Swelling Characteristics and Interaction Mechanism of High-Rank Coal during CO2 Injection: A Molecular Simulation Study.

In CO2-enhanced coalbed methane (CO2-ECBM) engineering, accurate knowledge of the interaction mechanism of CO2 and coal matrix is crucial for improving the recovery of CH4 and contributing to the geological sequestration of CO2. This study is performed to prove the accuracy of molecular simulation and calculate the variation characteristics of pore structure, volumetric strain, mechanical properties, Fourier transform infrared (FT-IR) spectra, and the system free energy by molecular dynamics (MD) and grand canonical Monte Carlo (GCMC) methods. According to the obtained results, a relationship between pore structure, swelling strain, mechanical properties, chemical structure, and surface free energy was established. Then, the correlation of various coal change characteristics was analyzed to elucidate the interaction mechanism between CO2 and coal. The results showed that (1) the molecular simulation method was able to estimate the swelling mechanism of CO2 and coal. However, because the adsorption capacity of the molecular simulate is greater than that of the experiment and the raw coal is softer than the macromolecular structure, the molecular results are slightly better than the experimental results. (2) As pressure increased from 0 to 4 MPa, the intramolecular pores and sorption-induced strain changed significantly, whereas when the pressure increased from 4 to 8 MPa (especially at 6-8 Mpa), there was an increase of the intermolecular pores and mechanical properties and transition from elastic to plastic. In addition, when the pressure was >8 MPa, the coal matrix changed slightly. ScCO2 with a higher adsorption capacity results in greater damage and causes larger alterations of coal mechanical properties. (3) The change of the coal matrix is essentially controlled by the surface free energy of the molecular system. E valence affects the aromatic structure and changes the volume of the intramolecular pores, thus affecting the sorption-induced strain change rate. E non affects the length of side chains and the disorder degree of coal molecules and changes the volume of the intramolecular pores, thus affecting the mechanical property change rate. Our findings shed light on the dynamic process of coal swelling and provide a theoretical basis for CO2 enhancing the recovery of CH4 gas in coal.

Introduction

CO2-enhanced coalbed methane (CO2-ECBM) recovery has a significant impact on CO2 sequestration and unconventional gas extraction,[1−3] which will contribute to long-term safety production and alleviation of global warming.[4,5] CO2 injected into the deep coal reservoir is converted into a supercritical state (ScCO2),[6,7] which has a complex physical or chemical reaction with the coal seam, leading to coal structure rearrangement, coal matrix deformation, and changed mechanical properties of coal seams.[8−12] The physical properties of coal reservoir, such as porosity and permeability, are also changed,[13,14] thereby triggering numerous unexpected responses in coal seams upon CO2 injection. Therefore, a proper understanding of the change mechanism of the coal matrix upon CO2 injection is essential for optimization of the CO2-ECBM techniques.

A number of investigations have explored the effect of CO2 adsorption on coal pore structure. Kang et al.[15] found that the micropore volume decreases after supercritical CO2 extraction. Liu et al.[16] found that, after the ScCO2 treatment, the volume of the micropores with width <0.46 nm in high-volatile bituminous coal increased slightly, but the decrease in the volume of micropores with width <0.46 nm in the low-volatile bituminous, semianthracite, and anthracite coal became more pronounced. Zhang et al.[17] found that the volume of mesopores and macropores increases in high-, medium-, and low-rank coals. Liu et al.[18] found that after CO2 treatment, the mesopore volume increased and the macropore volume decreased with the increase of coal rank. In other words, after the CO2 treatment, the pore structure is changed, but it is not clear how the micropores, mesopores, and macropores change in different rank coals.

Researchers have developed relationships between coal mechanical property and coal rank. Ranjith et al.[19] compared the effect of lignite and bituminous coal and found that the strength reduction in bituminous coal is about 4.5 times higher than the strength reduction in lignite, and the best fitting curves for unconfined compressive stress (UCS) reduction with CO2 saturation pressure exhibit different slopes for bituminous (15.3) and lignite coal (3.2), revealing that increasing CO2 saturation pressure significantly affects the strength of bituminous coal. Similarly, Perera et al.[20] saturated bituminous coal at 8 MPa and 33 °C for 7 days and found that UCS decreased by 77.70%, which is significantly higher than UCS found by Ranathunga et al.[21] for low-rank coal saturated with CO2 under 8 MPa and at 35 °C for 25 days. That is to say, CO2–coal interaction leads to a more significant alteration in the mechanical properties of high-rank coal.

