Study on the pore-fracture system of low-rank coalbed methane reservoirs in the Bayanhua Sag, Erlian Basin

1 Introduction

China possesses abundant low-rank coalbed methane (CBM) resources with significant exploration potential. The total reserves amount to approximately 14.7 trillion cubic meters, constituting about 40% of the country’s total CBM reserves (Xin et al., 2019; Li et al., 2024a). As the global energy structure transitions and the demand for low-carbon development increases, low-rank CBM, one of China’s richest unconventional natural gas resources, is gaining strategic significance in exploration and development (Muther et al., 2022; World Energy Council, 2016; Tao et al., 2025). The Erlian Basin in Inner Mongolia is a major area for the occurrence low-rank CBM, characterized by abundant coal reserves, thick coal seams, widespread distribution, good reservoir porosity, and stable caprock. The CBM resources in coal seams buried depths shallower than 2,000 m amount to 1.2 trillion cubic meters (Li et al., 2019a; Liu et al., 2005). Nevertheless, the typical low-rank coal in the basin is commonly characterized by low gas content, low gas saturation, and low permeability, which collectively impose substantial technical challenges on economic extraction. The efficient development of low-rank CBM is fundamentally constrained by the dual factors of pore structure connectivity and heterogeneity (Moore, 2012; Zhang et al., 2022). These characteristics directly govern the entire process of CBM behavior, including adsorption, desorption, and seepage (Chattaraj et al., 2016; Clarkson and Bustin, 1999). In this context, multi-scale quantitative characterization of coal pore structure and its heterogeneity is a critical prerequisite for accurately elucidating gas migration patterns within the reservoir and optimizing development strategies (Pant et al., 2015; Karimpouli et al., 2020; Li et al., 2023; Yang et al., 2023).

In recent years, various advanced technologies have been applied to characterize the pore network of coal reservoirs. Direct imaging technologies can visually present pore morphology (Chalmers et al., 2012; Loucks et al., 2012). Among them, scanning electron microscopy (SEM) is widely used for morphological characterization and genetic type classification of unconventional reservoir pores (Li X. C. et al., 2019). The comprehensive application of low-temperature nitrogen adsorption (LT-N₂), low-temperature carbon dioxide adsorption, high-pressure mercury intrusion porosimetry (MIP), and low-field nuclear magnetic resonance (LF-NMR) has become a mainstream technical method for systematically characterizing the comprehensive pore structure of coal reservoirs and quantitatively evaluating pore connectivity (Freedman et al., 2001; Jiang et al., 2019; Li et al., 2012). Yao et al. (2010a) improved the quantitative conversion model between T₂ relaxation time and pore size in LF-NMR, demonstrating that this technique can efficiently obtain comprehensive pore size distribution and connectivity parameters of coal reservoirs under non-destructive conditions, thereby significantly improving the efficiency and accuracy of unconventional reservoir pore characterization. Based on this, Li N. N. et al. (2024) and Tao et al. (2018) further clarified the characterization advantages of different analytical techniques within different pore size ranges in coal through cross-validation of various experimental methods, thereby establishing a comprehensive quantitative characterization system covering the entire pore size range.

In terms of quantifying pore heterogeneity, fractal dimension has become a core indicator for quantitatively evaluating the complexity and heterogeneity of coal reservoir pore structure due to its ability to accurately characterize the spatial heterogeneity of complex porous media. Pfeifer and Avnir (1983) first introduced fractal geometry theory into the evaluation of porous media surfaces. The FHH fractal dimension calculation model they established remains a common method for characterizing pore heterogeneity in LT-N₂ experiments to this day. The Menger sponge fractal model proposed by Katz and Thompson (1985) can accurately infer a two-dimensional pore image into a three-dimensional heterogeneous pore network, providing a key theoretical approach for quantifying the three-dimensional heterogeneity of complex pore systems in low-rank coal. Wang et al. (2018) and Li et al. (2015) applied these classical fractal theories to coal reservoir research and verified the significant correlation between fractal dimension and coal pore heterogeneity as well as gas adsorption capacity. Zou et al. (2021) further proposed that by comparing the fractal dimension results obtained from different experimental methods, the multi-scale pore characteristics of low-rank coalbed methane reservoirs can be more comprehensively characterized.

