Home >Documents >FINAL REPORT INTERNATIONAL ECBM … II... · swelling coefficients for the three gases ... Nitrogen...

FINAL REPORT INTERNATIONAL ECBM … II... · swelling coefficients for the three gases ... Nitrogen...

Date post:22-Mar-2018
Category:
View:214 times
Download:2 times
Share this document with a friend
Transcript:
  • FINAL REPORT

    INTERNATIONAL ECBM SEQUESTRATION CONSORTUIM TASK 2: LABORATORY CORE-FLOOD EXPERIMENTS

    Prepared by:

    Satya Harpalani Southern Illinois University Carbondale, Illinois 62901

    Submitted to:

    Advanced Resources International, Inc. Houston, Texas 77077

    February 2008

  • i

    TABLE OF CONTENTS

    EXECUTIVE SUMMARY ... ii 1. Introduction and Background 1 2. Objectives....... 1 3. Experimental Work ....... 2 3.1 Matrix Swelling/Shrinkage Experiments... 2

    3.2 Permeability Experiments . 4 4. Results and Discussion . 7 4.1 Matrix Swelling/Shrinkage ... 7

    4.1.1 Illinois Coal . 7 4.1.2 San Juan Coal .. 12 4.2 Coal Permeability Measurement . 19 5. Conclusions and Recommendations 27 REFERENCES .... 29 APPENDIX A: Report - CO2 Effect on Coal Strength, Implications for the Field, and Basis for Further Work ... 30

  • ii

    EXECUTIVE SUMMARY

    This research study was a collaborative effort between the US Department of Energy (DOE), Advanced Resources International (ARI), Southern Illinois University (SIU), University of Oklahoma, Oak Ridge National Laboratory (ORNL), and a consortium of organizations currently considering CO2 sequestration in deep coals, while enhancing the production of coalbed methane (CBM). An important DOE program goal for CO2 sequestration in unmineable coals is to perform enhanced coalbed methane recovery (ECBM) field demonstration whereby injectivity is maintained at 90% (or greater) of its initial value. In order to achieve this objective, it is critical to understand the response of coal permeability to CO2 injection, making it an obvious issue to address.

    As a part of this overall project, SIU performed methane production and CO2/N2 flood experiments in the laboratory to understand the sorption induced coal matrix strain behavior and the associated permeability changes. The objective of this part of the study was to determine the permeability of coal, its variation as a function of gas composition, and its matrix swelling/shrinkage characteristics when exposed to CO2/N2. Therefore, the study was divided into two tasks. First, the matrix shrinkage/swelling characteristics were determined to establish a correlation between matrix strain and the resultant permeability decay/enhancement. Second, a set of experiments was conducted by flooding methane saturated coal cores with CO2/N2 in order to determine the variation in coal permeability.

    The first task of the study involving measurement of the matrix shrinkage/ swelling characteristics of coal was conducted using core taken from Illinois and San Juan Basins. The matrix, or grain, compressibility was measured using helium. The samples were then subjected to increasing gas pressure, in steps of 200 psi, up to 800 psi for Illinois Basin sample and 1000 psi for San Juan Basin sample. For each basin, two samples were flooded with methane, one with nitrogen, and one with CO2. Matrix swelling coefficients for the three gases were calculated. The swelling was found to be highest in the case of CO2, followed by methane, and lowest for nitrogen, which is expected. The relative strain, that is, the strain ratio for CO2/methane was calculated to be between 2 to 3.5, and ~0.33 for nitrogen/methane. The measured volumetric strains were also fitted to the Langmuir-type model, thus allowing calculation of the sorption-induced volumetric strain at any gas pressure. The modeled volumetric strain matched very well with the measured values, suggesting that the sorption-induced strain is proportional to the amount of gas sorbed. However, the results showed that the sorption-induced strain is dependent not only on the volume of gas adsorbed but the gas type as well. In the case of CO2, after a certain amount of sorption, the induced strain was significantly higher than that given using linear dependence.

    Following the adsorption part of the experiment, the methane saturated samples were subjected to CO2/N2 flooding, replicating the two ECBM alternatives. The results showed that the volumetric strain in the sample subjected to N2 flooding was negative, whereas the sample subjected to CO2 flooding exhibited positive strain. Once again, this is expected due to the higher relative affinity of CO2 for coal, and lower for nitrogen.

    For the permeability task, a flow measurement experimental setup was designed with capability of independent control of stress conditions, gas pressure, strain, and

  • iii

    simultaneous measurement of gas flowrate. The unique feature of this setup was its ability to measure and control the horizontal strain caused due to swelling/shrinkage of the coal with injection of different gases. To replicate the conditions in situ, the sample was held under uniaxial strain, that is, no horizontal strain was permitted during the experiment, after application of mechanical stresses and saturation of the sample with methane; instead, the confining stress was adjusted to ensure zero strain. At constant conditions of strain, pressure and temperature, permeability enhancement or damage was estimated by subsequently switching the gas from methane to CO2, for ECBM coupled with sequestration, and N2 for ECBM option alone. Core of coal was first stressed vertically and horizontally, saturated with methane, and the horizontal strain was set to zero after achieving equilibrium. CO2 was then injected gradually and equilibrium was attained. As expected, the matrix swelling due to adsorption of CO2 resulted in positive horizontal strain. Effort was made to increase the horizontal stress to achieve the desired zero strain condition, but this became impossible because the excess stress required to maintain the uniaxial strain condition was found to be very high, as much as five times the gas pressure responsible for the strain. Such a large increase in the horizontal stress was simply not possible without sample failure. Hence, for subsequent work, nitrogen was injected to replace methane. As expected, this resulted in negative strain requiring substantial reduction in the stress level, clearly suggesting that nitrogen injection increases the permeability significantly. .

    Four experiments were carried out although the desired sequence could not be achieved for any of them. The behavior of the sample, qualitatively speaking, was as expected. With mechanical loading, there was a negative strain due to compression of the sample. With injection of nitrogen, there was positive strain due to ballooning of the sample as well as some sorption of gas. With injection of methane at same pressure, there was positive strain due to the sorption-induced swelling, which increased significantly when CO2 was injected. Methane permeability was measured and termed the base permeability. The gas was then switched to nitrogen. Again it was difficult to maintain the desired uniaxial strain condition because the required stress reduction was so large that it resulted in a complete loss of horizontal stress. However, using theoretical stress-dependent permeability relationship established in the past, the permeability of the sample to nitrogen was estimated to be >100 times the base methane permeability. The last phase of the experiment using CO2 flooding never succeeded. However, reduction in permeability even without satisfying the uniaxial strain condition was ten-fold, suggesting that the excess stress would have reduced it significantly further.

    Two critical pieces of information became available at the end of the study. First, the excess and reduced stresses associated with injection of CO2/N2 are undoubtedly large. Also, the response of coal permeability when N2 is injected is significantly positive. When CO2 is injected, the response might not necessarily be negative. It might, in fact, result in excess stresses, sufficient to cause microfracturing and increased permeability. Second, the differential swelling, a term that refers to the fact that different gases have different swelling characteristics, with important implications for the ECBM options became available for CO2/methane and N2/methane.

  • 1

    1. INTRODUCTION AND BACKGROUND

    The enhanced coalbed methane (ECBM) technique involves injecting a second gas into a coal reservoir in order to improve the recovery of methane over the primary pressure depletion method. Two techniques have been tried to achieve the enhancement of CBM recovery. The first technique involves injecting nitrogen into a CBM reservoir. Nitrogen flushes the gaseous methane from the cleats, thus creating a disequilibrium condition in a system containing both methane and nitrogen. As a result, methane desorbs and is drawn into the gaseous phase to achieve equilibrium partial pressure [1, 2]. The second technique involves injection of CO2, which gets preferentially adsorbed onto coal, thus displacing methane [3]. Injection of either gas improves the production rate and ultimate recovery of methane substantially. However, it is also well accepted that gas flow behavior of coal varies significantly when the flowing gas ad/de sorbs on coal due to changes induced in its microstructure. It is, therefore, important that the flow assessment be performed at in situ conditions, where coal can not swell or shrink due to lateral confinement.

    This study, in its entirety, was based primarily on the finding at the four pilots, one each in US, Canada, Poland and Japan, where CO2 was, or is being, injected in coals. A significant loss of initial permeability has been reported at all sites. This led the US DOE to focus its research at maintaining CO2 injectivity, as coal adsorbs CO2 and swells, at 90% of its initial value. This research study was, therefore, aimed at developing an integrated framework to accurately predict how coal permeability and injectivity change with CO2 (or any other gas) injection, and that correctly accounts for multi-component (CH4-CO2-N2) matrix shrinkage/swelling, competitive adsorption, bi-direction diffusion, and system pressure-volume-temperature (PVT) behavior.