The inconsistency of changes in coal pore structure and mechanical properties is due to the diverse chemical structure of coal ranks. Low- and middle-rank coals contain many branches, while high-rank coals contain fewer branches and more aromatic rings. High-rank coals have more adsorption sites for CO2 than low-rank coals and consequently more swelling potential upon CO2 adsorption.[22,23] According to Griffith et al.[24] and Larsen et al.,[25] the adsorption of CO2 with higher chemical potential reduces the surface energy of coal, which leads to the rearrangement of the macromolecules in coal to a more noncovalently associated structure. Therefore, the complex change process of the pore structure and mechanical properties is affected by the surface free energy of coal. To understand the different mechanisms coupled with each other, some scholars have used graphite and coal structure models to reveal specific processes of coal swelling caused by gas injection through molecular dynamics.[26−28] Zhang et al.[26] simulated the mechanical properties of methane bearing high-rank coal and found that with the increase of CH4 content, the total surface free energy decreases, which leads to the reduction of the strength. Wang et al.[27] simulated a macromolecular rearrangement caused by CO2 injection through molecular dynamics processes and found that the compression of closed pores and expansion of open pores together causes swelling of macromolecular volumes. Zhang et al.[28] used grand canonical Monte Carlo (GCMC) simulation to study the adsorption-induced deformation strain based on the deformation of organic carbon slit pore models and found that the molecules close to pore walls are parallel (tending to swell the pore) and play a dominant role in the deformation, and when the temperature increases, both swelling and shrinkage decrease due to a decrease in the adsorption density. However, coal is a complex compound with various functional groups and cross-linked bonds, and graphite and organic carbon slit pore models do not represent the structure of coal. Therefore, the simulation results are inconsistent. Wang et al.[27] showed that closed pores (diameter < 0.55 nm) decreased and the open pores (0.8 nm < diameter < 1.0 nm) increased. Zhang et al.[28] indicated that the pore aperture <0.55 nm shows no deformation and the pore aperture between 0.55 and 0.6 nm has the greatest swelling.

In this study, the high-rank coal Chengzhuang (CZ) from Qinshui Basin of China was selected as a research subject. The study mainly explores the influence of surface energy on pore structure, mechanical properties, and macromolecular structure after CO2 treatment. First, the results of the experimental method and the molecular simulation method were compared to see whether they were consistent. If they were consistent, the reliability of the molecular simulation method was determined. Second, the molecular simulation method was used to study the change in coal deformation, mechanical properties, molecular structure, and surface free energy. Lastly, the relationship of various deformation characteristics was analyzed to illustrate the CO2 swelling mechanism of high-rank coal. All molecular simulation calculations are performed in Materials Studio 2018 software.

Simulation and Calculation Methods

CZ Coal Molecular Structure

The macromolecular structure model of CZ (Figure ) coal was constructed based on the results of proximate and ultimate analyses, 13C NMR spectrum, and XPS spectrum. The proximate and ultimate analyses of the model and the experiment are shown in Table S1, the 13C NMR spectra of the model and the experiment are shown in Figure S1. On the whole, the model parameters agreed well with the experimental parameters, except for lower chemical shifts and a smaller area of the oxygen carbon zones in the CZ model compared with those in the experimental data.[29]

Figure 1

Model macromolecular structure in CZ coal.

CO2 Injection Simulation Process

Simulation Model Construction

First, a supercell containing 100 CZ coal molecules was constructed, and the structure of the supercell model was optimized.[30] The geometry optimization task in the Forcite module was used to optimize the structure of the model. The final configuration was an optimized coal model with the lowest energy. On the basis of the change in energy, Metropolis operation rules were applied to accept or reject the change for forming a new configuration. A periodic boundary condition was used in three dimensions, and the cell parameters were as follows: 6.77 nm × 6.77 nm × 6.77 nm (Figure ). The Nose thermostat and NVT ensemble with a 1 fs time step were performed. For full interaction, 500 ps simulations were conducted. The Ewald method with a precision of 0.001 kcal/mol was used for electrostatic interactions, while van der Waals interaction was calculated using the atom-based option with a cutoff of 12.5 Å.

Figure 2

Supercell containing 100 CZ macromolecules.

CO2 Injection Process

CO2 was simulated at 0, 2, 4, 6, 8, 10, 12, 14, and 16 MPa CO2 injection pressure. The amount of CO2 adsorbed at a fixed pressure is obtained by the fix pressure task in the Sorption module, and the COMPASS II force field,[31] a temperature of 298 K, and Metropolis operation rules[32] were used for the simulation.

To simulate the CO2 injection process in the coal, the Adsorption Locator module from Materials Studio 2018 was used to place some CO2 molecules in the box, according to the average adsorption in different fixed-pressure adsorptions. Then, the optimization of macromolecular structures, containing CO2, was carried out using the isothermal isobaric ensemble (NPT) and selecting the Andersen temperature-control method (298 K) and the Berendsen pressure control method. The duration of the simulation was 300 ps, the step size was 1 fs, and the sampling interval was 100 fs. The last 50 ps of data were collected for analysis.[27]

Isothermal Adsorption of CO2

The isothermal adsorption calculations were conducted by the adsorption isotherm task in the Sorption module. The force field used was COMPASS II. The convergence criterion for energy was 5 × 10–4 kcal/mol, and the fugacity step was 20. The maximum number of iterations and equilibration steps were set as 50 000 and 10 000, respectively. The Ewald sum method was used for electrostatic action, and van der Waals force was calculated via the atom-based method with cubic spline truncation. The cutoff distances for both electrostatic action and van der Waals force were set as 12.5 Å. To ensure balance in the system, 2 × 107 GCMC steps were adopted.

