Modelling heat and moisture transport in the Andra/SKB temperature buffer test

For Andra program for thermal nuclear waste geological disposal, there is an interest in improving the understanding of the Thermo-Hydro-Mechanical properties and behaviour of bentonite-based buffer materials during the saturation process, in particular in the high temperature range above 100 C. To meet the need for experimental data, a large-scale field test, the Temperature Buffer Test (TBT) has been designed and set up at 420 m depth below ground surface in crystalline rock at the Aspd Hard Rock Laboratory in Southeast Sweden. The test, which is jointly operated by Andra and SKB, the Swedish Nuclear Fuel and Waste Management, is in operation since March 2003. Two heaters, each 3 m long, 0,6 m in diameter and generating 1500 W of thermal power, are stacked on top of each other within a bentonite buffer inside a vertical, 8 m deep and 1,8 m diameter, deposition hole. Ring-shaped blocks of bentonite are directly in contact with the lower heater, while there is a 0,2 m sand shield between the upper heater and the surrounding bentonite. Cylinder-shaped blocks of bentonite are set below, between and above the heaters. An anchored plug seals the hole and confines the experiment mechanically in the vertical direction. A sand filter is installed between the bentonite buffer and the rock wall to allow for control of the hydraulic boundary. Temperatures, relative humidities, total stresses, pore pressures, water inflow and forces take n up by the plug anchoring are monitored continuously, logged hourly and reported monthly. Prior to finalizing the design, numerical simulations were performed in order to establish the necessary heat output and to get a first estimate of the desaturation / re-saturation time-scale and of the temperature levels. Prior to test start, a program for predictive modelling was defined. Modelling teams organize d by Andra, SKB and Enresa made blind predictions of the thermal, hydraulic and mechanical evolution of the test. A number of codes were used for the predictions: ABAQUS, Code Bright, Code Aster and CLEO. A subsequent evaluation modelling program has been defined to explore the relevance and the validity of the assumptions made in the predictions. The present paper describes the approaches taken by the different modelling groups in the predictive phase. Examples of modelling results are given and compared with measured data. The focus is largely on the densely instrumented regions around the two heater mid-sections. In particular the extent of buffer dehydration close to the surface of the lower heater appears to be sensitive to the choice of model used to represent vapour transport. (authors)