The Bayanhua Sag serves as a key experimental zone for developing low-rank CBM in the Erlian Basin. It possesses abundant low-rank CBM resources, suitable burial depths, substantial coal seam thicknesses, and six existing CBM wells, providing a solid research foundation (Li et al., 2022). Previous research by Li et al., (2022) completed a regional macroscopic geological evaluation of the Bayanhua structure. This study focuses on the micropore-fracture system of low-rank coal reservoirs in the Bayanhua Sag and conducts refined quantitative characterization targeting the unique pore heterogeneity of low-rank coal. By combining SEM, LT-N₂, MIP, and LF-NMR, the morphology, structure, and connectivity of the multi-scale pore-fracture system in coal reservoirs were precisely characterized. Coupled fractal evaluation was conducted for adsorption pores and seepage pores, providing a comprehensive quantitative characterization of pore heterogeneity. Based on this, this study identified significant controlling factors that dominate pore development at different pore size scales and clarified the coupled control mechanism behind the fractal characteristics of adsorption pores and seepage pores.

Built on the innovative design of the above multi-scale coupled characterization, this study has achieved core advances that feature both regional practical applicability and global universality. This research outcome can provide a directly reusable refined pore-fracture characterization paradigm for widely distributed continental low-rank coal blocks worldwide. It also addresses the long-standing industry common issue that traditional single-scale fractal characterization cannot fully reflect the heterogeneity of low-rank coal reservoirs. The proposed methodological framework can be directly migrated and applied to reservoir evaluation of other low-rank coal basins worldwide. The revealed control mechanism of pore structure in low-rank coal “three-low” reservoirs can provide a universal theoretical reference for the efficient development of similar low-rank CBM reservoirs across the globe, and has universal guiding significance for promoting large-scale utilization of low-rank coal unconventional natural gas resources. The relevant outcomes will provide a solid scientific basis for the efficient development of low-rank CBM in the Bayanhua Sag and other blocks with similar geological conditions.

2 Geological setting

The Bayanhua Sag is tectonically situated in the northeastern part of the Erlian Basin, and is geomorphologically characterized as a narrow, elongated half-graben‌ with a consistent NE-NNE structural trend (Figure 1). The main axis of the syncline strikes at NE 30°–40°, with its axis trace offset towards the northwest. All secondary anticlines and synclines within the study area exhibit a uniform NE-SW orientation, with relatively limited spatial extent. The strata on both flanks of the syncline show distinct asymmetry. The northwest flank strata are gentler, with a dip angle of approximately 6°, while the southeast flank strata are steeper, with most strata dipping more than 15°. The western flank is dominated by major boundary faults. The faults in the area are well developed across the study area, and all identified faults are classified as normal faults. Among these structures, the NE-striking faults that are sub-parallel to the strike of the syncline axis constitute the dominant fault population in this region.

The stratigraphic sequence of the study area, from oldest to youngest, is as follows: the Upper Jurassic Xing’anling Group (J₃xn), Lower Cretaceous Aershan Formation (K₁a), Lower Cretaceous Tenggeer Formation (K₁t), Paleogene (E₃), Neogene (N₂), and Quaternary (Q). The main coal-bearing strata are the Lower Cretaceous Tenggeer Formation. This formation can be further divided into four distinct layers from bottom to top: the basal conglomerate layer, lacustrine mudstone layer, coal-bearing clastic rock layer, and the upper mudstone layer. Five commercially recoverable coal seams were identified: 5-1, 5-2, 6-1, 6-4, and 6-7. In addition, eight locally recoverable coal seams exist: 5-3, 5-4, 5-5, 6-2, 6-3, 6-5, 6-6, and 6-8. The coal seams are predominantly deposited in sedimentary facies of braided river delta plains, marginal shallow lakes, and fan delta plains. The thick mudstone interval that forms the immediate roof of the coal-bearing succession is regionally stable, laterally continuous, and exhibits excellent sealing properties. It is considered as the effective regional caprock for the CBM reservoir system in this area.

3 Materials and methods

3.1 Samples and experiments

The coal samples used in this study were collected from four coalbed methane exploration wells in the Bayanhua depression. The coal cores were obtained using wireline coring technology, while the desorbed gas samples were collected through conventional water flooding. After on-site collection, all coal samples were immediately wrapped tightly with polyethylene plastic film. The desorbed gas samples were sealed and stored upside down in glass bottles saturated with saline water. All pretreated samples were then transported to the laboratory for subsequent experimental analysis.