Mechanical Property Simulation

The mechanical properties were simulated by the mechanical properties task in the Forcite module. The method is constant, the number of steps for each strain is 11, and the maximum strain amplitude is 0.1.

Young’s modulus, Poisson’s ratio, and stress–strain curves are usually used to characterize the mechanical properties.

Unconfined compressive stress (UCS)[33]where P is the maximum load at failure of the specimen, N, and A is the sample area, mm2.

Young’s modulus[34]

Poisson’s ratio[34]where E is Young’s modulus, GPa; v is Poisson’s ratio; and λ and μ are Lamé coefficients, GPa.

Stress–strain curves.

For each configuration, several strains are applied, resulting in a strained structure. The resulting structure is then optimized, keeping the cell parameters (and hence the strain) fixed to allow for internal relaxation. The number of steps for each strain is 11, the maximum strain amplitude is 0.003, and strain patterns are 100000 and 010000.

This defines a range of values, {−0.003, −0.001, 0.001, 0.003}, applied to each strain pattern[35,36]These are then used to generate the metric tensor: G, GPa.

Each strain pattern represents the strain matrix in Voigt notation. It is converted to the strain matrix, ε, such that ε(0,0) = e(0), ε(1,1) = e(1), ε(2,2) = e(2), ε(2,1) = ε(1,2) = 0.5*e(3),...where H0 is formed from the lattice vectors, I is the identity matrix, and H0′ is the transpose of H0.

From G, the new lattice parameters can be derived; these parameters are used to transform the cell parameters (fractional coordinates are held fixed).

Volumetric Strain Deformation Models

The total volumetric strain, including sorption-induced strain variation and mechanical strain variation, is as followswhere ε, ε, and ε are the total volumetric, sorption-induced, and mechanical strain, respectively.

The mechanical strain caused by stress can be written as[37,38]where Es and vs are Young’s modulus and Poisson’s ratio of the coal matrix, respectively, and P is gas pressure, MPa.

The linear strain was derived as[39]whereand where c = 8√2/3π = 1.2.

From this structure model, porosity can be calculated asSince the high-pressure gas compresses the coal solid, the strain caused by this pressure alone is[40]Combining adsorption and pressure compression strains and using γ = Φ/A, the overall strain can be derived asWith a Langmuir adsorption isotherm modeland assuming the fugacity is equal to pressure for calculation simplicity, the surface potential can be calculated asVa is the volume change for unit mass of adsorbent, cm3/g. Its relationship with strain isThe volumetric strain can be obtained from eqs and 14 being substituted into eq where ρs is the solid density, cm3/g. Es and vs are Young’s modulus and Poisson’s ratio of the coal matrix, respectively.

When the simple fitting method is adopted to describe the volumetric deformation variations, the model can be simplified as[41]where m is a fitting parameter determined by coal characteristics.

Surface Potential

The strain variations are linked with surface potential. The surface potential for compressible solid adsorbent can be written as[42]where Φ is the surface potential, Kcal/mol; Va is the solid volume change at the equilibrium adsorption pressure, cm3/g; P is the gas pressure, MPa; na is the amount of adsorbed gas, N; R is the gas constant, 8.314J/(mol·K); and T is the temperature, K.

Results and Discussion

Experiment and Molecular Simulation Comparison

The variation characteristics of the CO2-adsorption capacity and mechanical properties obtained by the experiment and molecular simulation have been compared. The experimental results of the CZ coal matrix swelling characteristics under different CO2 injection pressures were studied by Wang et al.[43]

CO2 Adsorption Capacity

Figure shows the results of the CO2 absolute adsorption capacity obtained by the experiment and simulation, respectively. The absolute adsorption capacity obtained by both methods increased with the increasing CO2 injection pressure, and the CO2 adsorption capacity increased fastest when the injection pressure was lower than 4 MPa but increased slowly when the pressure was greater than 4 MPa. Although the adsorption curves obtained by the experiment and simulation are analogous, the amount of CO2 adsorption obtained by simulation is greater than that obtained by the experiment. This is because samples used by Wang et al.[43] were core coals, and the volume of pores in the experiment was smaller than that of the molecular simulation, resulting in an underestimation of CO2 adsorption capacity in the volumetric methane adsorption experiment.[44]

Figure 3

Relationship between the absolute adsorption capacity of CZ coal and CO2 injection pressure.

Volumetric Swelling Strain

Figure shows the relationship between volumetric swelling strain and CO2 injection pressure from the experiment and simulation. Because the CO2 adsorption capacity calculated using molecular simulation is stronger than that of the experiment, the simulated volumetric swelling strain is larger than the experimental one.

Figure 4

Relationship between volumetric swelling strain and CO2 injection pressure obtained from the experiment and simulation.

