decay heat from the waste to the host rock,
• act as a barrier against inflowing brine or water,
• distribute the stacking load of the waste canisters to the surrounding rock.

Crushed salt was decided to be the most suitable backfill material. As a consequence of rock salt convergence, its permeability decreases continuously to very small values which are almost equal to the undisturbed rock. Furthermore, the mechanical behavior of the salt grains is similar to the surrounding rock salt. Therefore, crushed salt provides a continuously increasing resistance to drift closure and a relatively homogeneous stress distribution around drifts and boreholes. According to the present planning, the material for drift backfilling will consist of crushed salt that remained from the excavation works, i.e., a coarse material with a maximum grain size of 60 mm. The material for backfilling boreholes will be fine-grained crushed salt with a 10-mm maximum grain size.

For sealing emplacement panels, bulkheads will be constructed in the access drifts. Their function is to isolate the waste disposed of in the emplacement areas and to prevent brine from intruding into these areas during the time period in which the backfill is still a pathway for fluids in the repository. The effectiveness of the bulkheads will depend on the permeability of the excavation disturbed zone (EDZ) in the host rock around the drifts.

In order to investigate these phenomena, three in situ tests are carried out in the Asse salt mine. They were designed to simulate reference repository conditions of the emplacement concepts described above. Their purpose is to expand the understanding of backfill and rock behavior which earlier had been gained primarily from laboratory investigations. In these in situ experiments the combined effects of heat, convergence, rock pressure, and backfill compaction are being studied. The experimental results are used to refine and qualify the numerical models and to evaluate their capability to predict the performance of a repository over long time periods.

The objective of the Thermal Simulation of Drift Storage (TSS) test is to study the backfill behavior in emplacement drifts and to improve relevant computer codes and constitutive equations required for predictive calculations. According to the drift emplacement design, the Pollux casks will be deposited in 3.5-m-high and 4.5-m-wide drifts. The initial porosity of the crushed salt backfill will be about 35%. Therefore, its sealing capacity will be low in the beginning. As a consequence of host rock convergence, however, the backfill will be compacted until its permeability is almost equal to the surrounding rock salt. This process which is influenced by temperature, thermal conduction, rock convergence, and rock pressure has to be predicted for long time periods in repository performance assessments. The relevant models have to be calibrated and their predictive capability has to be evaluated.

The experiment is carried out in a test field consisting of two backfilled drifts and several observation and access drifts. In each test drift three electrically heated containers were deposited. The dimensions of the test drifts and the simulated casks as well as their heat power are as outlined above in the drift emplacement concept (Fig. 1).

• temperatures,
• stresses,
• backfill porosity and permeability,
• rock deformations,
• gas generation and transport, and
• corrosion of container materials.

The measuring instruments are placed on the heater casks, in the backfill, and in the surrounding rock. Measurements are conducted in the heated area around the heater casks as well as in the non-heated regions farther away from the casks where the influence of heating is less significant (Droste et al., 1996). The test is operational since September 1990. Heating was at a constant power output of 6400 W / Pollux cask or 19200 W / drift. In the following, some characteristic results of the measurements are presented.

At the surface of the casks a maximum temperature of 210°C was reached after five months. Because of the increasing thermal conductivity of the backfill during its compaction, the temperature subsequently decreased to 165 - 175°C at the moment (Fig. 2). Short-term irregularities in the temperature decrease were due to malfunctions of the heater power control system. At the drift floor the temperature has currently reached 125°C (Fig. 2). Temperatures of up to 100°C are recorded at the drift walls and up to 92°C at the roof above the casks. In a depth of 0.3 m below the heaters temperatures of 145°C are measured, decreasing to 107°C in 1.2 m depth. Between the heated drifts temperatures are reaching 81°C in the pillar centre. In the almost non-heated area 12 m away from the heaters the temperatures have increased up to 47°C in the meantime. Actually, the temperatures predicted by thermal model calculations agree very well with the measurements (Pudewills, 1997).

Fig. 2. Drift temperatures around a central heater and at the heater surface

With the start of heating, a significant acceleration of the drift closure rate was observed. Within three months, the rates increased by a factor of 8 to 10 compared to the non-heated sections of the test drifts (Fig. 3). Because of the increasing rigidity of the compacted backfill, the rates have been reduced again subsequently. Present rates in vertical direction amount to 0.6 %/a, whereas horizontal rates of 0.45 %/a are recorded. These values are still twice as much as before heating. In the non-heated area the drift closure increased gradually, currently reaching 0.5 %/a and 0.3 %/a in vertical and horizontal direction, respectively. Calculated drift closure rates in the non-heated sections show a rather good agreement with in situ measurements. In the heated area, however, the closure rates are overestimated by the calculation by about 40% (Korthaus, 1991). This may be caused by the modeling of the test geometry, the chosen values of the material parameters of rock salt and backfill, the constitutive equations describing the backfill compaction, and the assumed initial stress state in the test field.

