Permeability Loss of Bituminous Coal Induced by Water and Salinity Sensitivities: Implications of Minerals' Occurrence and Pore Structure Complexity.

Water sensitivity (WS) and salinity sensitivity (SS) are key issues to be investigated for instructing coalbed methane (CBM) production. This work studied the influences of minerals and pores on WS and SS of medium-volatile bituminous coal (MVBC) and highly volatile bituminous coal (HVBC) deposited in northwestern China by detecting and observing minerals using the TESCAN Integrated Mineral Analyzer, simulating WS and SS, and characterizing pore structural complexities using rate-controlled mercury penetration. The results show that (1) kaolinite is mainly distributed as irregular particles or fragile aggregates attaching on the bedding surface or filling in meso-pores or transition pores, showing a high potential for detachment; (2) MVBC and HVBC in this study are characterized as medium to weak WS and weak SS, respectively; (3) for HVBC during the WS or SS process, kaolinite distributed in meso-pores or transition pores first detaches and then migrates to the narrow throat of macro-pores and super macro-pores, leading to volume decreases of macro-pores and super macro-pores and loss of permeability; and (4) kaolinite filling in macro-pores of MVBC detaches, then migrates, and finally deposits in super macro-pores after WS and SS, leading to losses of super macro-pore volume and permeability. Results of this study can enhance the scientific knowledge on WS and SS of coal during CBM development.

Introduction

Coalbed methane (CBM) development is regarded as an effective method for supplying coal-based clean energy and alleviating the greenhouse effect induced by CH4 discharge during coal mining.[1−5] In 2020, 5.76 × 109 m3 CBM was produced in China (on the ground surface), and recently, the CBM industry in northwestern China has developed rapidly. A vertical well located in the southern Junggar Basin[6] yielded a peak production of 17,125 (m3/d). A CBM development block named Baiyanghe,[7] with an annual gas production of 30 Mm,[3] has been established as the first demonstration base for CBM utilization in the southern Junggar Basin. Although progress has been achieved, the production of a significant number of CBM wells in northwestern China was lower than 500 m3/d, which need to be urgently improved for better economic performance.

Reservoir damage can impede fluid flow during oil and gas production, leading to the generation of low-yield wells.[8,9] The contact between minerals and externally invaded fluids can result in decreased seepage capacity and permeability of the coal reservoir.[10] Minerals of the reservoir may be sensitive to water, salt, acid, and alkali.[11,12] Currently, research studies on those sensitivities controlled by mineral migration and pore structure complexities of bituminous coal are insufficient. Low- and high-salinity water is applied in CBM drilling and fracturing processes in northwestern China; hence, the water sensitivity (WS) and salinity sensitivity (SS) of coal reservoirs are to be urgently investigated for instructing the work fluid design there.

Mineral migration, swelling, and aggregation were thought to be the vital factors controlling WS and SS damage degrees. Fang et al. found that the number of small pores increases and that of big pores decreases after WS due to clay migration, leading to porosity reduction and permeability impairment.[13] Alhuraishawy et al. reported that the detachment of clay particles after WS has a significant effect on the permeability reduction of a reservoir.[14] Wang et al., Zhang et al., Tao et al., and Zhao et al. reported that WS mechanisms are hydration and swelling of clay minerals and particle migration.[10,15−17] Yang et al. first presented and visualized conclusive evidence for fines migration during the WS process by applying X-ray micro-computed tomography.[18] Zhao et al. reported that the main mechanisms of SS are that fines can easily detach, migrate, and finally plug the pore throats under a salt environment.[19] Zhang et al. presented that the change in the water film thickness, the enhancement of hydrophilia, the particle detachment, and the dissolution of quartz or albite contribute to WS damage under high-temperature conditions.[20] Wang et al. implemented that pH can control the change in the thickness of the water film during SS, leading to seepage space variation.[21]

The aforementioned recent research progress on WS and SS could be briefly summarized as that the detaching, expanding, dispersing, migrating, and depositing of minerals induced the permeability loss. However, three deficiencies still exist: (1) minerals’ migration pathways during WS and SS are still unclear; (2) the pore structure can control the reservoir damage degree;[22] however, it is unclear whether minerals’ plugging or pore structure complexity plays a dominating role in determining the WS or SS damage degree; and (3) WS and SS of bituminous coals deposited in northwestern China are scarcely investigated.

To address the aforementioned issues, the following content, aimed at revealing the implications of mineral migration and pore structure complexity on WS and SS of bituminous coals deposited in northwestern China, was analyzed: (1) medium-volatile bituminous coal (MVBC) and high-volatile bituminous coal (HVBC) of northwestern China were sampled; (2) mineral composition and occurrence characteristics were detected by the TESCAN Integrated Mineral Analyzer (TIMA) and scanning electron microscope as well as energy dispersive spectrometer (EDS); (3) WS and SS were simulated using HVBC and MVBC specimens, and low-field nuclear magnetic resonance (LFNMR) tests before and after WS and SS were conducted; (4) the relationship between the transverse relaxation time (T2) and pore diameter was determined, and then, space variations of pores on different scales and minerals’ migration pathways were discussed; (5) pore-throat structures were analyzed by rate-controlled mercury penetration (CMP), and the influences of the pore structure on WS and SS were revealed; and (6) WS of different ranked coals was discussed by combing predecessors’ published work. The results of this study can enhance the scientific understanding on the WS and SS of coal and provide instructions on CBM development in northwestern China.

