Gas migration in KBS-3 buffer bentonite. Sensitivity of test parameters to experimental boundary conditions

In the current Swedish repository design concept, hydrogen gas can be generated inside a waste canister by anaerobic corrosion of the ferrous metal liner. If the gas generation rate exceeds the diffusion rate of gas molecules in the buffer porewater, gas will accumulate in the void-space of a canister until its pressure becomes large enough for it to enter the bentonite as a discrete gaseous phase. Three long tenn gas injection tests have been performed on cylinders of pre-compacted MX80 bentonite. Two of these tests were undertaken using a custom-designed constant volume and radial flow (CVRF) apparatus. Gas was injected at a centrally located porous filter installed in the clay before hydration. Arrangements were made for gas to flow to three independently monitored sink-filter arrays mounted around the specimen. Axial and radial total stresses and internal porewater pressures were continuously monitored. Breakthrough and peak gas pressures were substantially larger than the sum of the swelling pressure and the external porewater. The third test was performed. using an apparatus which radially constrains the specimen during gas flow. Observed sensitivity of the breakthrough and peak gas pressures to the test boundary conditions suggests that gas entry must be accompanied by dilation of the bentonite fabric. In other words, there is a tendency for the volume of the specimen to increase during this process. The experimental evidence is consistent with the flow of gas along a relatively small number of crack-like pathways which propagate through the clay as gas pressure increases. Gas entry and breakthrough under constant volume boundary conditions causes a substantial increase in the total stress and the internal porewater pressure. It is possible to determine the point at which gas enters the clay by monitoring changes in these parameters. Localisation of gas flow within multiple pathways results, in nonuniform discharge rates at the sinks. When gas injection stops, the gas pressure in the clay drops rapidly and then continues to decrease slowly with time. The transient provides clear evidence of discrete gas pathway 'sealing' events. Gas flow must cease when excess gas pressure (i.e. relative to porewater pressure) falls below the capillary pressure. The capillary pressure for bentonite is shown to be approximately equal to the swelling pressure. There is no evidence from these long-term tests that the development of gas pathways in any way compromises the sealing capacity of the bentonite barrier