    The part of the project assigned to Southern Illinois University (SIU) was aimed at developing a good understanding of the changes in the physical structure of coal associated with release of methane, retention of CO2 or N2, and the impact of coal matrix volumetric strain on permeability. Hence, this report is limited to the study completed at SIU in order to: 1) determine the permeability of coal, and its variation as a function of gas composition (CH4-CO2/N2) under zero horizontal strain conditions, and 2) estimate the volumetric strain associated with CO2 and N2 injection. 2. OBJECTIVES Overall Research Objective: The overall objective of this project was to develop a means to accurately predict how coal permeability and injectivity change with CO2 (or any other gas) injection, correctly accounting for flow behavior in a multi-gas environment, and make a recommendation regarding the best strategy and potential location to conduct the field demonstration of CO2 injection in coal. Specific Objectives of this Study: In order to achieve the overall objective, the following specific objectives were pursued at SIU during the project period:

  • 2

    I. Determine the effect of individual gases (CH4/CO2/N2) on the physical structure of coal and how it varies with injection of a second gas, that is, the swelling and shrinkage associated with sorption of the second gas.

    II. Estimate the permeability of coal and its variation as a function of gas composition (CH4/CO2/N2) during enhanced coalbed methane recovery (ECBM) using inert gas stripping, or CO2 injection/sequestration.

    In order to achieve these two objectives, the work was divided into two tasks: TASK I: Coal Matrix Swelling/Shrinkage Compressibility Measurement: The purpose of these tests was to determine the swelling/shrinkage of coal with CO2/N2 injection and estimate the relative matrix strain when methane is displaced by these gases. Again, two different ECBM alternatives (CO2/N2 -ECBM) were replicated. The sorption-induced volumetric strain was measured to determine the effect of gas injection on coal matrix strain, and using these strains, the swelling/shrinkage coefficients for two coal types were estimated, along with relative strains for CO2/methane and N2/methane. TASK II: Coal Permeability Measurement: This task included developing a new type of experiment in order to flood methane-saturated coal cores with CO2/N2, while simultaneously monitor and control the shrinkage/swelling of the sample, and measure coal permeability. Importantly, ideal in situ conditions were replicated by performing the experiments under uniaxial strain conditions, that is, the core was not permitted to physically expand/shrink, just like the conditions are in situ due to lateral confinement. Rather, the stresses on the coal were increased/decreased as it tried to swell/shrink, thus resulting in permeability changes. 3. EXPERIMENTAL WORK 3.1 Matrix Swelling/Shrinkage Experiments Experimental Setup: The setup for measurement of matrix shrinkage/swelling and differential swelling consisted of four high pressure vessels, a gas chromatograph to measure the molar gas composition at equilibrium, pressure monitoring and recording system for each vessel, and a data acquisition system (DAS) to measure and record the sorption-induced strain. To ensure that the temperature over the entire duration of the experiment remained constant, the high pressure vessels were kept in a large constant temperature bath. A schematic of experimental setup is shown in Figure 1. Sample Preparation: Two sets of shrinkage/swelling experiments were performed, one using core taken from Illinois Basin and the other from San Juan Basin. Four samples were prepared for each complete set by splitting a core in to four quadrants. All four were used simultaneously in the four separate high pressure vessels.

    Three strain gauges were affixed on each sample in order to monitor strains in the

    three orthogonal directions. Figure 2 shows a typical sample schematically and

  • 3

    pictorially. The twelve strain gauges and four pressure transducers were connected to the data acquisition system. Gas Pressure and strain were monitored and recorded continuously. The algebraic sum of the three strains for each test sample gave the volumetric strain for the sample. The sum of the two horizontal strains (x and y) gave the horizontal strain since this is what truly affects the cleat aperture.

    Figure 1: Experimental setup for measurement of coal matrix shrinkage/swelling.

    Experimental Procedure: For each set of experiments, the samples were first subjected to increasing helium pressure. Helium being non-adsorptive, the measured volumetric strain was purely due to mechanical compression of solid coal resulting from changes in the external pressure. The measured volumetric strain was used to calculate the matrix, or grain, compressibility of the samples. After completing this, helium was bled out from the sample containers. Two samples were then subjected to step-wise flooding with methane, one with nitrogen, and one with CO2. Gas pressure in all four sample containers was increased in steps of ~200 psi to the final pressure, which was 800 psi for the Illinois sample and 1000 psi for the sample from San Juan Basin. For the experiment using the San Juan core, maximum pressure for CO2 injection was 955 psi due to the limitation posed by gas source. Gas injection between pressure steps was performed only after attaining strain equilibrium. Using the measured strains, the swelling coefficients for methane, N2 and CO2, and relative swelling were calculated. At the end of this part of the experiment, two samples were saturated with methane, one with nitrogen and one with CO2.

    Strain Gauge

    Pressure Gauge

    Coal Sample

    Pressure Cell

    Strain Data Acquisition System

    1 32 4

    GC

  • 4

    Figure 2: Schematic and pictorial diagrams of the sample for volumetric strain.

    Following the adsorption part, the two methane saturated samples were selected to

    replicate the two ECBM alternatives, that is, inert gas (N2) stripping and CO2 injection. In the first sample, methane was gradually replaced with N2, keeping the total gas pressure constant while, in the second sample, methane was replaced by CO2. Again, the gas switching, that is, decrease in methane concentration and simultaneous increase in N2/CO2 concentration, was performed by injecting the gas in steps. At the end of each step and prior to further injection, that is, after attaining strain equilibrium, a sample of gas mixture was taken from each vessel and analyzed using a gas chromatograph (GC) to determine the concentrations of methane and N2/CO2. Using the measured concentrations, the partial pressures of the components were calculated. The procedure was continued until the gas within, and surrounding the sample was pure N2 and CO2.

    3.2 Permeability Experiments Sample Preparation: For permeability experiments, blocks of coal were obtained from Herrin seam in the Illinois Basin. Cylindrical samples, two inches in diameter and three to four inches in length, were prepared using standard rock mechanics specimen preparation techniques. The two end surfaces of each specimen were polished to enable proper placement in the triaxial cell. Cores and samples were preserved in their native state to prevent any damage due to weathering, by storing them in an environmental chamber with no source of light and under controlled conditions of temperature and humidity. Experimental Setup and Procedure: In order to replicate the conditions in situ, controlling and monitoring of the external stress conditions and gas pressure are critical.

    1

    3

    2

    1 2

    3

  • 5

    The experimental setup for permeability measurement, therefore, included independent control of stress and strain conditions, gas pressure (upstream and downstream), and measurement of gas flowrate. The unique feature of the experimental setup was that it replicated the best in situ conditions under zero horizontal strain condition, that is, the core was not permitted to physically expand/shrink, just like it cannot in situ, due to lateral confinement. Instead, the horizontal stress was changed every time the sample started to swell/shrink. The setup consisted of a triaxial cell, a circumferential extensometer to monitor and control the shrinkage and swelling of core, a loading system, and a means to monitor and measure flowrate. A schematic of the experimental setup is shown in Figure 3. The setup enabled applying both confining and axial stresses initially to simulate the conditions in situ. The axial stress was applied by placing the cell in a universal testing machine (UTM). The triaxial cell was connected to the hydraulic system of the UTM to provide the confining stress. The perforated steel disks and porous metal plates were placed at both ends of the sample to distribute (and collect) the gas and prevent small particles from entering the tubing. The temperature of the triaxial cell was kept constant using a heating tape and temperature controller. The gas containers at the inlet and outlet were placed in a water bath, set at the same temperature as that of the triaxial cell. The gases entering and coming out of the triaxial cell were, therefore, at the same temperature. To keep the downstream pressure constant, a relief valve was used.

    Figure 3: Schematic of the permeability setup.

    1 Shrinkage Tubing 2 Triaxial Cell 3 Heating Tape 4 Insulation Pad 5 Pressure Transducer 6 Relief Valve 7 Constant Temperature 8 Humidifier Water bath 10 Regulator 9 Gas Cylinder 12 Perforated Disc 11 Hydraulic System 14 Circumferential 13 Porous Metal Plate Extensometer

    Gas

    Axial

    Confining

  • 6

    Figure 4: Sealing of core, perforated steel disks, porous metal plate with shrinkage

    tubing, extensometer attachment and placement in triaxial cell.

    Cores used for the permeability experiments were taken from a depth of 880 feet, with the gas pressure estimated to be ~500 psi. The reservoir temperature at the depth was estimated to be 72oF. Hence, tests were carried out at these conditions of temperature, maximum gas pressure, and external stresses. Prior to testing, the core, along with the perforated steel disks and porous metal plate, was sealed using PVC shrinkage tubing in order to avoid seepage of oil into the core during application of stress. The circumferential extensometer was placed around the core and PVC shrinkage tubing. The step-wise procedure is shown pictorially in Figure 4. The core setup was placed in the triaxial cell, which in turn, was placed in the load frame. The sample was then stressed triaxially.