To observe the thermally induced stress change in the host rock, about 50 stress monitoring stations with Glötzl-type hydraulic stress gauges have been installed in the test field. Different set-ups with 4, 7 or 8 components with different orientations are being used. The stations have been installed in the roof, the walls, and the floor of the drifts as well as in the pillar at different distances to the heater casks.

In most measurement directions a significant increase of stress was observed a short time after excavation of the test drifts. Subsequently, a decrease of stress due to creep and stress relaxation of the host rock was measured. With the start-up of heating in September 1990, the stresses increased again reaching a maximum after some weeks but then decreased continuously. Since 1994, a quasi steady state is observed around the test drifts. Because of the ongoing heat dissipation, however, an increase of stress is still being observed at farther distances north and south of the test field.

The increasing backfill compaction is creating a rising pressure between the backfill and the surrounding rock. In the heated area a maximum backfill pressure of 3.5 MPa is currently measured at the roof (Fig. 5). The average backfill pressure at the roof is also significantly lower than expected (Korthaus, 1991). The pressure has reached only 24 % of the initial vertical stress estimation of 12 MPa in the test field. In the non-heated backfill the current pressure increase ranges between 0.125 - 0.25 MPa.

Fig. 5. Backfill pressure in the heated area

In view of the great impact of the backfill on repository safety performance, it is necessary to predict its behavior under conditions similar to those expected for a repository. In order to perform repository safety assessment, different computer codes and user models have been developed in the past. These codes and models must be verified, validated, and analyzed with regard to their predictive capability. Therefore, a benchmark exercise "Comparative Study on Crushed Salt" (CS²) is carried out under participation of six modeling groups. The main emphasis is to further refine the constitutive models developed in the past to describe the backfill behavior. The objective is to

• assess the accuracy and reliability of predictions,
• improve the confidence in codes and models,
• identify future research needs,
• exchange and transfer technology.

Various constitutive models for crushed salt have been developed by the modeling groups. Because of the complex behavior of the crushed salt, the constitutive models contain a great variety of material parameters which have to be derived from laboratory tests. The benchmarking exercise provides a means of comparing and calibrating the constitutive models. The following table presents the participating modeling groups and their models.

The benchmark consists of three progressive stages:

• Stage 1: Benchmark calculations on simple theoretical examples to compare available codes and to evaluate their capability.
• Stage 2: Comparative calculations related to specified laboratory tests in order to further refine and to validate the constitutive models.
• Stage 3: Numerical modeling of the TSS experiment and comparison of the experimental results with calculated results. In this final stage the calibrated constitutive models derived from the stage-2 exercise are used.

The objective of this benchmarking is to provide modeling tools to predict the performance of crushed salt backfill under repository conditions and to create a basis for exchange and transfer of modeling expertise.

In 1999, after more than eight years of heating, the heat power will be continuously shut down over a six to twelve months period according to current planning. However, no definite timetable has been set yet. A post-heating investigation programme will be carried out from 1999 to 2001.

The experiment will improve the understanding of processes relevant to backfill and rock salt behavior in heated emplacement drifts. The numerical models that extrapolate forward these processes will be refined and calibrated.

The DEBORA (Development of borehole seals for radioactive waste) experiments investigate the backfill behavior in the annulus between HLW canisters and borehole wall and in the seals of HLW emplacement boreholes. The objective is to determine the interactions between backfill and near-field rock salt and to develop further the numerical models to simulate these processes (Rothfuchs et al., 1996). According to the borehole emplacement concept, the canisters will be stacked in 300-m-deep boreholes. Backfill will be emplaced around the canisters to distribute the stacking load of the canisters to the surrounding rock. The boreholes will be sealed against the drift above by emplacing backfill material in the upper 30- to 40-m part of the borehole.

In the early phase, the backfill will have a rather high porosity and permeability providing a flowpath for brine and gases. However, as a consequence of rock convergence, the backfill will be compacted and the flowpath closed. Numerical models that are based on laboratory data have been developed in the past to describe the performance of backfilled emplacement boreholes. These models are refined and calibrated by comparing in situ data with calculational results.

The experiments are carried out in two separate boreholes (Fig. 6). DEBORA 1 is carried out in a 600-mm-diameter and 14-m-deep borehole beneath a drift 800 m deep in the mine. A liner in the borehole represents the waste canisters in a disposal borehole. Electric heaters inside the liner produce 9 kW of power. In a 15-months testing period, the heat induced compaction of the backfill between the liner and the borehole wall and the related decrease of porosity and permeability are investigated. The experiment is operational since February 1997.