Mechanical Properties

Figure shows the distinction of Young’s modulus and Poisson’s ratio between the experiment and simulation at 4 and 8 MPa. The Young’s modulus values obtained by the experiment are larger than those obtained by the simulation, while the Poisson’s ratio values obtained by the experiment are smaller than those obtained by the simulation. The higher the Young’s modulus value, the smaller the elastic modulus deformation of the material, and the easier it is for the material to deform. This indicates that the raw coal sample is softer than the coal macromolecular structure. Both results show that in the subcritical CO2 injection stage, Young’s modulus decreased while Poisson’s ratio increased slowly. During the transition phase of CO2 injection pressure, the change values of Young’s modulus and Poisson’s ratio increased. This shows that the phase transition softened the coal matrix, and with the increase of CO2 injection pressure, the deformation of coal and rock changed from brittleness to plasticity.

Figure 5

Distinction of Young’s modulus and Poisson’s ratio between the experiment and simulation.

The results of the molecular simulation are better than those of the experiment because the molecular simulation is in an ideal environment; however, their changing trends are consistent. Therefore, molecular simulation is an effective method to study the coal matrix alterations during CO2 injection into coal. In addition, it can study the interaction mechanism of CO2–coal.

Coal Matrix Deformation

Pore Structure Deformation

Table shows the pore structure parameters of coal under different CO2 injection pressures. The pore size distribution was calculated using Poreblazer software.[45] When the injection pressure increased, the total pore volume, porosity, and intermolecular pore (>0.4 nm) volume increased, whereas the intramolecular pores (<0.4 nm) decreased. When the pressure was less than 4 MPa, the volume of intramolecular pores changed rapidly while the volume of intermolecular pores changed slowly. At 4–8 MPa, the pore structure parameters changed and the proportion of intermolecular pores exceeded that of the intramolecular pores; the porosity changed from 5.25 to 6.61%. After 8 MPa, the pore volume remained stable and the porosity decreased from 6.61 to 6.49%.

Table 1

Coal Macromolecular Pore Structure Parameters under Different CO2 Injection Pressures

gas pressure (MPa)	coal volume (nm3)	total pore volume (cm3/g)	intramolecular pore volume (cm3/g)	intermolecular pore volume (cm3/g)	porosity (%)
0	34.66	0.0501	0.0340	0.0161	5.12
2	35.31	0.0513	0.0332	0.0181	5.23
4	35.51	0.0530	0.0219	0.0211	5.25
6	35.51	0.0605	0.0194	0.0311	6.05
8	36.01	0.0672	0.0177	0.0495	6.61
10	35.80	0.0668	0.0166	0.0502	6.62
12	35.48	0.0666	0.0162	0.0504	6.63
14	35.49	0.0666	0.0161	0.0505	6.63
16	35.44	0.0664	0.0160	0.0504	6.63

The pore size distribution (PSD) of CZ coal determined from CO2 adsorption is shown in Figure . (When the pressure is >12 MPa, the curves of PSD do not change obviously, which is not shown in Figure .) The pores are mainly distributed in the pore size range of 0.3–0.9 nm, and the pore volume presents a multipeak distribution (generally two to three). There are two main peaks at 0.35–0.40 (intramolecular pore) and 0.65–0.75 nm (intermolecular pore). After CO2 treatment, both the peak positions and peak shapes are significantly altered. At 0, 4, 8, and 12 MPa, the peak values of intramolecular pore are 0.075, 0.080, 0.088, and 0.089 cm3/g/nm and the peak values of intermolecular pore are 0.0063, 0.053, 0.076, and 0.065 cm3/g/nm, respectively. The two peak values are higher in CZ coal after treatment, the width of the intramolecular pore peak is reduced, and the intermolecular pores are increased. This finding illustrates that after CO2 adsorption, the intramolecular pores were compressed, providing space for the formation of intermolecular pores, and ScCO2 treatment has a remarkable influence on the micropore distribution of high-rank coal. Chen et al.[46] studied the influence of ScCO2 on the pore structure of low- and high-rank coal and concluded that the largest increase in macropore volume is observed in low-rank coal and the greatest reduction of micropore volume in high-rank coal. This phenomenon can also be found in the studies of Gao et al.,[47] Cheng et al.,[48] Juan et al.,[49] and Liu et al.[50]

Figure 6

Pore size distribution of CZ coal at different CO2 injection pressures.

To compare the PSD results with conventional pore measuring technology, LP-N2-Ad were carried out using an automated gas sorption analyzer (Autosorb iQ-MP, Quantachrome Instruments, Germany), in accordance with the national standards (GB/T 21650.2-2008, 2008;[51] GB/T 21650.3-2011, 2011[52]), and the calculation results were obtained by DA and NLDFT models. Figure shows the PSD curves of simulation and DA and NLDFT models. The PSD curves of simulation and NLDFT are similar are significantly different from the DA model result. The DA model does not consider changes of these parameters or the influence of pore shape on gas molecule accumulation; therefore, the calculation results of this model may deviate further.[53] The analysis results of simulation and the NLDFT model, which describe the adsorption behavior in the narrow space of micropores, are more accurate. The pore volume obtained by the NLDFT model is slightly lower than the analysis result of the simulation, which might be caused by the effect of ash in the coal matrix.[54] This indicates that the molecular simulation method is accurate for analyzing the pore size distribution of micropores in coal.

Figure 7

Comparison and analysis of PSD by simulation and NLDFT and DA models.

