Compressed bentonite in the form of pellets or plugs is used for the abandonment of production wells for the oil and gas industry. The design of the abandonment systems is based on the hydro-mechanical behaviour of the compressed bentonite defined by mechanical parameters that are used from published data rather than quantified for the used material by laboratory investigations. This paper presents an experimental study on characterising the swelling and shear strength behaviour of raw and polymer (polyvinylpyrrolidone, PVP) treated bentonite. Dislodgement tests consist of three hydrated bentonite plugs inserted in steel casings with the failure mechanism characterised. The bentonite used comes from a local mine (in Queensland, Australia) and is comparable to other bentonites usually used for the abandonment of wells or for other problems where mineral sealing is required (e.g. basal clay barriers of landfills). The experiments have shown that polymer treated bentonite shows significantly larger shear strengths than raw bentonite with simultaneously less swelling. More compressed samples also showed higher shear strengths and less swelling. The dislodgement tests have characterised for the first time the cascaded failure mechanism of a series of plugs forming an abandonment system. This investigation is the first step towards the development of an improved design for abandonment systems for wells using bentonite plugs.
Highly compacted bentonite can be used as material to produce plugs or pellets for the abandonment of wells [
Geomechanical data used in the modelling of the hydro-mechanical behavior to date is based on literature reviews and includes high uncertainties because it is based on different bentonite materials [
・ An oedometer set-up to conduct swelling tests at constant normal stress that can observe the residual swelling for different stress states;
・ Direct shear tests to determine strength parameters at different hydration states. The results of the oedometer tests are required to define the hydration state at which the mechanical tests have to be conducted; and
・ Mechanical dislodgement tests of hydrated bentonite plugs in steel casings using an electro-mechanical self-reacting loading frame to measure the force required to dislodge the plugs within the casing.
Two different compositions of bentonite samples have been tested: samples with a composition of raw bentonite and a bentonite/polymer mixture (polyvinylpyrrolidone, PVP). The polymer was added as a binding agent [
Samples were prepared using a hydraulic press in the laboratory of the School of Chemical Engineering. Spherical discs of 71 mm diameter and approximately 4 mm height were produced. After preparation, samples were sealed and stored for further use. The gravimetric water content of the samples was determined to be 12% ± 1%.
The bentonite plugs for the mechanical dislodgement tests were compressed in a vertical hydraulic press. Plugs were prepared out of raw bentonite material and the bentonite/PVP mixture. The hydraulic ram applied 17.2 MPa (2500 psi) pressure, which is equivalent to 14.2 metric tons using the press geometries for the 140 mm (5.5”) diameter plugs. The compressional force was applied onto the bentonite in a bullet shaped mould (sample chamber) for 2 minutes.
The plugs were loaded into 80 cm long 177.8 mm (7”) outer diameter steel pipes (161.7 mm/6 3/8” inner diameter). Each pipe contained three plugs. Overall, 14 pipe sections were set up (7 raw bentonite/7 bentonite/PVP mixture) and hydrated over 42 days, maintaining wet conditions by filling water to the top of the pipe daily.
X-Ray Diffraction (XRD) analysis was conducted to determine the bulk mineralogy of the bentonite and clay mineral analysis on the separated <2 µm clay fraction.
The bulk sample was split, crushed then weighed and a specimen prepared for XRD analysis. A randomly oriented sample was prepared for XRD analysis spiked with an internal corundum standard (Al2O3) at 10 weight % for a quantitative phase analysis. The specimens were micronized in a McCrone mill using zirconia beads and ethanol as a fluid and dried in an oven at 40˚C to evaporate the ethanol.