The organic macerals and vitrinite reflectance of the coal samples were determined using a Leica DM2700P polarizing microscope, in accordance with the Chinese National Standards GB/T 6948-2008 and GB/T 8899-2013, respectively. The proximate analysis was conducted using an electric drying oven with forced convection and a muffle furnace, following the Chinese National Standard GB/T 212-2008. Porosity and permeability were measured using an HKY-300 fully automated pore-permeability integrated tester, complying with the Chinese National Standard GB/T 29,172-2012.

To visually characterize the pore and fracture microstructure of the coal reservoirs, thin sections of the coal sample were prepared and observed using a FEI Quanta TM 250 SEM and a Zeiss Axio Scope A1 polarimetric microscope, following the Chinese Coal Industry Recommended Standard MT/T 968-2005. For continuous and comprehensive pore size distribution characterization, three complementary techniques were employed: LT-N₂, MIP, and LF-NMR. The analyses were performed using an ASAP 2460 fully automated gas adsorption analyzer, an AutoPore 9500 mercury porosimeter, and a ReCore-5000 NMR rock analyzer, respectively. The testing procedures adhered to the Petroleum and Natural Gas Industry Standards of China: SY/T 6154-1995, SY/T 5346-2005, and SY/T 6490-2023, respectively.

• ‌Specific experimental steps:

LT-N₂: Prior to analysis, the coal sample underwent degassing treatment through vacuum heating. It was first heated at 90 °C for 1 h, followed by heating at 110 °C for 10 h, to eliminate moisture and pre-adsorbed gases. The sample tube was immersed in liquid nitrogen at 77 K, and the nitrogen adsorption isotherm was collected point by point under conditions with a relative pressure (P/P0) ranging from 0.01 to 0.995.

MIP: Agate mortars were used to gently separate intact coal matrix fragments, and particles of 2–5 mm size were precisely screened through standard test sieves. The screened coal particles were dried at 105 °C for at least 12 h until they reached a constant weight, then loaded into a sample holder, and the system was fully evacuated before the formal test. Mercury immersion was carried out under gradually increasing pressure, followed by mercury extrusion under gradually decreasing pressure. Notably, considering the low-rank coal’s loose texture and low mechanical strength, the termination pressure was set at 16,922 psi to prevent sample damage.

3.1.1 LF-NMR

The experiment was conducted on crushed coal particles with a diameter greater than 2 mm. The samples were pre-dried to constant weight at 105 °C and then vacuum-saturated with deionized water free of paramagnetic impurities for 24 h. The nuclear magnetic resonance measurement parameters were set as follows: the echo time was set to 0.2 milliseconds, the repetition time was set to 6000 milliseconds, and an average of 128 scans was performed to ensure sufficient signal-to-noise ratio for the T₂ spectrum.

3.2 Fractal method

This study employs the FHH fractal model to characterize the pore fractal features derived from LT-N₂ experiments, with the calculation formula expressed as follows (Pfeifer and Avnir, 1983):

lnV = A ln ln p / p 0 + C (1)

D = A + 3 (2)

Where: p is the equilibrium pressure, MPa; V is the gas adsorption volume at the corresponding equilibrium pressure, cm³/g; P₀ is the saturated vapor pressure of the adsorbate, MPa; A is the slope of the fitted line, C is a regression constant; D is the fractal dimension.

Based on the Menger theoretical framework, a quantitative correlation between the mercury intrusion difference (dV/dp) and the fractal dimension has been established to analyze the fractal characteristics of the seepage pore structure. The corresponding calculation formula is as follows (Katz and Thompson, 1985):

lg d V / d p = K lg ⁡ p (3)

D = K + 4 (4)

Where: V is the cumulative mercury intrusion volume, cm³/g; p is the applied mercury intrusion pressure, MPa.

Based on LF-NMR technology, fractal dimension has been widely used to quantitatively characterize the micro-pore structure of unconventional reservoirs. The widely adopted calculation formula is as follows (Yao et al., 2010b):

lg ⁡ S = K lg ⁡ T 2 + C (5)

D = 3 ‐ K (6)

Where: T₂ is the transverse relaxation time, ms; S is the percentage of the cumulative pore volume with transverse relaxation time smaller than a specific T₂ value, relative to the total pore volume of the sample.

4 Characteristics of coal reservoirs

Within the study area, the vitrinite content in coal samples ranges from 91.14% to 100% (mean: 98.4%), providing a favorable material basis for hydrocarbon generation and gas adsorption (Table 1). The inertinite content ranges from 0% to 8.1% (mean: 1.3%), while the liptinite content ranges from 0% to 3.1% (mean: 0.3%). The maximum vitrinite reflectance (R_{o, max}) ranges from 0.29% to 0.59% (mean: 0.40%). The coal types are mainly lignite and long-flame coal.