Volumetric Strain Deformation

The relationship between swelling strains and CO2 injection pressure is shown in Figure . The volumetric strain deformation caused by CO2 adsorption is mainly composed of two parts: the sorption-induced strain caused by CO2 adsorption and the mechanical strain caused by effective stress change. According to Section , the sorption-induced strain can be obtained by subtracting the mechanical strain from the volumetric strain. The volumetric strain and sorption-induced strain change values at pressure <4 MPa are 1.5 times those at 4–8 MPa. But the mechanical strain is different from them. The mechanical strain curve increased slowly at 0–6 MPa and increased rapidly at 6–8 MPa. When the pressure was >8 MPa, the deformation change in all of them was not obvious. It could be seen that the sorption-induced swelling is a dominant factor in volumetric strain. Zhao et al.[41] also obtained the same result from experimental research.

Figure 8

Relationship between swelling strain and different CO2 injection pressures.

Effects of Pore Structure on Swelling Strain

As shown in Table and Figures and 9, when pressure ranged within 0–4 MPa, we found that the absolute adsorption capacity increased from 0 to 0.8 mmol/g, the volumetric and sorption-induced swelling strain increased from 0 to 9000 and 0 to 4800, respectively, and the intramolecular pore reduced from 0.0340 to 0.0219. When pressure ranged within 4–8 MPa, the absolute adsorption capacity increased from 0.8 to 1.0 mmol/g, the volumetric and sorption-induced swelling strain increased from 0 to 12 000 and 0 to 5500, respectively, and the intramolecular pore reduced from 0.0219 to 0.0177. However, when pressure was at the >8 MPa stage, the absolute adsorption capacity, intramolecular pore, and swelling strain are not obviously changed.

Figure 9

Relationship of pore structure and swelling strain.

This phenomenon can be attributed to the CO2 adsorption characteristics of coal. The swelling of the coal matrix is related to the amount of CO2 adsorbed. The higher amount of CO2 adsorbed, the greater the increase in volumetric and sorption-induced swelling strain, and the pore structure affects the adsorption of CO2, thereby affecting the volumetric and sorption-induced swelling strain of coal. The swelling caused by the adsorption of CO2 changes the pore structure of coal, and this change of the pore structure of coal also affects the adsorption capacity of CO2. In the initial stage of adsorption, the coal matrix contains <0.4 nm pores, and there is no gas adsorbed in the coal matrix; CO2 tends to be adsorbed in pores <0.4 nm. Therefore, CO2 can be quickly adsorbed in the pore structure of coal in the low-pressure stage, and this causes volumetric and sorption-induced swelling strain to increase rapidly at 0–4 MPa. With the increase of the amount of CO2 adsorbed, the macromolecular structure of coal changes significantly, resulting in a decrease in the pore content of <0.4 nm with increasing pressure. Therefore, the rate of coal adsorption of CO2 slows down with the increase of pressure. Thus, when the adsorption amount of CO2 increases rapidly, the swelling stress changes significantly (0–4 MPa), then changes slowly when the <0.4 nm pore content decreases and the increase in CO2 adsorption decreases (4–6 MPa) and finally becomes stable.

The mechanical strain changes correlate with intermolecular pores and are discussed in the next section.

Coal Mechanical Properties

The following sections discuss the various mechanical properties of coal: unconfined compressive stress (UCS); Young’s modulus (E); Poisson’s ratio (v); and stress–strain features for three stages, subcritical stage (<6 MPa), transfer stage (6–8 MPa), and supercritical stage (>8 MPa).

UCS, Young’s Modulus, and Poisson’s Ratio

The modified characteristics of UCS are shown in Figure S2 and Table . With an increase in the CO2 saturation pressure, UCS of the coal decreased. At pressures from 0 to 6 MPa, UCS reduced slowly. When CO2 injection pressure increased from 6 to 8 MPa, CO2 transformed from a sub- to a supercritical state. The UCS reduction amplitude in coal was subjected to a sudden increase from 30.76 to 52.75%. However, when the pressure was >8 MPa, especially when the pressure was >12 MPa, there was no change in UCS with increasing pressure. According to Zeng et al.,[54] when pressure increased, the contribution to the fracture (cleat) aperture change increased from zero to a peak value and then dropped to zero, reducing the diffusion capacity of CO2 with increasing pressure. Thus, UCS changed less at higher supercritical CO2 pressures.

Table 2

Mechanical Parameter Changes of CZ Coal with Different CO2 Injection Pressures

pressure (MPa)	UCS (MPa)	ΔUCS (%)	E (GPa)	ΔE (%)	v	Δv (%)
0	68.71		6.46		0.26	0
2	57.57	16.20	5.34	17.33	0.3	15.38
4	52.74	23.23	3.81	41.02	0.33	26.92
6	47.57	30.76	3.51	45.66	0.36	38.46
8	32.46	52.75	2.37	63.31	0.43	65.38
10	31.89	53.57	2.06	68.11	0.44	69.23
12	31.31	54.44	1.95	69.81	0.45	73.07
14	31.20	54.59	1.93	70.12	0.45	73.07
16	31.22	54.59	1.91	70.43	0.45	73.07

