Hydraulic Testing of Compacted Bentonite Used for Plug and Abandonment Operations

The University of Queensland Centre for Coal Seam Gas (UQ CCSG) has in-vestigated plugging wells with bentonite through laboratory experiments and with field trials. This paper presents the laboratory tests, which were used to investigate the stability range of plugged sections for later well plug and abandonment operation designs. The plugs were tested on a specially built well simulator facility at The University of Queensland (UQ), School of Chemical Engineering. The bentonite material used for the plug production was treated with water and 1 weight% polyvinylpyrrolidone (PVP) which acted as a binder to allow the bentonite to be pressed into a cylindrical shape suitable for dropping into vertical wells. The experiments have shown that the best performing plug/casing size combination is able to hold pressure gradients of up to 5.9 bar/m (25.9 psi/ft) after 296 days of hydration before failing. Open hole si-mulations on the testing facility showed surprisingly high failure pressure gradients of 21.1 bar/m (93.3 psi/ft) after 146 days of hydration. The findings of this research indicate that the use of compressed bentonite is a viable me-thod for sealing wells, whether they can be coal seam gas wells, conventional oil and gas wells, water wells, or coal exploration wells.


Materials and Plug Production
Amcol Australia Pty Ltd supplied three of their mined raw bentonite materials from its Queensland Bentonite Mine (Gurulmundi, Australia). The three different bentonite materials were tested during the project to identify the most suitable material based on the comparison of mineralogical, geochemical, geomechanical properties and finally their plugging performance, tested with the well simulator. The best performing material was the Amcol 5D bentonite, followed by the 5A bentonite. They were used after PVP treatment (5D PVP/5A PVP) or as untreated material (5D raw/5A raw) for further testing.  Raw bentonite treated with potable water and 1 weight% polyvinylpyrrolidone (PVP)is pressed into a bullet geometry with a cone shaped tip and a center hole that allows fluid to pass through the plug and stabilize the fall through a water column to the final setting depth [5] [7]. A hydraulic pressure of 16.5 MPa (2400 psi) is applied during the production process, which is equal to 25.8 metric tons compression weight for the 139.7 mm (5.5") plugs. 15.2 MPa (2200 psi) was used for the 88.9 mm (3.5") plugs, which translates to 9.6 metric tons compression weight.