    After attaining stress equilibrium, the core sample was saturated with methane at the desired pressure by injecting gas through the inlet and keeping the outlet closed. The

  • 7

    circumferential strain was set to zero when the pressure and strain equilibrium were attained. From this point on, no horizontal strain was allowed during the experiment. Swelling/shrinkage due to switching of gases was prevented by varying the confining stress appropriately. For each gas, when equilibrium was attained, a pressure gradient of 40-60 psi was applied across the sample using the relief valve. The pressure, temperature, and flowrate at both ends (upstream and downstream) were monitored continuously. Using the measured flowrate, permeability of the sample for the gas was calculated. The testing procedure was the same for CO2 and N2. Since it took several days to attain equilibrium for each pressure step, a complete permeability test required approximately two months. 4. RESULTS AND DISCUSSION 4.1 Matrix Swelling/Shrinkage 4.1.1 Illinois Coal Helium Injection Results: The first part of the experimental work involved dosing all four samples with helium to a pressure of ~1000 psi while monitoring the strain continuously. As expected, the volume of coal matrix decreased with helium injection due to compression of the solid coal. The results for three samples are shown in Figure 5. The matrix, or grain, compressibility, defined as the fractional change in the volume of solid coal per unit change in pressure is given as:

    =

    dPmdV.

    mV1

    mC (1)

    where, dP is the change in applied pressure of a non-sorbing gas at both the external and internal surfaces of coal matrix. The average strain was measured and Cm was calculated to be 0.80 E-06 psi-1. Sorption Induced Matrix Swelling with CH4/N2/CO2: In sorptive environment, there is an additional sorption-induced matrix swelling phenomenon accompanying the flow of gas. Hence, an additional term, matrix shrinkage coefficient ( *mC ), is defined as follows:

    =

    dPmdV.

    mV1*

    mC (2)

    where, dP is the change in pressure of the sorbing gas.

    Figure 6 shows the volumetric strain induced in all four samples with injection of the three gases. Figure 7 shows the strain as a function of gas pressure. For pressure up to 800 psi, the volume of coal matrix increased by ~0.18% with N2 flooding, ~0.58% with methane, and ~2.0% with CO2 flooding. Using the measured strain for the three gases, matrix swelling coefficients for these gases were estimated using equation (2).

  • 8

    -0.0010

    -0.0008

    -0.0006

    -0.0004

    -0.0002

    0.00000 200 400 600 800 1000 1200

    Gas Pressure (psi)V

    olum

    etri

    c St

    rain

    (V

    /V)

    .

    Sample 1Sample 2Sample 3

    Figure 5: Volumetric strain with helium injection for Illinois samples.

    0.000

    0.005

    0.010

    0.015

    0.020

    0.025

    0 5 10 15 20 25

    Time (days)

    Vol

    umet

    ric

    Str

    ain

    (V

    /V)

    .

    Sample 1 (Nitrogen)Sample 2 (Methane)Sample 3 Sample 4 (Methane)

    (CO2)

    Figure 6: Volumetric strain induced data with injection of methane, nitrogen and CO2.

  • 9

    0.0000

    0.0050

    0.0100

    0.0150

    0.0200

    0.0250

    0 200 400 600 800

    Gas Pressure, psi

    Vol

    umet

    ric

    Stra

    in (

    V/V

    )

    .

    .

    Sample 1 (Nitrogen)Sample 2 (Methane)Sample 3 (COSample 4 (Methane)

    2)

    Figure 7: Volumetric swelling of coal samples with increase in gas pressure

    Illinois sample.

    As expected, the value of *mC was the highest for CO2 (26.1E-06 psi-1), followed

    by methane (7.3E-06 psi-1) and then N2 (2.3E-06 psi-1). Hence, the swelling coefficient for methane injection was ~10 times the calculated grain compressibility, suggesting that grain expansion with methane depletion or CO2 injection cannot have any significant impact on the overall strain. The two differential swelling coefficients at 800 psi, one for CO2/methane and the other for N2/methane, were calculated to be 3.56 and 0.32 respectively. It should be noted that these values are based on the assumption that the strain is linearly dependent on gas pressure, and it is apparent from the results that this is not the case. However, since this is how this parameter is used in the simulator, it was calculated as a first step. For the sake of completeness, more precise calculation of the pressure-strain dependence was also carried out although, at this time, this has no practical application.

    The sorption data for the coal type was measured in the laboratory for methane, nitrogen and CO2. Table 1 gives the value of Langmuir sorption constants (PL and VL) for the three gases. The measured sorption isotherms, along with the ones obtained using the Langmuir model, are shown in Figure 8. Excellent agreement between the two is apparent for methane and nitrogen. However, modeled data did not fit that well for CO2.

  • 10

    Table 1: Langmuir isotherm data for Illinois coal.

    Gas PL (psi) VL (scft) Nitrogen 1265 210 Methane 475 373

    CO2 126 781

    0

    100

    200

    300

    400

    500

    600

    700

    800

    0 200 400 600 800 1000 1200

    Pressure, psia

    Vol

    ume

    Ads

    orbe

    d, sc

    ft

    .

    MethaneCO2N2

    Figure 8: Experimental and modeled sorption isotherm data for Illinois samples. The strain results shown in Figure 7 clearly exhibit a strong similarity with typical

    sorption isotherms. Hence, the measured volumetric strains were fitted to a model similar to the Langmuir sorption model, given as:

    cPP

    Pc+

    = (3)

    where, is induced volumetric strain at gas pressure P, and c and Pc are the two Langmuir-type constants. This equation allows calculation of the sorption-induced volumetric strain at any gas pressure. The measured volumetric strains for all four samples were, therefore, processed in a way similar to that used to obtain the values of PL and VL using sorption data. The values of Pc and c for the three gases were given in Table 2. Using the values given in Table 2, the modeled results are plotted, along with the

  • 11

    measured volumetric strains. The two are shown in Figure 9. The plot shows excellent agreement between the modeled strain and experimental data, suggesting direct dependence of sorption-induced strain on the amount of gas sorbed. The swelling coefficient, which is a measure of change in strain with change in pressure, is then given as:

    2c

    cc

    )PP(P

    dPd

    += (4)

    Table 2: Langmuir-type strain constants for Illinois coal.

    Gas Pc (psi) c (psi-1) Nitrogen 1294 0.00496 Methane 583 0.01005

    CO2 296 0.02838

    0.000

    0.005

    0.010

    0.015

    0.020

    0.025

    0 200 400 600 800

    Pressure, psi

    Vol

    umet

    ric

    Stra

    in (

    V/V

    )

    .

    Sample 1 (Nitrogen) Expt. Data Sample 2 (Methane) Expt. DataSample 3 (CO ) Expt. Data Sample 4 (Methane) Expt. Data2

    Modeled

    Figure 9: Measured and modeled volumetric strain for Illinois coal. Injection of N2/CO2: Following the adsorption part of the experiment, one of the methane saturated samples was subjected to CO2 flooding and the other to nitrogen, replicating the two ECBM alternatives, where injection is carried out from the commencement of the operation at initial in situ pressure. The resulting volumetric strain with decrease in methane partial pressure is shown in Figure 10. The sample subjected to nitrogen injection showed a negative strain. It underwent less shrinkage than the swelling

  • 12

    induced during adsorption of methane. This small amount of incremental swelling may be due to adsorption nitrogen at 800 psi. On the other hand, the sample subjected to CO2 injection exhibited a significant increase in strain, almost three times that induced by methane at 800 psi. This indicates that injection of CO2 can result in considerable incremental swelling whereas injection of nitrogen results in shrinking of the coal matrix.

    0.000

    0.005

    0.010

    0.015

    0.020

    0 200 400 600 800

    Methane Total/Partial Pressure, psi

    Vol

    umet

    ric

    Str

    ain

    (V

    /V)

    .

    Methane Adsorption (Sample 2)Nitrogen Injection (Sample 2)Methane Adsorption (Sample 4)CO Injection (Sample 4)2

    Figure 10: Volumetric strain with methane total/partial pressure for Illinois samples. Sorption/Volumetric Strain Relationship: As a final step, sorption-induced volumetric strain was plotted as a function of the amount of gas sorbed. This is shown in Figure 11. It can be seen from Figure 11 that, for methane and nitrogen, there is a linear relationship between volumetric strain and amount of gas sorbed. However, for CO2, the linearity exists only up to sorbed amount of ~550 scft. Above that, the plot ceases to be a straight line. This is in agreement with the findings reported by other researchers [4, 5], suggesting that the matrix strain is not dependent only on the volume of sorbed gas but on gas type as well. Following the CO2 trend, the straight line obtained for methane was extrapolated as shown in Figure 11. The hatched line at the end of the last measured data for methane indicates that the strain might increase non-linearly. Of course, there is no practical application of this extrapolation for Illinois coals since the in situ gas content in the basin is below 250 scft. 4.1.2 San Juan Sample

    The second set of experiments was carried out using samples taken from San Juan Basin. Helium was injected in all four samples at a pressure of 700 psi to determine the coal matrix, or grain, compressibility (Cm). Volumetric strain could not be monitored for one of the four samples due to leakage in the sample container. The resulting volumetric

  • 13

    0.000

    0.005

    0.010

    0.015

    0.020

    0.025

    0 100 200 300 400 500 600 700 800

    Volume Absorbed, scft

    Vol

    umet

    ric

    Str

    ain

    (V

    /V)

    .