A small portion of the bulk sample was disaggregated and dispersed in water. The fine fraction (<2 μm) was transferred to a low background plate and allowed to settle and dry. This preparation is used to concentrate the fine (clay dominant) fraction and aids identification of the clays present. The fine fraction specimens were then further treated with ethylene glycol (which expands some clays) and re-examined. Additionally the sample was heated to 375˚C and measured again. This process caused the smectite (001) reflection to collapse, producing a diffraction pattern which is similar to illite [
Step scanned XRD patterns were collected using a PANalytical vertical diffractometer using cobalt Kα radiation. Intensities were measured at a step size of 0.0167˚ 2 Theta, 43.2 s/step and a scan range from 4˚ - 90˚ 2 Theta using a 0.5˚ fixed divergence slit. Generator voltage was 40 kV and tube current 40 Ma. The collected data was analysed using JADE software for phase identification and SIROQUANT software for quantitative phase analysis using the Rietveld method. The known concentration of corundum facilitates reporting of absolute concentrations for the modelled phases. The sum of the absolute concentrations is subtracted from 100 weight % to obtain a residual, also sometimes called “amorphous content”. The residual represents the unexplained portion of the pattern: it may be non-diffracting content but will also contain unidentified phases or poorly modelled phases.
The oedometer tests were conducted with a conventional loading frame. For the implementation of oedometer tests specific to bentonite, several modifications have been trialed and introduced compared to standard oedometer tests to avoid bentonite being squeezed out of the sample holder during swelling and to ensure repeatability of tests. As a result of these preliminary investigations, very thin (4 mm) cylindrical samples had to be prepared and used for all tests. Swelling tests have been conducted with different vertical stresses. For a vertical stress smaller than the swelling pressure, swelling occurred. Whereas the sample is compressed for a vertical stress above the swelling pressure. The swelling pressure of the compressed sample was then determined by fitting the strain at equilibrium versus the applied vertical stress. These results were used to quantify the swelling pressure and to set the initial conditions for the implementation of shear tests.
The testing program for the oedometer tests was developed through a series of pre-tests using different sample holders, disc dimensions and pre-conditioning procedures. The final testing procedure includes short-term tests over 48 hours and additional long-term tests over several days up to a week, which were then used for the final analysis of the swelling behavior. Three repetitions of short-term-tests were conducted to ensure repeatability of test results.
Shear tests were conducted using the direct shear module for circular samples of the loading frame also used for the oedometer tests. Three different normal stresses have been applied, namely 500, 1000 and 2000 kPa. The samples were preconditioned to establish a consistent hydration state for all tests, by applying a stress of 1000 kPa to all samples. Under this vertical stress, the sample swells until reaching equilibrium. The strain at equilibrium finally defines the hydration state of the sample.
After this preconditioning, a new stress state was adjusted and shearing initiated. No new stress condition was applied in the case of a normal stress during shearing at 1000 kPa. Changing the stress level after swelling results in further swelling if the stress level is reduced, or compression when the stress level is increased, which can alter the effective stress during shearing. However, shearing is induced at a fast rate (0.2 mm/min) to ensure the implementation of undrained conditions, i.e. shearing at constant volume within the shear zone. The influence of the overlapping processes of shearing and vertical deformation in the form of swelling and compression, especially in terms of the impact on effective stresses, is subject to future investigations. However, the primary aim of this procedure was to conduct tests at constant hydration states, which corresponds to constant densities. This was only possible by starting the shear tests immediately after changing the normal stress without awaiting a new equilibration. The hydration state did not change significantly during shearing, since shearing takes place faster than swelling or compression occurs.
The mechanical dislodgement tests were conducted using an Instron 250 kN Electro-Mechanical self-reacting loading frame (
Each test was deemed to be complete at the point where the applied load caused the plug inside the pipe to be dislodged and move continuously. This is observable from the Load/Displacement curve, which could be monitored in real time during each test. The raw data to generate a Load/Displacement Curve was recorded for each sample. The sampling rate was 10 Hz and the loading rate was defined to be constant at 2 kN/min.