TABLE 1. Basic parameters of the coal samples from the Bayanhua Sag.

min ⁡ ‐ ⁡ max mean sample count.

The results of proximate analysis indicate that the moisture content of coal samples ranges from 4.34% to 16.45% (mean: 8.32%), falling within the optimal “ moisture content window” for efficient CBM development (Ma et al., 2024) (Table 1). The ash yield is generally at a low-to-medium level, ranging from 3.47% to 39.49% (mean: 19.81%), which is conducive to methane adsorption. The volatile matter content ranges between 25.18% and 48.98% (mean: 38.9%), indicating a low coalification degree, which is beneficial for biogas generation and adsorption.

The porosity measured in coal samples ranges from 7.13% to 44.69% (mean: 21.90%), indicating relatively high porosity for low-rank coal reservoirs. The permeability ranges from 0.029 mD to 3.52 mD (mean: 0.95 mD), reflecting significant permeability variability and strong reservoir heterogeneity. The injection/falloff well testing results show that the permeability of the 5-2 coal seam (919 m) and 6-3 coal seam (1016 m) in Well B-T1 is 0.692 mD and 0.135 mD, respectively. For Well B-T2, the measured permeability of the 5-1 coal seam (818 m) and 6-7 coal seam (988 m) is 0.62 and 0.2, respectively. These field test results indicate a significant negative correlation between coal seam burial depth and permeability. Comprehensive analysis confirms that the study area is a typical low-permeability coal reservoir, requiring targeted reservoir modification to achieve effective commercial coalbed methane extraction.

5 Coal reservoir pore-fracture system

5.1 Multilevel pore micro-morphology

SEM observations reveal that the coal seams within the study area primarily develop plant tissue pores, accompanied by a small amount of mineral intergranular pores (clay minerals, brittle minerals), intergranular pores, and mold pores (Figure 2). Although plant tissue pores exhibit relatively poor connectivity, their abundance and large specific surface area make them the primary adsorption sites for methane molecules. Reflectance polarization microscopy further demonstrates the coexistence of original cleat, shear fractures, and tensile fractures within the coal matrix (Figure 3). Statistics show that the main fractures have a length of 0.005–3.3 cm, a height of 0.005–2.8 cm, a width of 1–440 μm, and a density of 2.6–6.6 fractures per centimeter; secondary fractures have a length of 0.005–2.3 cm, a height of 0.005–2.6 cm, a width of 1–200 μm, and a density of 2.5–15.8 fractures per centimeter. These fractures are mostly isolated, short fractures with poor connectivity, unable to form continuous three-dimensional seepage channels. Local observations indicate the presence of calcite mineral fillings, which may cause local permeability deterioration and strong heterogeneity in the reservoir. The connectivity of fractures ranges from poor to moderate, which is generally unfavorable for the percolation and commercial extraction of CBM.

5.2 Pore structure and fractal characteristics based on LT-N₂ method

According to the Hodot pore size classification method (Hodot, 1961), coal pores are divided into four categories: micropores (pore size <0.01 μm), transition pores (0.01 μm ≤ pore size <0.1 μm), mesopores (0.1 μm ≤ pore size <1 μm), and macropores (pore size ≥1 μm). LT-N₂ experiments measure pore size ranging from 1.7 to 300 nm, effectively revealing the pore structure of micropores and transition pores in coal reservoirs.

Within the study area, the average pore diameter of the micropores-transition pores system ranges from 3.61 to 7.85 nm (mean: 5.54 nm), with total pore volume between 0.02 and 0.11 mL/g (mean: 0.06 mL/g), and specific surface area ranging from 9.75 to 121.43 m²/g (mean: 47.18 m²/g) (Table 2). Micropores dominate the specific surface area, accounting for 84.14%–99.18% (mean: 95.28%), while transition pores contribute the majority of the total pore volume, ranging from 5.6% to 94.4% (mean: 63.38%). Notably, the 6-5 coal seam and 6-7 coal seam exhibit a significant reduction in transition pores proportion, while the proportion of micropores increases accordingly, which is closely related to the burial depth and the corresponding high in-situ stress environment.

TABLE 2. LT-N₂ results of coal samples from the Bayanhua Sag.