The coal matrix swelling caused by CO2 adsorption affected Young’s modulus and Poisson’s ratio (Figures S3 and S4 and Table ). Similar to UCS, these variations could be divided into three stages. At the subcritical CO2 saturation stage (<6 MPa), Young’s modulus decreased by 45.56% and Poisson’s ratio increased by 38.46%. In the transfer stage (6 MPa < pressure < 8 MPa), compared with the subcritical CO2-saturation stage, Young’s modulus was reduced by 1.5 times while Poisson’s ratio was increased by 1.5 times (Young’s modulus: 63.31%; Poisson’s ratio: 65.38%). When CO2-saturation pressure increased beyond 8 MPa, Young’s modulus and Poisson’s ratio did not cause a perceptible change.

The change in mechanical properties shown in the three stages is related to the brittle–ductile properties and the polymeric structure of coal. When CO2 was absorbed into the coal matrix, swelling changed its cross-linking and polymeric structure.[45] This improved the ductile properties of coal, reduced Young’s modulus, and increased Poisson’s ratio. Supercritical CO2 has a higher plasticizing effect than subcritical CO2, and the decrease in Young’s modulus and the increase in Poisson’s ratio become larger. When the saturation pressure of CO2 was further increased, the fracture (cleat) aperture of coal could be closed, and the diffusion capacity of CO2 was reduced. Thus, the plasticization capacity of coal was stable at high pressure (>8 MPa).

Stress–Strain Features

Figure shows the stress–strain curves of the coal matrix. (The curves of 14 and 16 MPa are overlapped with 12 MPa, which are not shown in Figure .) Under the influence of CO2 injection pressure, the stress–strain curves of CZ coal are different. The compression deformation was induced by stress. The stress–strain curves could be divided into four parts: compaction, elastic deformation, crack expansion, and peak strength stages.

Figure 10

Stress–strain curves of CZ coal with different CO2 injection pressures.

Compaction stage: with a gradual increase in axial pressure, the curve of this process is concave; the initial curve is relatively stable and then begins to rise, indicating that the slope of the curve increases and the stiffness of the coal rock increases. With the gradual increase in the axial stress, the pre-existing pores and cracks in the coal and rock were compressed or even closed. A small change in stress in this stage causes a large deformation. As the compressibility of the pores and cracks decreased, under the same stress, the amount of deformation gradually decreased, which caused concave stress–strain curves in the initial stage.

Elastic deformation stage: In the previous stage, the coal sample was compacted, the degree of pores and fissures was reduced, and the density of coal became larger. In this stage, there is a linear relationship between the strain and stress of the coal, and the stress–strain curves can be regarded as a straight line.

Crack expansion stage: with a further increase in the axial stress, coal enters the stage of crack expansion, and the stress–strain curves enter the stage of nonlinear change again. During this process, the slope of the stress–strain curves decreases, and the stiffness of the coal sample also begins to decrease. Due to the generation of cracks, the increase rate of the volumetric strain of the coal sample decreased.

Peak strength stage: the coal and rock rupture, and the stress–strain curve drops as the crack continues to increase. Since the macromolecular structure of coal cannot simulate coal rupture, the curve declines.

The stress–strain curves are different at subcritical and supercritical CO2 adsorption pressures. The elastic deformation in the supercritical CO2 stage was greater than that in the subcritical CO2 stage, indicating that the supercritical CO2 adsorption reduced the elastic section of the stress–strain curve and enhanced the plasticity of the coal matrix.

Effect of Pore Structure on Mechanical Property

Previous studies have shown that mechanical properties are related to the coal pore structure. For example, Masoudian et al.[55] observed a decrease in the size of the particles and considered that such changes caused a reduction in strength. Li et al.[56] showed that the mechanical strength weakened in relation to an increase in macropores. However, it should be noted that experimental methods can only describe qualitatively rather than quantitatively.

As shown in Figure , the relationship between the pore structure and UCS is described quantitatively. The change in intermolecular volume is negatively correlated with the change in UCS. At 0–6 MPa, the intermolecular pores increased from 0.0161 to 0.0311 cm3/g and UCS reduced to 30.76%. At 6–8 MPa, the intermolecular pores increased from 0.0311 to 0.0495 cm3/g and UCS reduced to 52.75%. When the pressure was >8 MPa, the intermolecular pores and volume remained stable. The increased intermolecular pores indicated that the original pores were enlarged by swelling, the plasticity of the coal matrix increased, and the mechanical properties of coal weakened. The relationship between the pore structure, Young’s modulus, and Poisson’s ratio is shown in Figure S5, in which the correlation is similar to that with UCS. Supercritical CO2 has higher diffusivity and lower viscosity[57] and adsorbs pores more easily than subcritical CO2, and thus, the impact of supercritical CO2 on coal is greater than that of subcritical CO2.

Figure 11

Relationship of pore structure and UCS.