Bulk mineralogy of the bentonite samples was determined by powder diffraction measurements on randomly oriented samples using an internal corundum standard (Al2O3) at 10 weight % for a quantitative phase analysis. An example for the powder diffraction of bulk bentonite samples is shown in
The XRD measurements confirm the existence of montmorillonite, quartz and traces of kaolinite. Traces of calcite and clinoptilolite, a zeolite mineral, were also identified. The average bulk mineralogical composition of the raw bentonite is shown in
An example for a clay diffraction pattern (oriented specimens < 2 µm) is shown in
The position of the (001) reflections of the montmorillonite of the air-dried sample (dark blue curve in
The (001) reflections of the montmorillonite in the air-dried samples show asymmetric shoulders (
Mineral content | Average weight % ± 2σ |
---|---|
Quartz Kaolinite Plagioclase (Albit, Ca) Calcite Clinoptilolite Montmorillonite | 10.8 ± 0.8 3.4 ± 0.5 6.5 ± 1.1 0.3 ± 0.2 0.7 ± 0.3 78.3 ± 1.3 |
The <2 µm fraction of the Gurulmundi bentonite consists dominantly out of Na-montmorillonite. The composition of the montmorillonite is remarkably consistent over the full range of the tested material (
As a result of the oedometer tests, vertical strains were observed for different normal stresses. Expansion occurred for vertical stresses below the swelling pressure expansion, while compression was caused with stresses larger than the swelling pressure.
The results illustrated in
・ The PVP treated bentonite samples show up to 20% reduced swelling than raw bentonite samples;
・ Swelling of samples compressed with 18.3 tons is up to 30% larger than for samples compressed with 9.2 tons; and
・ Swelling of the PVP treated samples is faster at the beginning of the test. It takes longer for the raw bentonite sample to reach comparable strains. A constant strain value is reached for all samples after a similar hydration time.
The strain monotonically increased eventually reaching positive values for higher normal stresses, as expected. Only the raw sample compressed at 18.3t (solid red curve) showed a decrease in the strain for the highest applied normal stress. Even for another repeated test, this result was confirmed. It is not entirely clear why the strain reduced for a higher stress for this sample, and it was not further investigated, since the aim to quantify the swelling pressure was reached.
As can be seen from
The summary of all shear strengths derived from the shear tests is shown in
From the results in
・ PVP treated samples show a much larger frictional behavior than raw bentonite samples. The friction angle varies between 6.0˚ for 9.2t compressed
samples and 6.7˚ for 18.3 t compressed samples. The friction angles for raw bentonite samples are much smaller, with 1.4˚ for 9.2t and 1.8˚ for 18.3t compressed samples.
・ Samples compressed with 18.3 t show a larger shear strength than samples compressed with 9.2 t, regardless of the material composition. The difference in shear strength is larger for PVP treated samples than for raw bentonite samples. The difference is visible in friction angle and cohesion, though the influence on cohesion is significantly larger.
・ The coefficient of internal friction µ varies from 0.024 (9.2 t) to 0.031 (18.3 t) for the raw bentonite and from 0.105 (9.2 t) to 0.117 (18.3 t) for the polymer treated samples.
One needs to keep in mind that no measurements have been made for stresses smaller than 500 kPa while analysing the results shown in
The experiments using the Instron loading frame for dislodgement of hydrated bentonite plugs showed gradual failure of the stacked plugs within the tested pipe section. The plugs were dislodged after 42 days of hydration. Summary graphs for Load/Displacement curves are shown in
Some tests were interrupted due to technical issues. Additional continuation measurements were run when these tests were recommenced after failure of the apparatus (e.g. #6 cont in
Load/Time graphs were used to identify the maximum force at peak failure (compare
These tests showed for the first time that all stacked bentonite plugs are failing gradually. The final dislodgment of the bentonite/PVP blend happened on average 75 seconds later than the raw bentonite samples (
Each of the red marked areas in the sketch on the left-hand side of
in
This process is repeated with the third plug until final failure of the complete set of plugs occurs and they are pushed out of the casing (yellow arrow in
This observation represents major progress in understanding of the failure process when compared with the results derived from earlier experiments using hydraulic presses equipped with a digital pressure gauge (e.g. Towler et al., 2015).
These tests using the Instron loading frame have allowed the characterisation of the detailed dislodgement process for the first time. It is planned to extend these experiments to different casing sizes and tubing with different internal roughness (corrosion states).