The N₂ adsorption-desorption isotherms of all coal samples can be divided into two distinct types, with Type I being the predominant one (Figure 4). Taking the 5-1 coal seam as an example, the characteristic of Type I isotherm is that the adsorption curve rises steadily and accelerates at higher relative pressures (Figure 4A). The corresponding desorption curve exhibits a pronounced hysteresis loop and drops sharply when the relative pressure (P/P0) is approximately 0.5. The Type I pore system is predominantly micropores, featuring ink-bottle pores (with a bottle neck radius typically of 4 nm), exhibiting a “single-peak” pore size distribution. This pore structure is favorable for CBM adsorption and storage, but it may increase the difficulty of CBM desorption and diffusion (Figure 4B). Taking the 5-3 coal seam as an example, the overall adsorption curve trend of Type II isotherm is similar to that of Type I (Figure 4C). Its desorption curve also shows a pronounced hysteresis loop, but lacks the plateau segment observed in Type I. The Type II pore structure is also dominated by micropores, but exhibits a “double-peak” pore size distribution within a relatively large average range of pore diameters (Figure 4D). These pores are mainly open cylindrical or slit-shaped pores, which are conducive to the full process of CBM adsorption, desorption, and diffusion.

The FHH fractal model defined in Equations 1, 2 was employed to perform linear fitting on lnV and ln [ln (p₀/p)] datasets of coal samples, and the calculated fractal dimensions are summarized in Table 2. The fractal dimension (D1) ranges from 2.65 to 2.82 (mean: 2.73), and the determination coefficient (R1²) ranges from 0.78 to 0.98. This indicates that the adsorption pore (≤100 nm) surface is rough and the pore structure is highly complex, providing a large number of effective sites for methane adsorption, thus significantly promoting the adsorption of methane.

5.3 Pore structure and fractal characteristics based on MIP method

MIP can characterize pore sizes ranging from 3.29 to 360,000 nm, covering the full size range of transition pores, mesopores, and macropores. The statistical results of MIP tests are summarized in Table 3. The total pore volume ranges from 0.535 to 0.980 ml/g (mean: 0.818 ml/g), the specific surface area ranges from 0.185 to 4.716 m²/g (mean: 1.460 m²/g), the mean throat radius ranges from 4.72 to 17.61 μm (mean: 12.50 μm), and porosity ranges from 7.30% to 25.10% (mean: 18.04%). The mercury withdrawal efficiency ranges from 34.0% to 64.0% (mean: 49.24%), which is relatively low, indicating uneven pore size distribution and poor pore connectivity. The total pore volume is primarily contributed by transition pores and macropores, with transition pores accounting for 37.48%–81.03% (mean: 58.49%), and macropores accounting for 12.20%–53.55% (mean: 32.90%). Mesopores are relatively underdeveloped and have poor connectivity, accounting for only 5.18%–13.96% (mean: 8.67%) of the total pore volume. The mercury intrusion curve exhibits a typical three-stage pattern. In the initial stage (pressure <0.1 MPa), mercury intrusion volume increases rapidly, indicating the presence of numerous pores with diameters exceeding 12.6 μm. In the intermediate pressure stage (0.1–10 MPa), the curve becomes relatively flat, and the mercury intrusion rate decreases significantly, reflecting the underdevelopment of pores in the pore size range of 0.2–12.6 μm. When the pressure exceeds 10 MPa, mercury intrusion volume rises sharply, indicating a high degree of development of transition pores (Figure 5).

TABLE 3. MIP results of coal samples from the Bayanhua Sag.

The mercury intrusion-extrusion curves of all coal samples can be divided into four types, with Type I and Type II being the main ones, followed by Type III and Type IV (Figure 5). All mercury extrusion curves show that mercury is difficult to be discharged during the low-pressure stage, and mercury saturation is mainly concentrated in the high-pressure stage.

Type I (Figure 5A): At pressures above 10 MPa, the mercury intrusion and extrusion curves show minimal disparity. As the pressure decreases, the hysteresis loop gradually expands, corresponding to a low mercury withdrawal efficiency. The pore-throat system is narrow and uniformly distributed, making it difficult for mercury to flow out after intrusion. This type is dominated by micropores with a high pore-to-throat ratio, mostly semi-closed or closed pores, poor connectivity, and high fluid migration resistance.

Type II (Figure 5B): the mercury intrusion and extrusion curves are close to each other, indicating a relatively high mercury withdrawal efficiency. However, the curve characteristic of “difficult mercury discharge at low pressure, continuous mercury intrusion at high pressure” indicates a poorly sorted transitional pore structure. Despite the presence of macropores, they are restricted by narrow throats, and the overall pore system is still dominated by neck pores, resulting in limited permeability.