    MethaneCONitrogen

    2

    Extrapolated (Methane)

    Figure 11: Volumetric strain as a function of sorbed gas concentration - Illinois samples.

    strain for the other three samples is shown in Figure 12. The average value of Cm for the three samples was calculated to be 0.85 E-06 psi-1, which is very close to the value obtained for the sample from Illinois Basin.

    -0.0008

    -0.0006

    -0.0004

    -0.0002

    0.00000 200 400 600 800

    Gas Pressure, psi

    Vol

    umet

    ric

    Stra

    in (

    V/V

    )

    .

    Sample 1Sample 2Sample 4

    Figure 12: Volumetric strain with helium pressure for San Juan samples.

  • 14

    After bleeding the helium out, two samples were subjected to increasing gas pressures, in steps of 200 psi, up to 1000 psi with methane and one with nitrogen. The last sample was subjected to a step-wise increase in CO2 pressure, in steps of 200 psi, but only up to a final pressure of 955 psi. The step-wise volumetric strain with gas injection over time is shown in Figure 13. The linear strains induced in the three directions (x-y-z) over time are shown for one of the samples subjected to methane injection in Figure 14. The volumetric strain as a function of increasing pressure for the three gases is shown in Figure 15. Figure 16 shows the best fit Langmuir-type modeled results, along with the measured strains, and Table 3 provides the values of the constants, Pc and c for the three gases.

    The results clearly indicate that the linear strains in the two horizontal directions

    are very close but in the vertical direction, it is higher. This was observed in other results as well. Hence, total horizontal strain (i.e., sum of x and y) was calculated for all four samples at all gas pressure steps to obtain an estimate of the horizontal strain with increasing pressure. Figure 17 shows the horizontal strain with increasing gas pressure.

    0.000

    0.003

    0.006

    0.009

    0.012

    0.015

    0 4 8 12 16

    Time (days)

    Vol

    umet

    ric

    Stra

    in (

    V/V

    )

    .

    Sample 1 (Nitrogen)Sample 2 (Methane)Sample 3 (CO )Sample 4 (Methane)

    2

    Figure 13: Volumetric strain with time for San Juan samples for various pressure steps.

  • 15

    0.0000

    0.0006

    0.0012

    0.0018

    0.0024

    0.0030

    0 4 8 12 16

    Time (Days)

    Lin

    ear

    Stra

    in ( L

    /L)

    .

    xyz

    Figure 14: Linear strain with time for methane injection in San Juan sample.

    0.000

    0.003

    0.006

    0.009

    0.012

    0.015

    0 200 400 600 800 1000

    Gas Pressure (psi)

    Vol

    umet

    ric

    Stra

    in (

    V/V

    )

    .

    Sample 1 (Nitrogen)Sample 2 (Methane)Sample 3 (CO )Sample 4 (Methane)

    2

    Figure 15: Measured volumetric strain for San Juan sample with increasing pressure.

  • 16

    0.000

    0.004

    0.008

    0.012

    0.016

    0 200 400 600 800 1000

    Methane Total/Partial Pressure (psi)

    Vol

    umet

    ric

    Stra

    in (

    V/V

    ) .

    Sample 1 (Nitrogen) - Expt. Data Sample 2 (Methane) - Expt. Data

    Sample 3 (CO ) - Expt. Data Sample 4 (Methane) - Expt. Data2

    Modeled

    Figure 16: Measured and modeled volumetric strain for San Juan samples with increasing gas pressure.

    Table 3: Langmuir-type constants for San Juan Basin samples volumetric strain.

    Gas Type Pc (psia) c Nitrogen 1226 0.0056 Methane 603 0.0107 CO2 296 0.0165

    Table 4: Langmuir-type constants for San Juan Basin samples horizontal strain.

    Gas Pc (psi) c Nitrogen 1460 0.00364 Methane 640 0.00660

    CO2 280 0.00955

    The matrix swelling coefficients for CO2, methane and nitrogen were calculated.

    As expected, the swelling coefficient was the highest for CO2 (13.1 E-06 psi-1), followed by methane (6.7 E-06 psi-1) and then nitrogen (2.5 E-06 psi-1). The differential swelling coefficients for CO2/methane and nitrogen/methane were calculated to be 1.9 and 0.37

  • 17

    0.000

    0.002

    0.004

    0.006

    0.008

    0 200 400 600 800 1000

    Gas Pressure (psi)

    Hor

    izon

    tal S

    trai

    n(H

    /H)

    .Sample 1 (Nitrogen)Sample 2 (Methane)Sample 3 (CO )Sample 4 (Methane)

    2

    Figure 17: Horizontal strain for San Juan samples with increasing gas pressure.

    respectively. The relative swelling for nitrogen/methane is close to that obtained for Illinois coal but that for CO2/methane is not.

    One of the methane saturated samples was then subjected to CO2 flooding, and

    the other to nitrogen, replicating the two ECBM alternatives practiced in the field. At the beginning of this exercise, the gas pressure within both samples was ~1000 psi. However, for the sample subjected to CO2 flooding, the gas pressure was reduced to 800 psi since the available pressure did not allow injection at pressure above 950 psi. The gas exchange exercise for both samples was performed in five steps. For each injection/ bleeding step, the volumetric strain resulting from the gas exchange and the gas composition within each pressure vessel were measured at equilibrium condition. The resulting volumetric strain as a function of methane partial pressure is shown in Figure 18. The sample subjected to nitrogen was found to undergo less shrinkage than the swelling induced during adsorption of methane, once again due to adsorption of nitrogen at 1000 psi. On the other hand, the sample subjected to CO2 injection exhibited an increase in strain. However, this incremental strain is less than that produced by pure CO2 adsorption at 800 psi due to simultaneous desorption of methane from 800 to 0 psi. Also, it was less than that produced for Illinois coal.

    Finally, sorption-induced volumetric strain was plotted as a function of the

    amount of sorbed gas. This is shown in Figure 19. For the San Juan coal, the deviation from linear dependency is apparent, confirming the extrapolation of the measured results for Illinois Basin coal.

  • 18

    0.000

    0.002

    0.004

    0.006

    0.008

    0.010

    0.012

    0 200 400 600 800 1000

    Methane Total/Partial Pressure

    Vol

    umet

    ric

    Stra

    in (

    V/V

    ) .

    Methane Adsorption (Sample 4)Methane/CO Exchange (Sample 4)Methane Adsorption (Sample 2)Methane/Nitrogen Exchange (Sample 2)

    2

    Figure 18: Volumetric strain for San Juan samples as a function of methane pressure with CO2/nitrogen injection.

    0.000

    0.001

    0.002

    0.003

    0.004

    0.005

    0.006

    0.007

    0.008

    0 100 200 300 400 500 600 700Gas Conc., scft

    Vol

    umet

    ric

    Stra

    in,

    V/V

    Methane - As Received

    Methane DAF

    Figure 19: Volumetric Strain as a function of sorbed gas concentration SJ coal.

  • 19

    4.2 Coal Permeability Measurement

    This part of the project was unique since, to the best knowledge of the project participants, similar work has not been reported prior to this investigation. Following the procedure described, a stainless steel cylinder was first used in place of the actual core to calibrate the experimental setup and understand the behavior of the hydraulic, mechanical and electronic components of the entire system. The data acquisition system was calibrated simultaneously. Trial tests were then carried out using sandstone samples due to their simple, well understood, and documented behavior. Calibration of the triaxial testing setup for its use in the permeability experiment was carried out by the MTS (manufacturer of the load frame and circumferential extensometer) personnel. The purpose of this entire exercise was to understand and fine-tune the procedure and ensure proper working of the entire setup. Based on these tests, several modifications were made to improve the setup. To gain some experience with operation of the setup prior to using valuable cores, four samples from Herrin seam in the Illinois Basin were used for permeability tests. The samples were prepared using a block of coal obtained from an underground mine and well preserved in the laboratory environmental chamber.