Coal Macromolecular Structural Properties

The effect of CO2 on coal macromolecules is mainly seen in two ways. (1) The difference in the macromolecules of low–middle- and high-rank coal. In the coalification process, the number of side chains and functional groups were constantly reduced due to the weak binding force of side chains and functional groups.[58] (2) The diversity of interactions between CO2 and coal functional groups. The adsorption order of CO2 is aromatic rings >−C2O> C=O > −CH3 > −OH.[59]

FT-IR spectra can be obtained to derive molecular structure information by analysis of the position and intensity of absorption peaks of different functional groups.[60] The basic structural units in coal, such as alkyl side chains, oxygen-containing groups, or aromatic rings, can be analyzed in detail on the basis of changes in the absorption peak intensity. To determine changes in the characteristic absorption peaks in the infrared spectrum after ScCO2 injection with different pressures, the FT-IR spectrogram is divided into four parts: (1) hydroxyl groups (−OH, 3700–3100 cm–1), (2) aliphatic structures (−CH, 3000–2800 cm–1), (3) aromatic structures (1600 cm–1), and (4) aromatic out-of-plane structures (−C–H–, 900–700 cm–1); the FT-IR spectra of the functional-group region are shown in Figure .

Figure 12

FT-IR spectra of CZ with different CO2 injection pressures.

The intensity of the −OH group absorption peak for CZ coal had no obvious changes after CO2 treatment (Figure , gray circle). As coal rank increases, the carbon content gradually increases and the oxygen content decreases. CZ, which is high-rank coal, has low oxygen content and high carbon content, and the coal matrix swelling caused by CO2 adsorption did not change the oxygen content. As a result, the −OH group showed almost no changes with CO2 treatment.

The 3000–2800 and 1600 cm–1 wavelengths changed more significantly under the supercritical CO2 pressure than under the subcritical CO2 pressure, resulting in a sudden enhancement of aliphatic side chains (−CH) and a reduction of aromatic carbon (−C=C−) (Figure , green and blue circles). −C=C– of aromatic structures, particularly polycyclic aromatic hydrocarbons (PAHs), have low bond dissociation energy and are easily broken. The longer the aliphatic side chains, the smaller the energy need for bond dissociation, and R2–CH2 breaks easier than −CH3. CZ coal has high PAH content and low aliphatic content, and the length of the aliphatic side chains is relatively short. After CO2 treatment, the aliphatic side chains in their original composition are expected to show almost no changes, as the PAHs are broken and changed into aliphatic side chains. As a result, the aromatic structures are obviously weakened and aliphatic side chains are increased.

The transmittance peak in the 900–700 cm–1 wavelength region indicates the vibration of aromatic out-of-plane C–H.[46,47] We observed that there was a slight increase. As explained by Lin et al.,[60] the coal samples under ScCO2 treatment predominantly displayed substitution reactions, which led to the increase of substituents. As the coal rank increased, the bond dissociation reactions also increased. Because CZ is high-rank coal, the change of substituents is not strong. Therefore, the band change of 900–700 cm–1 is not obvious.

Coal System Energy Properties

The total energy (Etotal) consists of valence (EValence) and nonvalence (Enon) (Etotal = Evalence+Enon) energies. From Table , Evalence and Enon decreased with different CO2 injection pressures, indicating that the position of the atom in the macromolecular structure and the distance between the molecules are changed.

Table 3

Energy Comparison of CZ Coal with Different CO2 Injection Pressures (Kcal/mol)

		EValence				Enon
gas pressure (MPa)	ETot	EB	EA	ET	EI	EvdW	Ee
0	24197.37	1361.24	1342.14	20566.71	181.2	873.45	–127.37
2	24169.25	1360.11	1339.63	20565.18	167.89	863.84	–127.4
4	24116.86	1352.94	1325.11	20554.88	160.18	851.16	–127.41
6	24072.91	1345.98	1320.93	20552.71	157.84	829.23	–133.78
8	23994.15	1341.91	1315.02	20548.01	154.91	768.24	–133.94
10	23949.84	1321.61	1314.99	20547.87	152.74	753.19	–140.56
12	23916.2	1321.55	1314.27	20547.33	151.14	722.97	–141.06
14	23913.01	1321.38	1314.07	20547.21	151.08	722.08	–142.81
16	23909.64	1321.25	1313.99	20547.17	151.04	721.45	–145.26

Variation of EValence of Coal under Different CO2 Injection Pressures

EValence is the main form of the stable molecular structure. The change of valence energy causes atom displacement, and the displacement of atoms leads to bond stretching (EB) and angle (EA), torsion (ET), and inversion (EI) changes in the model structure. The greater the aromaticity of coal, the higher the EValence of coal. From Table , Evalence reduced from 23 451 to 23 333 Kcal/mol, which implies a decrease of aromaticity, causing the molecular structure to loosen.

Variation of Enon of Coal under Different CO2 Injection Pressures

Enon is composed of van der Waals (EvdW) energy and electrostatic energy (Ee), which represents the interaction between paired electrons. Ee is greater when the molecule has multiple charges or multipole moments. Coal molecules have high aromatic content and a large number of aromatic rings, which leads to the π–π stacking interaction in the molecular arrangement and controlled multipole moments of coal molecules. From Table , Ee changes from −127.3 to −145.26, which means that the multipole of coal molecules decreases, indicating that the aromatic structure is destroyed.