• ‌Type III (Figure 5C): the mercury intrusion-extrusion curves show a significant hysteresis loop, indicating low mercury withdrawal efficiency. It presents a typical “easy intrusion, difficult discharge” structure, reflecting the development of ink-bottle pores, which are semi-closed pores. Although these pores possess a certain gas storage capacity, their permeability is significantly limited.
• ‌Type IV (Figure 5D): At pressures above 10 MPa, the difference between mercury intrusion curves and extrusion curves is minimal. As the pressure decreases, the hysteresis loop significantly expands, resulting in a low mercury withdrawal efficiency. Unlike the other three types, mercury intrusion in type IV mainly occurs within the pressure ranges from 1 to 10 MPa, and the amount of mercury intrusion above 10 MPa is very limited. Although mercury can intrude into some pores during the medium pressure stage, it is extremely difficult to be discharged, reflecting the typical characteristics of closed pores or blind holes with high tortuosity‌.

Overall, the pore structure typically exhibits characteristics of “difficult intrusion and difficult discharge”, and is dominated by a pore system with narrow pore throats, high pore-to-throat ratio, and poor connectivity.

The Menger sponge fractal model defined in Equations 3, 4 was employed to perform linear fitting on lgp and lg (dV/dp) datasets of coal samples, and the calculated fractal dimensions are listed in Table 3. The fractal dimension of adsorption pores is all greater than 3, which exceeds the theoretical range of fractal dimension porous media (2–3). This indicates that the MIP method is not suitable for fractal characterization of adsorption pores. The fractal dimension (D2) of seepage pores (pore size >100 nm) ranges from 2.63 to 2.99 (mean: 2.82), and the corresponding determination coefficient (R2²) ranges from 0.90 to 0.99. The seepage pores system exhibits pronounced fractal characteristics, with a complex and highly heterogeneous pore structure, which is conducive to CBM adsorption.

5.4 Pore structure and fractal characteristics based on LF-NMR method

LF-NMR determines pore size distribution by analyzing hydrogen nuclear relaxation signals. It is an advanced non-destructive testing technology for coal reservoir characterization. The distribution pattern of transverse relaxation T₂ spectrum in LF-NMR can effectively provide quantitative information on the pore and fracture structures in coal reservoirs. Specifically, the continuity of spectral peaks directly reflects the connectivity between corresponding pores and fractures, while the transverse relaxation time corresponds to different pore types in coal: adsorption pores (0.1 ms < T₂ ≤ 10 ms), seepage pores (10 ms < T₂ ≤ 100 ms), and fractures (T₂ > 100 ms) (Yao and Liu, 2013). In the T₂ spectra of coal, distinct relaxation time intervals correspond to specific pore categories: micropores (T₂: 0.5–2.5 ms), mesopores (T₂: 20–50 ms), and fractures (T₂ > 100 ms) (Yao et al., 2010b).

The T₂ spectrum of the coal samples in the study area exhibits a typical bimodal pattern, with the first peak distributed in the relaxation time ranging from 0.1 to 3 ms, and the second peak distributed within a range of 10–100 ms. The discontinuity between the first and second peaks indicates poor connectivity between the corresponding pore size ranges (Figure 6). The porosity ranges tested by LF-NMR are 12.23%–17.72% (mean: 15.56%). The pore volume is primarily contributed by micropores, accounting for 77.15%–88.29%, followed by mesopores and macropores, accounting for 11.11%–18.15%, with no significant fracture development (Table 4).

TABLE 4. LF-NMR results of coal samples from the Bayanhua Sag.