Heating of the DEBORA 1 borehole led to a rapid increase of the backfill temperature to about 180 °C. The temperature is only slightly rising during the further course of the experiment. In the DEBORA 2 experiment the temperature increase is not that fast because the heaters are located outside the central borehole at a radial distance of 1.1 m. In both experiments the heat induced borehole convergence (closure) is being measured at three levels at depths of 10.5, 12, and 13.5 m below the drift floor. The decrease of backfill porosity is determined on the basis of borehole convergence measurements. For both experiments, Figure 7 shows the comparison of the measured and predicted development of the porosity 12 m below the drift. The predictions were made with the code Supermaus (Breidenich, 1993) under consideration of a constitutive equation describing the thermomechanical behavior of crushed salt published by Hein (1991). In both cases, the measured compaction rates are smaller than predicted; so a further improvement of the used models seems to be necessary.

The DEBORA experiments are scheduled to be terminated by the end of 1998. It is planned to uncover the boreholes by a new access drift and to analyse the spatial distribution of porosity and permeability in the compacted backfill.

The project ALOHA deals with the investigation of the excavation disturbed zone (EDZ) around cavities in salt formations. The objectives of the project include the characterization of the hydraulic behavior of the EDZ and the investigation of its extension. Its development with time, i.e., the reduction of the EDZ after placing a bulkhead and subsequent stress build-up, cannot be investigated directly. Instead, the relation between stress state and hydraulic properties of the salt has to be considered.

The investigation program consists of a series of in situ measurements backed by laboratory and modeling works. The in situ activities include

• the measurement of the salt permeability to gas by gas injection at various distances from a test drift,
• the injection of liquid (brine) into the salt rock, and
• the investigation of the spreading of the liquid in the rock by geoelectric measurements.

The in situ measurements are performed in a test drift at the 875-m-level of the Asse salt mine. A plan view of the test drift (cross section about 4.5 m by 4 m) and a schematic view of the test borehole arrangement are given in Figure 8. Two parallel vertical boreholes of 25 m depth and 56 mm diameter in the drift floor are used for the injection tests with gas and liquid, respectively. The liquid injection borehole is surrounded by another five boreholes containing the electrodes for the geoelectric measurements. Additionally, three electrode profiles are located on the drift floor.

During 1997, a series of gas injection tests were conducted at borehole depths between 1 m and 20 m. For the tests a hydraulic multipacker system that can be moved in the borehole was used. It consists of four sealing elements of 0.5 m length each. They are separated by an upper and lower observation interval of 0.3 m length and an intermediate test interval with a length of 0.8 m. The sealing elements are inflated by oil up to a pressure of about 8 MPa. After the packers have settled (about one day after inflation) the test gas (nitrogen) is injected at a constant flow rate of 500 - 550 ml/min into the test interval. When the maximum pressure of 3 MPa (2 MPa for the measurement at 1 m depth) is reached, the test interval is shut in and the pressure decay is monitored. Flow rate and pressure data are recorded by a PC-based data acquisition system. From the measured data permeability is derived using the computer code WELTEST (Intera, 1992) which iteratively minimizes the root mean square deviation between measured and calculated pressure values.

For the liquid injection tests a double packer system similar to the one for the gas injection tests was used. The brine was injected by a pump; the injected amount was measured both by a flowmeter and by continuous weighing of the brine container. An overview of the complete system is shown in Figure 10.

The geoelectric measurements are performed by injecting a current by two electrodes into the rock and measuring the potential difference between two other electrodes. By varying the injection and measurement electrodes a distribution of the electric resistivity in the rock can be obtained. The resistivity of rock salt is connected to its moisture content by a well-established proportionality. Thus, the geoelectric measurements yield the moisture distribution in the rock.

The sealing capacity of backfill and rock salt in a salt repository are investigated by different In-situ experiments in the Asse salt mine by simulating drift and borehole emplacement.

In the TSS test the backfill behavior in emplacement drifts is investigated. Measurement results are available over more than seven years. Temperatures predicted by thermal model calculations are confirmed by the measurements. The drift closure was accelerated considerably by heating. Drift closure rates and backfill compaction, however, are significantly slower than expected. The backfill pressure has reached only 24 % of the initial vertical stress estimation of 12 MPa in the test field. This value is significantly lower than the expected lithostatic pressure of 18 MPa due to the influence of former mining activities in the Asse salt mine. Additional calculations and improvements of the numerical models by benchmark calculations with different codes are planned to refine a prediction of the behavior of crushed salt backfill under repository conditions.

In the DEBORA project model calculations were performed to predict the heat induced compaction and porosity decrease in simulated HLW disposal boreholes in the Asse mine. Preliminary results of the experiments started in February 1997 and scheduled to be terminated by the end of 1998 indicate a slower compaction and thus a slower decrease of the porosity and permeability of the crushed salt than predicted in the calculations. The used models are to be improved by further laboratory investigations on crushed salt compaction behavior.

In the ALOHA project an excavation disturbed zone extending less than the drift radius into the salt rock was found. The EDZ provides a significant pathway to both gas and liquid with a permeability contrast of at least four orders of magnitude compared to the undisturbed rock.

The works were funded by the Federal Ministry for Education, Science, Research, and Technology (BMBF). The TSS test and the DEBORA tests are co-sponsered by the European Commission.