    For the first test using coal, the sample was subjected to axial and confining stresses of 800 and 400 psi respectively. After achieving the desired stress level, the extensometer was set to zero in order to prevent any further strain in the sample. The sample was then flooded with CO2 at ~300 psi. This test run was unsuccessful due to breakage of the coal sample being tested. This was attributed to the automated controls being used to adjust the horizontal stress to ensure that there was no horizontal strain induced in the sample. The horizontal stress probably became excessive. It was, therefore, decided that manual controls will be used to maintain the uniaxial strain conditions. The second test was started to monitor the strain induced with injection of CO2 and nitrogen. The core was subjected to axial and horizontal stresses of 1000 psi and 650 psi respectively. The circumferential extensometer was set to zero for these conditions of stress. CO2 was then injected gradually to prevent a sudden build-up of pressure in the sample, a possible rupture of the shrinkage tubing, or breakage of the sample. The CO2 pressure was allowed to reach equilibrium at ~120 psi. This resulted in a positive circumferential strain (C/C, where C is the circumference of the sample) of 0.072%. The measured strain is shown in Figure 20. This strain was the combined result of the effect of physical ballooning of cleats, and swelling of the coal matrix due to adsorption. Keeping the axial stress constant, the horizontal stress was increased gradually in an effort to bring the strain to its initial value (zero-value), that is, bring the sample back to its original diameter, and simultaneously, measure the amount of excess stress required to do so. However, after increasing the horizontal stress by 350 psi, only 50% of the induced strain was recovered. The excess stress is, therefore, significantly greater than the CO2 pressure responsible for the strain, and estimated to be as much as five times the gas pressure. Also, at this stage, the axial and horizontal stresses were almost equal (~1000 psi). Hence, any further increase in the horizontal stress was not possible without increasing the axial stress as well. This rather unexpected finding

    1

    2

    3

    4

    5

  • 20

    -0.08

    -0.06

    -0.04

    -0.02

    0

    0.02

    0.04

    0.06

    0.08

    0.1

    0 50 100 150 200 250

    Time, hours

    Cir

    cum

    fere

    ntia

    l Str

    ain

    ( C

    /C ),

    %

    h = 650 psia = 1000 psi

    CO2 injected @ 200 psi

    CO2 [email protected] 120 psi h increase initiated

    h = 1000 psiOutlet valve opened to bleed CO2

    N2 injected @150 psi

    N2 equilibrium @120 psiN2 flow initiated

    Atmospheric N2 pressure equilibrium

    h = 1000 psia = 1000 psi

    Figure 20: Circumferential strain with injection of CO2/N2.

    confirmed that the experimental conditions of uniaxial strain cannot be maintained precisely by the automated system since this would result in horizontal stresses exceeding the axial stress when bringing the strain back to its initial value, and thus sample failure. Without bringing the sample back to its original dimension, nitrogen was injected in order to determine the sample response. The downstream end was first opened to flush out the CO2 and facilitate the injection of nitrogen. The flushing out of CO2 resulted in a negative strain of 0.035%, which can be attributed to desorption of CO2 and mechanical de-ballooning of the sample. Nitrogen was then injected at 150 psi, and the horizontal strain was monitored continuously. There was a negative strain of 0.0035% immediately after injection of nitrogen. After achieving nitrogen pressure equilibrium at 120 psi, an additional negative strain of 0.013% was measured, totaling to a negative strain of 0.0165%. The downstream end of the sample was then opened to atmosphere, resulting in a through-flow of nitrogen through the sample, and this resulted in a further decline of 0.049% in the strain, bringing the total strain due to nitrogen injection down to 0.0655%. At the end of the experiment, there was an overall negative strain of -0.0545%. This is expected since the horizontal stress at the end of the experiment was 350 psi higher than the starting stress of 650 psi. The strain behavior with injection of CO2/N2 is as expected but the stresses required to maintain uniaxial strain conditions are not.

    Since the concept of excess stress is relatively new and not well understood, it was decided that the permeability changes with nitrogen injection will be evaluated first. Nitrogen, being less sorptive than methane, would require reduced stress to maintain

  • 21

    the uniaxial strain condition, and this would be easier to achieve experimentally. Only after measurement of nitrogen permeability is completed, the CO2 phase of the experiment would be started. Also, the fact that the experimental conditions were no longer representative of the in situ conditions was discussed with consortium members.

    For the third test, the initial experimental conditions were similar except for the stress levels, which were changed to ensure that the experiment could be performed successfully. A vertical stress of ~1400 psi and horizontal stress of ~800 psi were applied to the sample. After achieving the strain equilibrium due to mechanical stressing, methane was injected at a pressure of ~400 psi and a pressure gradient of 50 psi across the sample was applied. The strain induced due to methane injection is shown in Figure 21. The observed trend is very similar to that measured during previous experiments.

    After achieving methane equilibrium, permeability of coal to methane was measured at a mean pressure of ~390 psi, and a pressure gradient of ~50 psi. Three permeability values were measured and averaged. This was considered the base permeability. Methane was then replaced with N2, all other experimental conditions remaining unchanged, thus replicating the N2-ECBM conditions. Displacement of methane by N2 resulted in desorption of methane, and thus a significant decrease in the horizontal strain, clearly indicating shrinking of the sample. The sample was allowed to shrink. After achieving equilibrium, N2 permeability was measured and, once again, three different measurements were made and averaged. The permeability to nitrogen was twice that to methane. After permeability measurement, the horizontal stress was reduced to bring back the strain to its initial value. However, with a reduction in horizontal stress by 300 psi, only 35% of the strain was recovered. The horizontal stress could not be

    0.00000

    0.00005

    0.00010

    0.00015

    0.00020

    0.00025

    0.00030

    0 2000 4000 6000 8000 10000 12000

    Time, min

    Cir

    cum

    fere

    ntia

    l Str

    ain

    (C

    /C)

    . .

    Circumferential Strain

    Figure 21: Circumferential strain with methane injection at 400 psi (v ~ 1400 psi and h ~ 800 psi).

  • 22

    reduced any further as this would have resulted in gas pressure becoming higher than the horizontal stress, a condition when the measurements become unreliable, since shrinkage tubing can easily burst due to the pressure differential. Three permeability measurements for N2 to coal were made and averaged. The permeability was 11 times higher than the methane permeability measured at initial conditions, suggesting that with decrease in horizontal stress, the permeability increased by an order of magnitude even though the desired uniaxial strain condition was not maintained. Figure 22 shows the strain for the entire duration of the experiment. Prior to terminating this part of the experiment, the gas pressure was reduced to zero and the horizontal stress was decreased to estimate the magnitude of reduced stress necessary to recover the induced strain with nitrogen injection. Surprisingly, the stress had to be reduced to almost zero. Although permeability measurement at zero stress is not possible due to lack of lateral confinement, effort was made to estimate its value using theoretical stress-permeability relationship, which is fairly well understood and accepted. The relationship between stress and permeability has been reported by several researchers [6, 7, 8, 9] to be of exponential in nature, given as:

    k = koeB (5) where, ko is the permeability at zero stress, is the applied stress, and B is a constant depending on coal type. A typical relationship is shown in Figure 23. Since two measured permeability values were available, one at 800 and the other at 500 psi, permeability at zero stress was calculated using eq (5). This was found to be ~200 times the methane, or base, permeability value under initial stress conditions. Although an interesting finding, it is based on one set of experimental results, two measured permeability values, and an empirical relationship established in the past.

    The fourth experiment was started with application of vertical and horizontal stresses of ~1400 and 800 psi respectively. Mechanical equilibrium was achieved in ~55 hours. This is shown in Figure 24. This was followed by methane injection. Equilibrium in horizontal strain was established in about 37 hours. Flowrate was then measured at a mean gas pressure of ~360 psi and methane permeability was calculated. This was followed by CO2 injection at the same pressure, keeping all other experimental conditions unchanged. The switching from methane to CO2 required ~210 hours to reach equilibrium, and induced a significant increase in horizontal strain, as shown in Figure 24. After achieving equilibrium, three flowrate measurements were taken and CO2 permeability was estimated. The reduction in the permeability was almost ten-fold. Finally, the sample was subjected to gradually increasing horizontal stress to recover the strain induced due to CO2 injection, and bring it back to zero strain condition. However, the sample failed almost immediately with an increase in the horizontal stress of only ~100 psi.

  • 23

    -30

    -20

    -10

    0

    10

    20

    30

    0 100 200 300 400 500 600 700Time, hrs.

    Cir

    cum

    fere

    ntia

    l Str

    ain

    (C

    /C) x

    10-

    5

    .

    Mechanical Stress Effect

    Methane injection and Sorption Effect

    Methane Permeability

    measured Methane relapcement with Nitrogen started

    Nitrogen Permeability

    measured

    h reduced from 800 psi to 500 psi

    v = 1400 psih = 800 psi

    Nitrogen Pressure

    reduced to atmospheric

    h reduced from 500

    psi to 0 psi

    Figure 22: Circumferential strain with mechanical stressing, methane injection and methane-N2 exchange.

  • 24

    0.01

    0.1

    1

    10

    100

    1000

    200 300 400 500 600 700 800 900Hydrostatic Stress, psi

    Perm

    eabi

    lity,

    X 1

    0 -

    18 m

    2 ( X

    10

    - 3 m

    d)

    .Stressing - Methane

    Stressing - Nitrogen

    Best Fit - Methane

    Best Fit - Nitrogen

    k = koeB

    Figure 23: Typical stress-permeability relationship for coals.

    -0.0010

    -0.0006

    -0.0002

    0.0002

    0.0006

    0.0010

    0 30 60 90 120 150 180 210 240 270 300

    Time (hrs.)

    Cir

    cum

    fere

    ntia

    l Str

    ain

    (C

    /C)

    .

    Mechanical Compression

    Methane Sorption

    Methane/CO2 Exchange

    Figure 24: Circumferential strain with methane sorption and methane/CO2 exchange.