Linear fitting was performed on the lgT₂ and lgS datasets of coal samples according to Equations 5, 6, and the corresponding fractal dimensions are listed in Table 3. The fractal dimension of adsorption pores (D3) ranges from 0.64 to 2.01 (mean: 1.37), with the determination coefficient (R3²) ranging from 0.68 to 0.84. The fractal dimension D4 of the seepage pores ranges from 2.88 to 2.95 (mean: 2.90), with the determination coefficient (R4²) ranging from 0.91 to 0.98. The results indicate that seepage pores exhibit more stable and regular fractal features compared with adsorption pores. The seepage pores system demonstrates highly complex structures, possibly containing numerous narrow throats, tortuous channels, or secondary pores. Compared with D3, D1 is significantly higher and more stable, with a larger determination coefficient. This indicates that the LT-N₂ method is more efficient and reliable for fractal characterization of adsorption pores. The R ² values for seepage pores fractal fitting obtained by both MIP and LF-NMR methods are above 0.9, indicating that both methods exhibit good linear fitting degrees for the fractal characterization of seepage pores. The seepage pore structure satisfies statistical self-similarity under both testing systems, and the fractal calculation results are reliable. The stability of LF-NMR results for fractal characterization of percolation pores is significantly better than that of MIP, with less interference from sample heterogeneity and testing operational errors. However, the D2 value range of MIP method is relatively wide, with some samples having D2 values as low as 2.63. The stability of LF-NMR results in the fractal characterization of seepage pores is significantly better than that of MIP, and it is less affected by sample heterogeneity and testing operational errors. The strong heterogeneity of seepage pores in coal samples from the study area has a significant negative impact on CBM seepage and migration, which is a key geological factor restricting the efficient development of such low-rank CBM reservoirs.

5.5 Full-scale pore size structure characteristics

Based on previous studies on multi-method coupled pore structure characterization (Li N. N. et al., 2024; Tao et al., 2018; Lin et al., 2019), this study systematically evaluated the advantages and limitations of different pore testing methods and full-scale pore size splicing techniques. In the study area, centimeter-scale main fractures and fractures with opening widths up to hundreds of micrometers were developed in the coal samples, but these fractures were not effectively detected by LF-NMR. Combined with the unique pore development characteristics of coal seams in the study area, LT-N₂ was employed to characterize micropores, while MIP was used to characterize transition pores, mesopores, and macropores. By organically combining the two sets of data, a comprehensive quantitative characterization of the full pore size range of coal reservoirs was achieved (Table 5).

TABLE 5. Pore size distribution reconstruction results for coal samples from the Bayanhua Sag.

The total pore volume is primarily contributed by transition pores, accounting for 23.69%–80.29% (mean: 55.96%), followed by macropores that account for 12.09%–53.05% (mean: 32.71%). Mesopores are relatively underdeveloped, accounting for only 5.13%–38.16% (mean: 10.10%), exhibiting poor connectivity. Micropores volume accounts for only 0.42%–4.68% (mean: 1.39%), but it contributes the majority of the specific surface area, ranging from 86.29% to 99.28% (mean: 96.09%).

6 Controlling factors

6.1 Burial depth of coal seam

In Bayanhua Sag, there is no significant correlation between coal seam burial depth and total pore volume, while a significant positive correlation exists between burial depth and specific surface area of micropores (Figure 7A). As burial depth increases, compaction compresses the volume of primary pores, while the hydrocarbon generation of organic matter and dissolution by diagenetic fluids generate a large number of new secondary pores, especially micropores and transition pores. In addition, the spatial heterogeneity of coal’s internal mineral components, filling states, and fracture network significantly interferes with and weakens the sole dominant control of the temperature-pressure field on pore volume evolution. The low-rank coal in the study area is characterized by high initial porosity and strong matrix plasticity. Thus, changes in stress increment and short-distance burial depth are insufficient to significantly alter the pore volume of various pore size segments in a directional manner. For the specific surface area, continuous compaction continuously compresses macropores and mesopores, significantly increasing the relative proportion of micropores. As the pore size decreases, the specific surface area of the pore system increases exponentially. During this geological evolution process, the heterogeneity of adsorption pores continues to increase, while the fractal dimension gradually rises (Figure 7B). Meanwhile, the originally randomly distributed seepage pores and microfractures gradually transform into straighter and more connected flow channels, leading to a decrease in the heterogeneity of seepage pores and a continuous decline in the fractal dimension (Figure 7B).

6.2 Coal quality parameters

Proximate analysis parameters of coal samples in the study area are only correlated with micropores volume. Ash yield is positively correlated with micropores volume (Figure 8A), as relatively low ash yield is more favorable for the development of micropores and transition pores, as well as the formation of complex structures (Yan et al., 2022)‌. Volatile matter and fixed carbon content both exhibit significant negative correlation with micropores volume (Figures 8B,C). Since the study area is in the early stage of coalification, the molecular structure of organic matter has not yet completed the rearrangement process to form a stable micropores network. Instead, the original pore space has been partially compressed and reduced due to overlying compaction (Chang, 2024; Huang et al., 2025). This also leads to no significant correlation between the heterogeneity of adsorption pores and industrial analysis parameters. The heterogeneity of seepage pores is positively correlated with ash yield and fixed carbon content, while negatively correlated with volatile matter content (Figure 8D). The increase in ash yield leads to an increase in mineral pores, while the increase in fixed carbon content leads to secondary organic matter pores. These combine with the originally compressed pores to form a complex pore size distribution across multiple scales. When the volatile matter content is high, a large amount of undecomposed organic matter fills the pore gaps, smoothing out the differences in pore size distribution.