An additional outcome of the experiment was the proof of different dislodgement pressure regimes for the materials used. The bentonite/PVP blend showed dislodgement pressure gradients of 599.5 kPa/m ± 29.5 (26.5 psi/ft ± 1.4) after 42 days of hydration, whereas the raw bentonite only went up to a gradient of 400.4 kPa/m ± 43 (17.7 psi/ft ± 1.9). This accords to 6 MPa (870 psi) for the bentonite/PVP mix or 4 MPa (580 psi) for the raw bentonite using a 10 m plug section. The tests show that 33% higher pressures are needed for dislodgment of the bentonite/PVP blend plugs than those composed of raw bentonite.
These observations seem to be in conflict with test results from the oedometer measurements, which showed that the bentonite/PVP material show much reduced swelling when compared with the raw bentonite samples. However, it reflects as well the fact that PVP treated samples have a more pronounced frictional behaviour than samples made of raw bentonite.
An increase in swelling does not necessarily result in an increased shear strength. The shear strength is composed of cohesion and friction. Cohesion always exists while friction depends on the swelling pressure. Friction angle and cohesion are both influenced by the sample structure formed by the clay minerals and influenced by additives such as PVP. The influence of the hydration and the hydration history, especially on the shear strength of bentonite is subject to future experiments. They will help to reveal the influence of PVP in the hydro-mechanical behaviour of treated and untreated bentonite in terms of swelling and shear strength.
Standard geotechnical investigations have been conducted with raw and PVP treated bentonite to investigate the swelling and shear strength behaviour of plugs compressed at two different weights for assessing their applicability for the abandonment of wells. Furthermore, dislodgement tests with several hydrated plugs inserted in steel casings have been conducted for characterising the failure process of a short section of an abandonment system. A clay mineralogical investigation gave insight into the composition and nature of the bentonite used in this study. The results of the presented study can be summarised as follows:
・ In terms of the swelling behaviour, raw bentonite shows larger swelling than PVP treated samples with the largest swelling occurring for samples compressed with the larger weight. The swelling pressure of both, raw and PVP treated bentonite, varied only insignificantly between 1580 kPa and 1740 kPa.
・ With regards to the shear strength measured in the direct shear tests, PVP treated samples have revealed a more pronounced frictional behavior compared to raw bentonite. The friction angle of the PVP treated samples is nearly four times larger than the friction angle of the raw bentonite samples. It should be noted that the cohesion of raw bentonite shows larger values than for PVP treated samples. However, as no tests have been conducted at lower stresses, these values might be falsified and are therefore not discussed in detail. As expected, the samples compressed at larger weight always showed the higher shear strength.
・ The dislodgement tests of an abandonment segment consisting of three hydrated plugs revealed for the first time the cascaded failure mechanism of a series of plugs. This mechanism is characterised by the compaction of the soft material between the plugs and the failure of individual plugs within the chain of plugs. PVP treated samples revealed 33% higher pressures required to bring the system to failure than raw samples, which confirm the results of the shear strength investigations, but to some degree contradict the expectations from the swelling tests.
These investigations represent a first step into improving not only our understanding of the coupled hydro-mechanical behaviour of PVP treated bentonite, but will also help to better design more efficient abandonment systems based on compressed bentonite plugs.
Additional investigations were recently conducted to test dislodgment behavior using a well bore simulator. Plugs will be hydraulically dislodged using water pressures up to 1000 psi. The simulator is also designed to perform dislodgment experiments using gas pressure to mimic downhole conditions of Coal Seam Gas Wells.
Modifications of the geotechnical tests are in preparation to improve the accuracy of the parameterization of the compressed bentonite. Observations of the pore water pressure during shear tests will allow measurement of the effective stresses and will create therefore more confidence in the resulting shear strength parameters. Furthermore, the swelling pressure will be directly measured using adjusted oedometer molds.
This research was funded by The University of Queensland Centre for Coal Seam Gas (UQ-CCSG) and its industry members (Arrow Energy, Australia Pacific LNG, Santos and Shell/QGC) and the Queensland Government Advance Queensland Innovation Partnerships program.
Holl, H.-G. and Scheuermann, A. (2018) Characterisation of Geomechanical Properties of Bentonite Clay Used for Plug and Abandonment Operations of Coal Seam Gas Wells. Journal of Minerals and Materials Characterization and Engineering, 6, 218-234. https://doi.org/10.4236/jmmce.2018.62016