  • 25

    The last experiment was carried out using similar vertical and horizontal stress conditions as for the previous experiment. The variation in the circumferential strain for the entire duration of the experiment is shown in Figure 25. Mechanical equilibrium was achieved in ~70 hours, which was followed by nitrogen injection. Nitrogen equilibrium was established in ~190 hours. After attaining equilibrium, the flowrate was measured at a mean pressure of ~410 psi and nitrogen permeability was calculated. Methane was then injected at the same pressure, keeping all other experimental conditions unchanged. Methane equilibrium was attained in ~260 hours, inducing a horizontal strain, as expected, due to swelling of the coal matrix. The flowrate was measured and methane permeability reduction was determined to be 25%. This was followed by CO2 injection, keeping all other experimental conditions unchanged. As expected, the duration of this phase was significantly longer, ~315 hours. A significant increase in the horizontal strain was observed during this phase of the experiment. CO2 permeability was determined to be ~40% less than the methane permeability. At this point, the experiment failed. Analysis of Results Understanding Observations

    Based on the results obtained, there are several interesting findings. First, excess and reduced stresses induced when gas is switched from methane to CO2 or nitrogen are significantly higher than the gas pressure causing the matrix strain. For the tests completed, increase in horizontal stress of 350 psi recovered 30% of the strain

    -0.0010

    -0.0005

    0.0000

    0.0005

    0.0010

    0.0015

    0.0020

    0.0025

    0 100 200 300 400 500 600 700 800 900

    Time (hrs.)

    Cir

    cum

    fere

    ntia

    l Str

    ain

    (C

    /C)

    .

    Mechanical Compression

    Nitrogen Adsorption

    Methane/Nitrogen Exchange

    CO2/Methane Exchange

    Figure 25: Circumferential strain over the duration of experiment.

  • 26

    induced by switching the gas from methane to CO2 at a pressure of 500 psi. Extrapolating this suggests that ~1000 psi increase in stress would be required to maintain the uniaxial strain condition. Similarly, when nitrogen was injected, decreasing the stress by 300 psi recovered 35% of the induced strain, suggesting that it would require ~1000 psi reduction in stress to maintain the uniaxial strain condition. This was further shown by reducing the stress and pressure down to zero. Hence, there appears to be no question that nitrogen injection would increase the permeability of coal significantly due to stress reduction alone. Similarly, CO2 injection can reduce the permeability due to increased horizontal stress although it is not very clear at this time whether CO2 reduces the permeability permanently, or it is a temporary effect.

    There has been evidence presented recently that exposure of coal to CO2 alters the flow properties in a manner that is not easy to explain. Of particular interest is the finding in the field, reported by Gunter et al (10) that even though the absolute permeability of the coalbed decreases with continued injection of SAG (strongly adsorbing than methane), the SAG injectivity, surprisingly and unexpectedly, increases while injecting at pressures greater than reservoir pressure, but less than fracture pressure or fracture extension pressure. Of course, this is surprising and unexpected if one starts with the premise that the permeability of coal decreases as a result of CO2 injection. Furthermore, it was reported that CO2 injectivity at the Yubari site was higher in 2005 than it was in 2004 (11). The explanation given for this increase is the nature of stress-dependent permeability of coal. Finally, a patent, developed by AMOCO in the early nineties, stated that coal weakens when exposed to CO2 (12) although the claim is not based on scientific evidence in the opinion of the PI.

    Given the unexpected results, effort was made to understand the phenomenon of

    CO2 injection in methane saturated coal. Dr. Ian Palmer, Higgs Technologies, was contacted. The results were forwarded to him for analysis and possible recommendations for modification of the procedure for future experiments. Under a separate effort, a preliminary analysis of the results was carried out by him. The measured swelling and shrinkage of coal with injection of CO2 in a methane saturated sample were modeled successfully. However, modeling of the failure of sample with CO2 injection was not successful. The results were presented to the Consortium members and their suggestions/ recommendations sought.

    As a follow-up of one of the Consortium recommendations, two samples of coal were prepared from a block of coal. One was exposed to CO2 at 500 psi for ~100 hours and the other was not. The samples were then shipped to TerraTek for geomechanical testing uniaxial compressive strength (UCS), Youngs Modulus (E) and Poissons Ratio () to determine if exposure to CO2, in fact, weakens the coal. TerraTek reported the results but these were inconclusive because the clean sample had a major crack along its length and it failed prematurely. These tests were repeated and, prior to testing, the two cores were scanned. The results were obtained and provided to Dr. Ian Palmer, Higgs-Palmer Technologies, along with all of the supporting information, for analysis. Dr. Palmer completed the analysis. His report of the analysis completed by him is attached in Appendix A.

  • 27

    5. CONCLUSIONS AND RECOMMENDATIONS

    A major part of the effort during the project period was devoted to development of the experimental work to enable conducting permeability tests under uniaxial strain condition. Since this was the first effort ever to conduct tests of this nature, designing and modifying the experimental setup, and developing appropriate procedure, was a challenge. The conclusions are, therefore, based on the work completed under the given timeframe.

    Two categories of conclusions can be made, some that are specific to the coals

    tested and others that are general. The general conclusions are as follows: At this time, there is no certainty that flow experiments that involve switching of gas

    from methane to CO2, replicating in situ conditions, can actually be carried out successfully. The excess stresses required to maintain uniaxial strain condition might be too high to conduct these experiments. One way of getting around this complexity would be to use higher applied stresses but this would mean deviating from the desired and appropriate conditions. The positive aspect of the finding is that, depending on the depth of the coal and associated overburden stress, coal might actually develop fractures with CO2 injection, resulting in increased permeability. This would explain the improved injectivity reported at some of the ECBM pilots.

    Similarly, nitrogen injection cannot be accurately replicated in the laboratory since there may be a complete loss of horizontal stress, making it impossible to measure the flowrate through the sample.

    Nitrogen injection can result in large increases in permeability, estimated to be more than 100 times the methane permeability. This has a very positive implication for the deeper and highly stressed basins worldwide since the primary reason for lack of commercial CBM activity in these regions is low permeability. If injection of nitrogen can reduce the in situ stresses significantly, the result would be drastically enhanced permeability.

    The sorption-induced volumetric strain follows the sorption isotherm trend. It would, therefore, be more appropriate to determine the strain using a Langmuir-type model rather than a single coefficient assuming a linear dependence of strain on pressure. This would require changing the way this parameter is used when conducting simulation using commercial CBM simulators although this would probably not be a major challenge.

    Swelling of coal in the axial/vertical direction is much higher than in the horizontal direction, clearly exhibiting the anisotropic nature of swelling. There is anisotropy in the two horizontal directions as well although the difference is small. Hence, with CO2 injection, since the vertical stress is always higher than in the lateral direction, the resistance to swelling would be greater in the vertical direction. This would result in increased lateral swelling, causing more permeability damage due to a further increase in the excess stress. However, if the resulting horizontal excess stress

  • 28

    exceeds the overburden stress, the coal might actually fail in situ resulting in increased permeability.

    Preliminary conclusions specific to the coals tested are as follows:

    Swelling of coal matrix with injection of CO2 is two to three times that caused with

    sorption of methane. Similarly, shrinkage with N2 injection is one-third of that with methane. For San Juan Basin, the relative swelling/shrinkage of coal with injection of CO2/N2 in methane saturated coal is ~2 and 0.33 respectively. For Illinois Basin, the corresponding relative strains are 3.5 and 0.33. Hence, the relative shrinkage with nitrogen injection is similar for both coal types but the relative swelling with injection of CO2 is significantly different.

    Similarly, the matrix, or grain, compressibility for the two coals tested is very similar, 0.80-0.85 E-6 per psi. The matrix shrinkage coefficient for methane for the two coal types tested is also fairly close, 6.7-7.3 E-6 per psi. However, the same is not true for CO2 where, for Illinois coal, the swelling was twice that for San Juan coal.

    N2-ECBM, as a technique, has technical promise although the incentive to sequester CO2 is not there. In Illinois Basin, given that CBM contains a significant proportion of nitrogen, necessitating separation at the downstream end, N2-ECBM might be a viable option that merits further consideration.

    Based on the work completed, the following recommendations are made for continued research in this area:

    The most important recommendation for future work is to continue working on developing a technique to conduct laboratory tests to measure flow under uniaxial strain condition. It would be interesting to couple these tests with the ability to see inside the test sample to verify development of new fractures resulting from CO2 injection.

    The laboratory technique developed has excellent application in measuring the pore volume compressibility (Cp) and matrix shrinkage/swelling coefficients (Cm) under field replicated conditions, that is, under triaxial strain. There appears to be conflicting opinions about the variation in the values of Cp and Cm with depletion in CBM operations, or when a second gas is injected. Given the impact of these two parameters on CBM production and successful injection of CO2/N2, the nature of these variations should be evaluated.