6.3 Organic macerals

In the study area, the vitrinite content in coal samples exhibits no significant control over the overall pore structure, possibly due to the abnormally high background value of vitrinite content (Deng et al., 2022). As the inertinite content increases, both the pore volume and specific surface area of transition pores present a trend of decreasing first and then slightly increasing, with a relatively small overall fluctuation. The specific surface area of micropores gradually decreases with the increase of inertinite content (Figures 9A,B). Vitrinite content is significantly positively correlated with the fractal dimension of adsorption pores, while it is significantly negatively correlated with the fractal dimension of seepage pores (Figure 9C). The effect of inertinite on pore fractal dimensions is opposite to that of vitrinite (Figure 9D). During the coalification process, vitrinite tends to generate a large number of secondary micropores, significantly increasing the structural complexity of adsorption pores (Wang et al., 2025; Zhang et al., 2025c). In contrast, inertinite itself has a rough internal pore surface, with macropores (>50 nm) as its dominant pore type, and its inherent pore structure is relatively complex (Giffin et al., 2013).

7 Conclusion

• Full-scale pore structure characteristics of coal reservoirs in the Bayanhua Sag.

The coal reservoirs of the Bayanhua Sag are characterized by significant development of plant tissue pores, with fractures primarily consisting of original cleat, shear fractures, and tensile fractures. The pore morphology is mainly closed and semi-closed pores, resulting in limited permeability. The adsorption pores morphology is dominated by ink-bottle pores, typically with a bottle neck radius of approximately 4 nm and exhibiting a “single-peak” distribution in pore size. Micropores constitute the majority of the total specific surface area, while transition pores and macropores contribute to the bulk of the total pore volume. The underdevelopment of mesopores results in poor connectivity between adsorption and seepage pores. Overall, the pore structure exhibits the typical characteristic of “difficult intrusion and difficult discharge”, mainly developing pore systems with small pore throats, high pore-throat ratios, and poor connectivity. The comprehensive pore characterization of this system fills the long-standing basic data gap in the study of coal reservoir pore structure in the Bayanhua Sag, providing empirical evidence for subsequent precise evaluation of CBM resources and optimization of development parameters.

• Fractal characteristics and heterogeneity of pore structure.

The pore structure of coal reservoirs is complex and highly heterogeneous. Both the adsorption pores and permeability pores exhibit obvious fractal characteristics, with fractal dimensions ranging from 2.65 to 2.82 and 2.88 to 2.95, respectively. This provides key parameters for the subsequent precise evaluation of CBM resources and the matching analysis of reservoir-forming elements, and also serves as a reference for the evaluation of pore structure in similar low-rank coal reservoirs in the Erlian Basin. With the increase in coal seam burial depth and vitrinite content, the heterogeneity of adsorption pores increases, while the heterogeneity of seepage pores decreases. As the content of inertinite increases, the heterogeneity of adsorption pores decreases, and the heterogeneity of seepage pores increases. The heterogeneity of seepage pores is positively correlated with ash yield and fixed carbon content, while negatively correlated with volatile matter content. This corrects the one-sided understanding in traditional research that “increasing burial depth will simultaneously enhance the heterogeneity of pores across all scales,” and clarifies the inverse evolution law of fractal characteristics between adsorption pores and permeable pores.

• Main controlling factors of pore development.

The burial depth of coal seams, coal quality parameters, and organic macerals of coal jointly control the pores strcture. The study area has an extremely high vitrinite content, with well-developed micropores in the vitrinite. The development of micropores shows a continuously enhancing trend with the increase of coal seam burial depth. Ash yield shows a significant positive correlation with micropore volume, while volatile matter and fixed carbon content exhibit an opposite trend to micropore volume. This study clarifies the internal control mechanism of pore development, provides a new practical index for the fine division of CBM sweet spots in the Bayanhua Sag, and can be extended to other low-rank coal-bearing basins with similar characteristics.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: John Hooker, University of the Incarnate Word, United States

Reviewed by: Anmin Wang, China University of Mining and Technology, Beijing, China

Yulong Liu, Yangtze University, China