First Hydraulic Tests Using the Well Simulator Facility
The Well Simulator Facility consists of one meter long pipe sections of 97.2, 139.8 and 173.1 mm internal diameter (ID) (red, blue and green pipes in Figure   2). These were stacked and flanged to each other to create taller test sections.
First trials were commenced in July 2017 after calibrating the digital pressure transducers, using the pre-set proportional pressure relief valve with a nominal set pressure of 6.9 MPa (1000 psi).
Spacers were plunged into the pipe sections to ensure that the pressure transducers in the lowermost flanges are still operational, even after filling the pipes up to 20 cm below the top flanges with hydrating bentonite plugs. The pipe section was loaded with bentonite plugs by dropping them from the top into a casing partly filled with potable water. The pipe section was filled with water to the top after installing the plugs. The first trials were commenced without installing blind flanges on the top to allow visual inspection of the failure that occurred during pressuring the plugged pipes up while pumping potable water into the bottom flange. Figure 2. The Well Simulator Facility at the University of Queensland. Three different pipe sizes can be installed and tested up to a maximum pressure of 6.9 MPa (1000 psi). Sections of up to 5 m height can be installed. One pressure transducer is installed at the flange of each pipe and allow pressure monitoring while pumping. days of hydration using potable water. The constant pump rate during the test was 4.8 l/h (1.3 ml/s). We expected to see a very similar result to those provided by the mechanical dislodgement tests using a loading frame [5].
We could not identify the same stepwise geometry due to gradual shear failure of each single plug. We registered a first distinct signal caused by slip related pressure loss after 150 seconds (10.8 bar/157psi). There was no water production visible on top of the plugged section. Pressure was building again slowly and a couple of smaller peaks could be recorded. After 810 seconds final failure occurred at a maximum pressure of 19 bar (275 psi) and water production was detected at the top of the plugged section. The smaller peaks between 150 and 800 seconds are believed to be triggered by stick-slip phenomena of multiple or single bentonite plugs at different times within the plugged section. Material heterogeneities within the compacted bentonite plugs might be responsible for the "masked" dislodgment patterns compared to the patterns detected in the mechanical loading frame tests [5]. The pump was running a further 460 seconds after failure of the plugged section before being shut down. The shut-in pressure readings stabilized finally at 11.1 bar (161 psi), which are very close to the pressure peak, which was detected at the beginning of the test.
The second hydraulic test in the Well Simulator was conducted using the same material (5D PVP) and the same test geometry/height as used in the first trial, but with a lower constant pump rate of 2.4 l/h (0.7 ml/s). The results are presented in Figure 4 and show a lot more similarity to the data derived from the mechanical dislodgement tests using a loading frame [5]. Again, we can't assign single pressure peaks to single plugs. Failure of the plugged section caused a water leakage at 5 bar (72 psi) after pumping 1920 seconds. The water flow was detectable at the interface plug/steel pipe. The central hole within the plug was   We decided to keep the tested plug interval in the simulator and tested it again a week later after further hydration (58 days), but this time with a much higher constant pump rate of 36 l/h (10 ml/s). The results are shown in Figure 5. The rig pump was operated at 80 percent of its pumping capacity and the pressure data readings are quite noisy because of vibrations of the pump, which is directly installed beneath the rig table in contact with the metal structure. Final failure happened after a pumping period of 125 seconds at a pressure of 39.8 bar (577 psi). There was no visible water production at the top of the plugged section, instead the hydrated bentonite failed in shear mode (frictional failure) and was pushed out of the casing. The pump was shut down after 225 seconds. At this stage, 400 mm of hydrated bentonite plug material (two of the ten loaded plugs) was already pushed out of the pipe and was still moving until we depressurized the system using a manual pressure relief valve and the plug movement stopped.
The high failure pressure was not triggered by the extended hydration time of 1 week. The results of the long-term tests presented later in this paper clearly show that maximum strength of the plugs is reached after much longer lasting hydration periods. The fact that the bentonite plug was pushed out of the casing shows that we are dealing with a different failure mode than observed in the two tests mentioned before. Hydrated bentonite seems to show a similar behavior like non-Newtonian Fluids while failing (shear thickening effect). High pressure/flow rate will trigger shear failure at the complete circumference of the Figure 5. Second hydraulic test continued. Ten 88.9 mm OD 5D PVP treated plugs installed in 2 m of 97.2 mm ID pipe are tested after 58 days of hydration using a constant pump rate of 36 l/h (10 ml/s). hydrated plug at a high failure pressure, whereas low flow rates creating single micro channels at the interface plug/steel pipe at much lower applied pressures, as shown in the following tests.
The third and fourth hydraulic test have been done using single 1 m pipe sections (97.2 mm ID) with pre-hydrated bentonite. Single pipes were loaded with bentonite plugs, hydrated in the laboratory and later installed on the simulator tooptimize the testing capacity of the rig. Figure 6 shows a comparison of the 5A bentonite as PVP blend and in its untreated form. They were tested after 48 days of hydration using a constant pump rate of 2.4 l/h (0.7 ml/s). The recorded curve geometries are quite different compared to the test results using water (blue curve in the background of Figure 6). There was a small volume of air (0.5 l) at the bottom of the plugged section above the water level within the installed pipe, because of the pre-hydration procedure. This air was used to test failure in presence of gas flow. The gas volume was compressed while pumping water. The plugged section finally failed as in the second hydraulic test with water, after the complete air volume escaped through the micro channels at the interface plug/steel pipe.
Both materials were tested after 48 days of hydration. Failure was observed when air bubbles were detected at the interface plug/steel pipe (   . Failure behaviour of PVP treated 5A and raw 5A bentonite after 48 days of hydration. Left: 5A-PVP in a 97.2 mm ID pipe (third hydraulic test). Air bubbles (arrows) occur at the interface plug/steel pipe using micro channels as pathways, generated at failure pressure. Right: 5A-raw in a 97.2 mm ID pipe (fourth hydraulic test). Air bubbles (arrows) occur at the center of the plug and at the interface plug/steel pipe at failure pressure.
psi). The 5A raw bentonite failed in the center and at the plug/steel pipe interface after 1265 seconds and 0.4 bar (5.8 psi) failure pressure (Figure 7, right side). The center of the 5A raw plug was not cured with expanded bentonite after 48 days of hydration. Direct comparison of PVP blended bentonite with the raw material confirms that the polymer blend is 60% stronger than the raw material and the failure pressures are roughly 50% higher using water than failure pressures using air.
The decision was made to pre-hydrate plugs in single pipe sections for the long-term testing to complete the planned experimental program in the two year project time frame. The 97.2 mm ID pipe sections were loaded with 5A PVP and 5D PVP bentonite plugs and stored in the laboratory with a 20 cm long PVC pipe extensionon top of the pipe to guarantee a continuous hydraulic head on top of the plugged section. The PVC pipes were filled to the top with potable water and covered with a lid. The water level was checked 3 times per week and topped up if necessary. The pipes were installed on the rig after 3, 6 and 9 months hydration time and tested.