    In the area of modeling, there is a need to develop a reliable model for ECBM alternatives where a second gas is injected into coal, and test it rigorously. To start with, laboratory studies should be carried out to generate the required data for model validation.

  • 29

    REFERENCES

    1. Puri, R., and Yee, D. 1990. Enhanced Coalbed Methane Recovery. Proceedings,

    65th Annual Technical Conference and Exhibition of the SPE, New Orleans, Louisiana, pp. 193-202.

    2. Reeves, S.R. 2003. Enhanced CBM Recovery, Coalbed Sequestration Assessed. Oil and Gas Journal, No. 14, pp. 49-53.

    3. Arri, L.E., et al 1992. Modeling Coalbed Methane Production with Binary Gas Sorption. SPE 24363, Proceedings, SPE Rocky Mountain Regional Meeting, Casper, Wyoming.

    4. Pekot, L. J., and Reeves, S. R. 2003. Modeling the Effects of Matrix Shrinkage and Differential Swelling on Coalbed Methane Recovery and Carbon Sequestration. Proceedings, 2003 International Coalbed Methane Symposium, Tuscaloosa, Alabama.

    5. Connell, L., 2006. An Overview of CSIRO Research on CO2 Sequestration in Coal, Coal-Seq V, Houston, November.

    6. Somerton, W. H., et al 1975. Effect of Stress on Permeability of Coal. International Journal of Rock Mechanics Mining Science and Geological Abstracts, No. 12, pp. 129-145.

    7. Harpalani, S., and McPherson, M. J. 1985. Effect of Stress on Permeability of Coal. Methane from Coal Seams Technology, Vol. 3, No. 2, pp. 23-29.

    8. Koenig, R.A. et al 1989. Application of Hydrology to Evaluation of Coalbed Methane Reservoirs. Final Report, GRI Contract No. 5087-214-1489.

    9. Puri, R., and Seidle, J. 1991. "Measurement of Stress-Dependent Permeability in Coals and its Influence on Coalbed Methane Production." Proceedings, 1991 Coalbed Methane Symposium, Tuscaloosa, Alabama, pp. 415-424.

    10. Gunter, et al, 2005. Process for Predicting Porosity and Permeability of a Coalbed, US Patent, 6,860,147.

    11. Fujioka, M., 2006. Japan CO2 Geo-sequestration in Coal Seams Project, Coal-Seq V, Houston, November.

    12. Puri et al, 1991. Method of Increasing the Permeability of a Coal Seam, US Patent 5,014,788.

  • 30

    APPENDIX A

  • 31

    CO2 Effect on Coal Strength, Implications for the Field, and Basis for Further Work Harpalanis results at Coal-Seq V in November 2006 (ref 1) showed that when Illinois basin coal core was saturated with methane in a triaxial cell, and followed by CO2 injection, while holding the sample under uniaxial strain conditions, the core failed. The experiment was repeated several times, with only minor differences. In a followup presentation, Palmer (ref 2) showed that the failure was quite unexpected, and speculated that the core was weakened by the CO2 swelling. To confirm this, it was decided to measure unconfined compressive strength (UCS) in two sister coal cores, one of which would be saturated with CO2, the other without. Two sister cores from a block of coal taken from the HVB rank Herrin seam, Illinois basin, were prepared. One was exposed to CO2 at 500 psi for 96 hours. The two cores were sent to Terra Tek, where UCS tests were conducted on each core, anticipating that the CO2 core would be weaker. The results of the first UCS testing are given in Table 1, where UCS is called effective compressive strength. The CO2 core was actually stronger, but the clean sample (ie, without CO2) was compromised by a weak plane that led to early failure. Nevertheless, before that failure, Youngs modulus was lower than in the CO2 core, and UCS and modulus generally trend together, which suggests that the CO2 core was indeed stronger. Table 1: First UCS tests on cores.

    Sample ID

    Diameter (in)

    Length (in)

    Bulk Density (g/cm3)

    Effective Compressive

    Strength (psi)

    Youngs Modulus

    (psi) Poissons

    Ratio

    241,100 0.31 Sample A (clean) 1.9945 2.81 1.294 876 129,100 0.15

    Sample B (CO2) 1.9885 2.94 1.294 3429 387,400 0.26

    In the second testing, the coal was moisture saturated prior to exposure to CO2. As Table 2 shows, the CO2 core was definitely stronger, and substantially so. Youngs modulus is also larger, which supports this observation. Images of the cores are given in Appendix A. Table 2: Second UCS tests on cores.

  • 32

    Sample ID

    Diameter (in)

    Length (in)

    Bulk Density (g/cm3)

    Effective Compressive

    Strength (psi)

    Youngs Modulus

    (psi) Poissons

    Ratio

    Sample A (CO2)

    1.9965 2.4098 1.245 5620 230,000 0.28

    Sample B (Clean) 1.9817 2.183 1.302 2550 169,500 0.43

    Possible explanations for the unexpected UCS results at TerraTek:

    1. The CO2 has to remain in the core to experience the core weakening: in this case the core was shipped to TerraTek, and then delayed at least several days before testing, so the CO2 would have escaped from the core. Based on the Thermal Cracking section below, we do not think this is the answer.

    2. Plastification of coal occurred during the CO2 saturation, and therefore weakened. We do not think this happened, because lab work indicates no plastification up to a temperature of 80F, and very high pressures (ref 3).

    3. The CO2 injection for 96 hours would have dried out the coal, and less water can mean stronger coal, since water can have the effect of lubricating the cleats and allowing them to slip more easily. In one lab measurement (ref 4), the UCS of coal was reduced from 2425 psi to 693 psi, and cleat lubrication was invoked as one explanation. Note: an opposite view has also been suggested, that removal of water causes weakening of coal as the macrostructure collapses (ref 3). This illustrates the complexity and uncertainty of this work, because it is so new. Despite this, we favor explanation 3. as the answer to the unexpected UCS results at TerraTek. That is, coal that has been saturated with CO2 in the lab (but then the CO2 removed) becomes stronger because it has been dried, and the removal of the water lubrication effect dominates the CO2 weakening due to differential expansion (which is discussed below).

    Implications for double-U coal strength guide: The two clean samples of HVB coal rank (without CO2) gave UCS values of 876 and 2550 psi. However, the 876 psi value appears to have been compromised by a pre-existing plane of weakness. The higher UCS = 2550 psi falls below that predicted by the double-U curve (see Figure 1), where UCSmin = 4800 psi. The double-U curves provide a guide to coal strength versus rank when no core tests are available. This result suggests the double-U curve can sometimes be an overestimate of UCS in cores.

  • 33

    Figure 1: Double-U curve for predicting UCS of coal. Explanation for coal failure in the uniaxial strain tests at SIU: Harpalani and Palmer (refs 1, 2) describe the details of these tests. When CO2 is adsorbed, differential expansion occurs on a local scale due to the heterogeneous and anisotropic nature of coal. This can cause localized failure. If enough localized failures occur, they can link up and we can get gross failure, and the coal pops. The uniaxial strain condition may also be a key, for when CO2 is adsorbed under constant stress, the core expands but without apparent failure. We suggest that the localized failures are induced to link up more quickly under uniaxial strain, because the confining stress acting on the core plug has to be increased to maintain this condition. Differential expansion: A good example of this is provided in Figure 2 (ref 5), which shows expansion of different coal particles as they adsorb a liquid solvent, and swell in size. The amount of swelling was measured , and has been plotted in Figure 3, where it can be seen that the swelling differential can reach a factor of 10.

    HVB

  • 34

    2.0 2.3 2.3 2.8 2.4 2.4 1.9 2.5

    FIGURE 2: Eight NMR vials of Waterberg arranged by increasing colour with corresponding relative swelling extent (courtesy ref 5)

    FIGURE 3: Single-particle swelling results of Waterberg (top) Highveld coal (bottom) (courtesy ref 5)

    Waterberg

    Highveld

  • 35

    Coal is a very heterogeneous material, and when CO2 is adsorbed differential swelling will occur between different macerals: eg, a particle of vitrinite will expand much more than a particle of inertinite. The differential expansions will cause interparticle stresses to be set up, and these can lead to localized failure. Studies have been made of coal samples, kept under constant effective stress, and subject to CO2 sorption, and examined under X-ray CT (refs 12,13). The swelling and CO2 sorption in coal are heterogeneous processes. For example, vitrite showed the highest degree of swelling, while the clay (kaolinite) and inertinite region was compressed in response. The volumetric strains associated with swelling and compression were between 15%. Although the effective stress on the sample was constant, it varied within the sample as a result of the internal stresses created by structural changes induced by CO2 sorption. Figure 4 is a visual picture that shows a situation where coal and non-coal units are adjacent. This sample was held under uniaxial strain during the CO2 injection test. The image shows cracks developed between coal and non-coal units. No preexisting fractures were seen in separate high resolution (pixel= 6-7 micron) micro-tomography image, from the sample under uniaxial stress. It appears that hairline cracks appear in the coal units after CO2 cycles. A hairline crack is all that is needed at a localized failure site, before many such sites are linked up by stress increase to maintain a uniaxial strain condition.