EvdW is to stabilize the molecular system. EvdW alternations include the following three situations: ① For molecules with a similar composition and structure, the greater the relative molecular mass, the greater the EvdW, ② for a similar molecular structure, the higher the number of molecular side chains and the greater the distance between molecules, the smaller the EvdW, and ③ uneven charge distribution of molecules reduces EvdW. It can be seen from Table that there is EvdW reduction during the interaction between carbon dioxide and coal, indicating that the length of the aliphatic chain increases, the space obstruction effect becomes stronger, and the disorder degree of the molecular arrangement increases.

Mechanism of CZ Coal Swelling

Surface free energy is the potential energy of the molecules or atoms on the surface. Coal adsorbs CO2 and induces swelling of the coal matrix, reducing coal surface free energy and leading to instability of the system. The molecules/atoms in the coal matrix tend to repel other molecules/atoms to achieve balance, leading to a rearrangement of the macromolecules in coal and changing the distance between the surface molecules and adjacent molecules, as shown in Figure .[61] The coal pore structure is related to the macromolecular structure. In high-rank coal, the intramolecular pores are associated with the interlayer spacing of the aromatic layers, and intermolecular pore volumes are determined by the directional arrangement of aromatic crystallites and the length of aliphatic hydrocarbon. The mechanical properties are not only related to the macromolecular structure but also related to the pore pressure; aliphatic chains lengthen, the aromaticity decreases, or the pore pressure increases, ultimately reducing the strength of coal. The relationship between pore structure, swelling strain, mechanical properties, chemical structure, and surface free energy is shown in Figure .

Figure 13

Conceptual diagram of sorptive matrix shrinkage and the swelling mechanism with intermolecular and surface forces. Reprinted (adapted or reprinted in part) with permission from AAPG Bulletin [Compressibility of sorptive porous media: Part 1. Background and theory]. Copyright [2014] [Shimin Liu;Satya Harpalani].

Figure 14

Relationship between pore structure, swelling strain, mechanical properties, chemical structure, and surface free energy.

When pressure is at 0–4 MPa, the coal matrix adsorbs a large amount of CO2, which causes the coal matrix to swell rapidly. Evalence decreased from 23 451 to 23 333 Kcal/mol, and this shows that the content of the aromatic structure is reduced, leading to the intramolecular pore volume changing from 0.0340 to 0.0219 cm3/g (corresponding to S1 changing to S2 in Figure ). EvwD decreased from 873.45 to 851.16 Kcal/mol, the distance between surface molecules and the adjacent molecules changes from D1 to D2 (D2 > D1), leading to the intermolecular pore volume changing from 0.0161 to 0.0211 cm3/g. The decrease of aromatic structure and the increase of gas pressure are also accompanied by the weakening of mechanical properties.

When the pressure was at 4–8 MPa, the CO2 pressure in pores increased, but the CO2 adsorption amount increased slowly, and the coal matrix swelling induced by CO2 adsorption can be resisted by external pressure. The higher pressure helps to form aliphatic hydrocarbons and increases the length of side chains and exacerbates system instability leading to the obvious decrease of EvdW (from 851.16 to 768.24 Kcal/mol). The attraction force of CO2 and coal molecules increased (the D2 increases), resulting in the intermolecular pore increase (from 0.0211 to 0.0495 cm3/g). Hal/H and CH2/CH3 increased, Aar/Aal decreased, and pore pressure increased; especially when pressure is at 6–8 MPa, carbon dioxide is converted into a supercritical state, and the mechanical properties of coal change greatly.

Above 8 MPa, the pressure causes the internal structure to become very regular and compact. Hence, the volumetric swelling stops, and the pore structure and mechanical properties remain stable.

Conclusions

In this work, the various characteristics of pore structure, swelling strain, mechanical properties, chemical structure, and surface free energy were evaluated, and the mechanism of the CO2–coal interaction in high-rank coal by molecular simulation was established. The conclusions of this study are as follows.

By a comparative study of the experimental and simulation data, molecular simulation is found to be an effective method to study the coal matrix alterations during CO2 injection into coal. The raw coal sample is more flexible than the coal macromolecular structure.

CO2 adsorption causes a significantly greater coal parameter alteration in high-rank coal. With the simulated coal’s porosity increasing from 5.12 to 6.63%, UCS strength is reduced by up to around 54.59% with gas CO2 adsorption and Young’s modulus and Poisson’s ratio are reduced and increased by around 70.43 and 73.07%, respectively. Furthermore, ScCO2 with a higher adsorption capacity results in greater damage and causes larger alterations of coal parameters.

The coal-swelling properties are controlled by molecular structure and surface free energy. Evalence is a dominant factor in controlling the intramolecular and volumetric swelling strain. With decreasing Evalence, the aromaticity and size of the aromatic structure decrease, affecting the intramolecular pores and volumetric swelling strain. Enon maintains the stability of coal structure and mechanical properties, and the decrease of Enon indicates that the distance between the surface molecules becomes larger, leading to intermolecular pores increasing and mechanical properties weakening.