Hydraulic Long-Term Testing
The two most promising bentonite materials to be tested in the long term test

Results of the 97.2 mm ID Pipe Sections
Three 97.2 mm ID pipe sections were loaded with five 5A PVP and 5D PVP bentonite plugs each and pre hydrated in the laboratory. The results of the six hydraulic tests of the 97.2 mm ID pipe long term testing are shown in Figure 8 and Figure 9. The samples were tested after hydrating for 99, 183, and 296 days.
After all three tests the 5D PVP bentonite performed significantly better than the 5A PVP. The failure pressure of 5D PVP is nearly double as high after 296 days as the one from the 5A PVP material. Water production on top of the plugged sections could be detected in all tests.

Results of the 139.8 mm ID Pipe Sections
127 mm OD diameter plugs were prepared by stripping 12.7 mm from the originally produced 139.7 mm OD plugs (Figure 10). A section of 4 m was filled with the plugs and hydrated. Water was refilled weekly at the top to assure a constant water head on the installed bentonite. This experiment was designed for two purposes: to test the long term failure behavior of the plugs and to examine the "self-healing" ability of the plugged section after being dislodged [7]. It was expected that the sealing performance of the bentonite plug during/after the healing process would actually increase.
The pump rates were increased during the third and fourth test, because no water production could be detected at the top of the plugged section after the first pressure drops (black arrows in Figure 11). The massive increase of pressure at final failure seems to be impacted by the effect similar to non-Newtonian fluid behavior, as already described in Chapter 3 (second hydraulic test, continued, Figure 5).    The original data were height converted for better comparability with the 97.2 mm ID pipe long term test results, assuming that the failure pressure is directly proportional to the height of the plug [6].

Results of the 173.1 mm ID Pipe Sections
able to expand enough to create a proper seal in the 173.1 mm ID pipe section.
The decision was made not to proceed with this testing geometry combination because the moisture content of the hydrated plugs was simply too high to create a proper seal.

Simulation of Testing Bentonite Plugs in Uncased Holes
An 80 cm long sandstone cylinder with an ID of 94 mm was drilled out of a block of Helidon Sandstone out of the Ipswich-Moreton Basin (Figure 14). This geometry is close to HQ size (96 mm), which is the usual size of continuously

Summary of the Results of the Hydraulic Long-Term Testing
A summary of the failure pressures of all long-term hydraulic test results is given in Table 1 and Figure 18. The erroneous results of the experiment "139.8 mm ID pipe, 5D PVP, high flow" are also included in the table and figure to show the pitfalls arising in this kind of experiments.
The comparison of the two best performing bentonite materials show a clear result: the best performing material is the 5D PVP blend that is able to hold pressure gradients up to 5.9 bar/m (25.9 psi/ft) ( height of a plugged section for a given maximum reservoir pressure. The results of the 139.8 mm pipe testing are not as conclusive. Figure 18 shows two blue curves for the 139.8 mm pipe 5D PVP plugs: The bright blue curve with the abnormally high pressure readings for the two last data points, is impacted by the effect similar to non-Newtonian fluid behavior triggered by increased pump rates. The dark blue curve (5D PVP, 139.8 mm pipe, min flow) represents failure after using a constant pump rate of 2.4 l/h (0.7 ml/s).
It was expected that the dark blue curve would plot below the grey curve representing a pipe/plug combination with a smaller diameter. Bentonite is more stable in its plugging performance if it has to fill smaller "voids" after hydration which is directly related to the in situ moisture content [6]. The higher the final moisture content, the weaker the plugged section. The increased pressure volumes of the two last readings of the dark blue curve (alternative interpretation) are well above the 97.2 mm pipe results. This is caused by re-healing of the bentonite plugged section. The plug increases in strength after being disturbed due to the testing and later rehydration [6].

Conclusions
Different failure modes depending on flow rates were identified during the hy- Slow pumping creates single micro channels at the interface of the plug/steel pipe at much lower applied pressures. At low flow rates, failure modes are more complex and material/hydration time-dependent. The 5D bentonite hydrates more quickly than the 5A material. The hole in the center of the plugs is sealed earlier when using the 5D material.
The identification of the micro channeling effect influences the design for well abandonment planning. The lower failure pressure levels for gas compared to water detected in this study have to be taken into account for the abandonment planning of different well/reservoir types.
Failure pressures are roughly 50% higher for the bentonite/PVP blend plugs while using water than failure pressures using air. The failure gradients reported in Table 2 are based on failure using water as the pumping medium. They have to be corrected to 50% lower pressure gradients for free gas containing wells.
The failure gradients shown in Table 2 can be used for the design of abandonment operations using comparable plug/pipe diameters. The 5D PVP blend is easily able to withstand 69 bar (1000 psi) reservoir pressure in a 97.2 mm casing after installing a 30 m high plugged section, even after reducing the reported gradient by a 50% safety/contingency margin for the basis of design.
Nuclear magnetic resonance (NMR) experiments might help to create a better understanding of the process of the micro channeling generation. A future project should use this or an alternative suitable detection technique to try to map/visualize the growth of these channels at the interface of the plug/steel pipe during a hydraulic test.