    FIGURE 4: Visual of polished surface showing cracks induced by differential swelling as CO2 is adsorbed in heterogeneous coal (courtesy ref. 6).

  • 36

    Two Australian papers have addressed the weakening of weak lignite coal (UCS

  • 37

    Steve Bauer at Sandia did a thermal cracking PhD dissertation. This indicates for granites that thermal cycling (unconfined, to high temperatures in the range 600-800 C) reduces the UCS by about 50%, due to microcracking caused by differential thermal expansion at the grain scale. However, he notes that a confining pressure of 2900 psi or more is sufficient to eliminate this effect of the microcracks on strength. For comparison, the mean effective stress in San Juan basin is about 1100 psi, which is a lot smaller then 2900 psi. One might think that it might take less stress to eliminate microcracks in softer coal, but the crack inducement may be a lot stronger in coal because it is more heterogeneous. So more lab testing needs to be done to see how CO2-induced cracking depends on in-situ stresses acting on coal. This information may be used by analogy, because the effect of a large temperature change (600 C) on strong rock (granite UCS ~30,000 psi) would likely be similar to a much smaller temperature change on a much weaker rock (coal). In the CO2 lab test at SIU we are not dealing with temperature change of course, but rather with swelling induced by CO2, and we would have to convert this to an equivalent temperature change to take this approach any further. The important thing is that cracking by thermal cycling in granite, and weakening of this very strong rock, is an analog to the situation of CO2 adsorption/cycling in coal and the weakening of coal. Implications for the field: This has important implications for CO2 sequestration in the field, since most fields are likely to be under uniaxial strain or close to it. If coal fails easily and early when CO2 is injected, it is likely to affect the permeability of the formation encountered by the CO2, and the injectivity at the well. For example, shear failure in brittle rock is generally accompanied by dilatancy, when the permeability of the rock is increased (although this may be mitigated by fines creation and plugging). A net permeability increase would counteract the swelling of the coal by CO2 adsorption, and the injectivity at some point may increase rather than decrease. This could be very desirable. If failure of coal is induced by CO2 injection, we will need to improve the current reservoir models for perm change and injectivity, both of which are important to describe correctly. The current reservoir modeling of CO2 sequestration, which has reached a high level of sophistication, may be inadequate. Role of huff-and-puff: It has been reported (refs 8, 9) that huff-and-puff cycles of CO2 injection caused coal cores to fail in the lab, and that more cycles gave more breakage (each cycle was a few days). Some experiments were done to explore if CO2 huff-n-puff could be used to enhance near-wellbore permeability, by inducing coal to fracture. Some coals are fragile enough that the swelling and shrinking of coal by CO2 adsorption and desorption can induce sufficient stress to cause the coal to fall apart. Each cycle was a few days: injection followed by time to reach equilibrium. The failure seemed to be progressive,

  • 38

    and became more apparent after several cycles. None of this was published, but the idea was transmuted into a patent, by Raj Puri, as described in Appendix B. This information supports the thermal cracking analog discussed above. Each cycle of CO2 injection would be accompanied by differential swelling, and induce microcracks, but possibly in different local sites than the time before. That is, the extent of failure is likely to be increased during each cycle of huff-and-puff. In a micropilot in China, a well produced better after a huff-and-puff treatment with CO2, and this may indicate failure of coal (ref 10). In a CO2 sequestration project in Japan, there is evidence for an injectivity increase over time (ref 11). Further study should be directed at finding the best procedure for sequestering CO2 in coal, eg. it might be a series of huff-and-puff injections, optimally timed to increase the coal failure, and boost the injectivity. Conclusions:

    1. We have offered explanations for the behavior of coal under CO2 saturation. These explanations need to be confirmed by further lab testing which is carefully designed.

    2. The double-U curves of UCS versus rank provide a guide to coal strength when no core tests are available. The lab UCS measurements in this report suggest the double-U curve can sometimes be an overestimate of UCS in cores.

    3. Coal that has been saturated with CO2 in the lab (but then the CO2 removed) becomes stronger (UCS measurement) because it has dried, and the removal of the water lubrication effect dominates the CO2 weakening due to differential expansion.

    4. Coal fails very easily and very early in the lab when CO2 is adsorbed under uniaxial strain conditions. Differential expansion occurs on a local scale due to the heterogeneous and anisotropic nature of coal. This can cause localized failure. If enough localized failures occur, they can link up and cause gross failure.

    5. Studies have been made of coal samples subject to CO2 sorption, and examined under X-ray CT. Cracks developed between coal and non-coal units, and hairline cracks appear in the coal units after CO2 cycles. A hairline crack is all that is needed at a localized failure site, before many such sites are linked up by stress increase to maintain a uniaxial strain condition.

    6. Two Australian papers give interesting results on CO2-induced failure and permeability changes. The paper concludes that weakening by CO2, failure by in-situ stresses, and permeability increases will not be an issue for sufficiently deep coal seams. However, this is based on lab tests of very weak lignite, and its risky to apply the same conclusion to much stronger bituminous coals, such as HVB, HVA, or MV. Further study of how CO2 weakening depends on coal rank and in-situ stress will be important.

    7. Cracking by thermal cycling in granite, and weakening of this very strong rock, is an analog to the situation of CO2 adsorption/cycling in coal and the weakening of coal.

  • 39

    8. This has important implications for the field, which is likely to be under uniaxial strain: if coal fails easily and early when CO2 is injected, it is likely to affect the permeability of the formation encountered by the CO2, and the injectivity at the well. The injectivity may increase rather than decrease.

    9. The reservoir modeling of CO2 sequestration, which has reached a high level of sophistication, may be inadequate, and may need to be corrected.

    10. A basic theoretical/modeling study needs to be done to understand coal failure, and to predict the effect of permeability changes due to coal failure induced by CO2 injection (as well as perm changes due to swelling, which are fairly well understood). This will lead to improved reservoir modeling of CO2 injectivity, which is of paramount importance to the economic success of CO2 sequestration.

    11. The study should also be directed at finding the best (optimal) procedure for sequestering CO2 in coal, eg. it might be a series of huff-and-puff injections that increase the coal failure, and boost the injectivity.

    12. A program of study needs to be defined and funded to study the many aspects of this whole issue.

    References:

    1. Harpalani, S. Core flooding experiments, Presented at Coal-Seq V, Houston, November 2006.

    2. Palmer, I. Geomechanical analysis of core flood experiments, Presented at Coal-Seq V, Houston, November 2006

    3. Keleman, S., private communication, 2007. 4. Zheng, Khodaverdian, and McLennan, Static and dynamic testing of coal

    specimens, SCA Conf, Paper 9120, 1991. 5. Van Niekerk and Mathews, Solvent swelling of similar rank South African

    vitrinite-rich and inertinite-rich coals: swelling extent and maceral influence, to be published, 2007.

    6. Karacan, O., private communication, 2007. 7. Higgs, N., private communication, 2007. 8. Unpublished Amoco work 9. Puri, R., private communication, 2007. 10. Gunter et al, Report on CO2 micropilot in China injectivity may increase 2004

    or 2005? 11. Fujioka, M. Yubari project, Coal-Seq V, Houston, November 2006. 12. Karacan, O. Heterogeneous sorption and swelling in a confined and stressed coal

    during CO2 injection, Energy and Fuels, 17, 1595-1608, 2003. 13. Karacan, O. Swelling-induced volumetric strains internal to a stressed coal

    associated with CO2 sorption, Intl J. Coal Geology, in press, 2007 14. Viete, D.R. and Ranjith, P.G. The effect of CO2 on the geomechanical and

    permeability behavior of brown coal: implications for coal seam CO2 sequestration, Int J. of Coal Geology, 66, 204-216, 2006.

    15. Viete, D.R. and Ranjith, P.G. The mechanical behavior of coal with respect to CO2 sequestration in deep coal seams, Fuel, in press, 2007.

  • 40

    Acknowledgements: We have benefited from discussions on the subject of coal failure with Zissis Moschovidis (who first suggested the CO2 weakening), Nigel Higgs (who suggested the thermal cracking idea), Jonathan Matthews (who provided the solvent swelling information), Ozgen Karacan (who provided the 2D pictures of core cracking), John Olson, John McLennan, and Wes Martin.

    23 August 2007 Ian Palmer (Higgs-Palmer Technologies) On behalf of Ian Palmer (Higgs-Palmer Technologies) and Satya Harpalani (Southern Illinois University) Appendix A: Details of second set of UCS results by TerraTek. The figure shows X-ray images of the whole core, and cross-sections. The CO2 core appears to have more cleats, although most of these are calcite-filled. This core is also a little longer, and a bit less dense (see Table 1). All these effects would tend to make the CO2 core weaker, but in fact it is stronger, so we can rule out these effects as contributors to the strength discrepancy.

    Appendix B: Extracts from patent number 5,014,788

  • 41

  • 42

Click here to load reader

Reader Image
Embed Size (px)
Recommended