OR/17/067 Gas testing
| Daniels, K A, and Harrington, J F. 2017. The response of compact bentonite during a 1-D gas flow test. British Geological Survey. (OR/17/067). |
Before gas breakthrough
Gas testing began on day 39. Additional helium was added to the IV to increase pressure to 3 MPa (Table 3). Gas pressure was then held constant for a further 7 days to allow the system to equilibrate with the DI water in the IV. At day 46, the injection pump was set to a constant flow rate of 500 mL/h. The injection pressure gradually increased for the next 8 days from 3 MPa to 5 MPa whilst the volume of fluid in the injection pump decreased from 102.7 ml to 6.25 ml. Data from the axial and radial load cells and the porewater pressure sensors is presented in Figure 7 and Figure 8. At day 54, the DI water in the injection pump was refilled by 95.9 ml to 102.15 ml and the flow rate was reduced to 375 mL/h to create a consistent ramp of pressure following an increase in volume (Figure 8).
At day 46, as the gas pressure ramp was initiated, and the pressure registered by the transducer attached to the injection end-closure filter showed an immediate increase in pressure (Figure 7). This is in contrast to the radial porewater pressures, which initially decreased, and in the case of arrays 1 and 2, then slowly increased again. The midplane porewater filter however, showed no obvious change between the start of gas testing and gas breakthrough; its value remained just below zero for the whole of this first part of the test. Although not able to measure an accurate value of suction, the small negative values demonstrate that there is a suction at this point with water being drawn from the filter into the clay. Detailed inspection of the data in Figure 7 shows that the porewater pressure in radial array 1, closest to the injection end of the sample, increased most quickly as gas breakthrough approached. The cause for this rise is unclear and may relate to a hydrodynamic effect as the gas acted against the injection face of the clay. In contrast, the rate of change in porewater pressure of radial array 3, closest to the backpressure end of the sample, showed a steady decrease between days 46 and 61, followed by a slight increase until gas breakthrough, indicating that the sample was not in hydraulic equilibrium at the start of gas testing.
At day 61, further helium was added to the IV to ensure there would be sufficient gas to complete the experiment (Table 3). This was done before the pressure in the injection filter reached the breakthrough pressure and gas started to flow through the sample; after breakthrough, it would not have been possible to provide additional helium to the IV without disrupting the flow of gas through the sample. Therefore, at day 61, just before gas breakthrough occurred, approximately an additional 0.21 mol helium gas was added to the IV, causing the injection pump volume to increase from 40.75 ml to 100.88 ml (Table 3).
Gas entry, gas breakthrough and shut-in behaviour
Gas entry occurred at 63 days, and the injection pressure at this point was 10.5 MPa. After gas entry, the porewater pressure in arrays 2 and 3 rose sharply to a value close to the injection gas pressure (Figure 9a and Figure 10a). Interestingly, despite beginning to increase first, the increase in pressure in radial array 1 (closest to the injection face) lagged behind that of the other two arrays by just over 0.2 days, which could indicate that the gas flow was non-uniform. Alternatively, the gas pathway may have carried gas to the edge of the sample, pressurising arrays 2 and 3 more quickly. After a small drop in porewater pressure around day 64.5 (Figure 7 and Figure 9a), porewater pressures then generally tracked the gas pressure for the following 11 days, with small deviations in the pressure traces probably reflecting changes in the stability, aperture and/or configuration of the pathways. At day 71, the injection pump was stopped. Between day 71 and day 76, the injection pressure decreased very slightly, whilst the porewater pressures decreased substantially (Figure 10a) with porewater pressure in radial array 1 continuing to fall until day 81.
Examination of the axial and radial load cell data during gas entry and breakthrough (Figure 9b and Figure 10b) indicates that the swelling pressure (stress) within the sample increased at the same time as gas breakthrough was occurring in the backpressure filter. The largest increase in radial stress was observed in load cell 3. This, supported by the porewater pressure data above, indicates a rapid increase in porewater pressure around this location in the sample. Following gas breakthrough, the system approached a quasi-steady state as gas pressure tended towards an asymptote and flow out of the sample more closely matched the gas flow into the sample (Figure 9b). The outflow reached a level that was just lower than the inflow, and it is possible that this discrepancy could have been caused by a very slight leakage. The continued increase in the stresses measured by the axial and radial load cells after the gas breakthrough event had occurred, mirroring the increase in injection pressure, suggested that the sample was not in complete hydraulic equilibrium at the start of the test and a redistribution of the fluid in the sample caused the clay to continue to expand.
At day 79, the injection pressure began to reduce at a slightly faster rate than it had been between days 71 and 79, the pressure in the three radial porewater pressure arrays and at the midplane started to increase again. This may suggest that another breakthrough event to the filter array was occurring. Alternatively, the increase in pressure measured by the midplane filter may be an artefact of the starting test conditions. If a filter is full of water, when gas reaches it at pressure an instant change in the pressure is observed, whilst if a filter is full of air, more gas is required at that filter to change the pressure. Radial filters 1–3 were filled with water at the start of the test, whilst the midplane filter was filled with air. Gas may therefore have been reaching the midplane filter since day 63 but the change in pressure recorded at that location would have been initially very small. Thereafter, the porewater pressure traces from each radial array continued to track the injection pressure response.
The dip in porewater pressure between days 71 and 81 also corresponded with a dip in swelling pressure and a reduction in outflow to zero (Figure 10b). Radial load cell 3 shows a decrease in pressure most clearly over this time interval. This event appears to occur close to the cessation of pumping. However, by day 81, porewater pressures had rebounded suggesting the cessation of pumping was not the cause for the spontaneous change in porewater pressure.
Following gas breakthrough, the stress measured by the load cells and the porewater pressure transducers appeared to be integrally linked to the gas pressure within the clay; both the stress and outflow show a similar form as the gas pressure (Figure 9), which was not the case prior to breakthrough. This continued following the cessation of pumping, as gas pressure, porewater pressure (Figure 10a) and swelling pressure (Figure 10b) began to decay. After day 81, the outflow was sporadic with positive fluctuations and spikes, despite the continued reduction in the injection pressure, suggesting new gas pathways continued to open and close. Some of these outflow events correlated with observed changes in the swelling pressure and porewater pressure, while others did not. The fluctuations in outflow are not related to changes in temperature (Figure 11). This suggests a dynamic process was operating within the clay, governing the local development of permeability.