Geological Background

HVBC is accumulated in the Mesozoic coal measures in the southern Junggar Basin, northwestern China, where samples for this study were collected. The southern Junggar Basin, with a length of 500 km and a width of 40–90 km,[23] is characterized by multi-depositional and multi-typed structural systems including reverse faults and a series of anticlines as well as synclines.[24] The Lower Jurassic Badaowan Formation (J1b) and the Middle Jurassic Xishanyao Formation (J2x) are the main coal-bearing strata.[25,26] The J2x primarily consists of conglomerates, fine-grained sandstones, siltstones, black mudstones, and coal seams.[27] The J1b mainly consists of glutenite, siltstone, silty mudstone, gray mudstones, and interbedded coal seams.[28] CBM resources are 3.54 × 1012 m3 in the southern Junggar Basin.[23] MVBC is reserved in the northwestern Turpan-Hami Basin located in northwestern China. Its dimensions are approximately 660 km from east to west and 60–100 km from north to south.[29] The Middle or Lower-Middle Jurassic Badaowan Formation (J1b) and the Xishanyao Formation (J2x) are the main coal-bearing strata.[30−32] CBM resources are 6.21 × 1012 m3 in the Turpan-Hami Basin.[33] To date, the CBM development in the Turpan-Hami Basin is at an early stage of exploration.

Methods

A pore-size distribution classification scheme proposed by Cai et al.,[34] stating that micro-pores, transition pores, meso-pores, macro-pores, and super macro-pores are 2–10, 10–100, 100–1000, 1000–10,000, and >10,000 nm in diameter, respectively, was applied in this research. Figure depicts the research workflow of this study: (1) the mineral composition and occurrence state were analyzed by TIMA and SEM, respectively, as well as EDS technologies; (2) the pore-size distribution was determined by combining corrected mercury intrusion porosimetry (MIP) data and low-temperature nitrogen adsorption (LTNA) data, and the relationship between the transverse relaxation time (T2) and pore diameter was determined by comparing the pore-size distribution and T2 spectrum obtained by the LFNMR test; (3) If WS and SS simulations as well as LFNMR tests before and after those simulations were conducted, then volume variations of pores with different diameters could be analyzed by transferring T2 to the diameter; and (4) influences of mineral migrations and pore structure complexity on WS and SS were discussed.

Figure 1

Work flow diagram of this study (abbreviations: TIMA = TESCAN Integrated Mineral Analyzer; MIP = mercury intrusion porosimetry; LTNA = low-temperature nitrogen adsorption; LFNMR = low-field nuclear magnetic resonance; CMP = rate-controlled mercury penetration; PSD = pore-size distribution; WS = water sensitivity; SS = salinity sensitivity; HVBC = high-volatile bituminous coal; MVBC = medium-volatile bituminous coal).

Sample Collection and Treatment

Coal blocks (30 cm × 30 cm × 30 cm, approximately) were sampled from the Wudong coal mine and Kuangou coal mine located in the southern Junggar Basin as well as the Ewirgol coal mine located in the northwestern Turpan-Hami Basin, northwestern China. The mean maximum vitrinite reflectance, proximate analysis, and coal composition tests were conducted according to the standards of ISO 7404-5-2009, ISO 17246-2010, and ISO 7404-3-2009, respectively (Table ). According to the American Society of Testing Materials Standard ASTM D388-2015, samples from the Wudong coal mine and Kuangou coal mine are HVBC, named as HVBC1 and HVBC2, respectively. Samples from the Ewirgol coal mine are MVBC, and these samples are named as MVBC1. Samples were treated into seven types from the same coal blocks, which were used for MIP, LTNA, LFNMR, WS and SS simulations, TIMA, SEM and EDS, and CMP, respectively. Details on these types of samples were introduced in Table . It should be noted that, to weaken the influences of heterogeneity, samples used for MIP, LFNMR, WS and SS simulations, and CMP were obtained by cutting the same long cylindrical sample drilled along the direction parallel to the bedding plane from the coal block.

Table 1

Results of Proximate Analysis and Coal Composition of Coal Samplesa

				proximate analysis (%)			coal composition (%)
sampling sites	coal type	macroscopic coal petrography	Ro,max (%)	Mad	Ad	Vdaf	V	I	E
Wudong	HVBC	semi-bright coal	0.72	2.53	4.12	32.24	34.2	62.2	3.6
Kuangou	HVBC	semi-bright coal	0.73	3.66	5.38	36.73	57.2	41.0	1.8
Ewirgol	MVBC	semi-bright coal	1.26	1.42	33.64	30.27	90.7	9.3	0.0

Abbreviations: HVBC = high-volatile bituminous coal; MVBC = medium-volatile bituminous coal; Ro,max = maximum reflectance of vitrinite, %; Mad = moisture content on an air-dried basis, %; Ad = ash content on a dried basis, %; Vdaf = volatile matters yield on dry and ash-free basis, %; V = vitrinite content, %; I = inertinite content, %; and E = exinite content, %.

Table 2

Allocation of Coal Samples for Different Testsa

targeted experiments	specifications
MIP	cylindrical cores 10 mm in height and 25 mm in diameter
LTNA	60–80 meshed powder coal samples
LFNMR	cylindrical cores 25 mm in diameter and 30 mm in length
WS/SS simulations	cylindrical cores 45 mm in height and 25 mm in diameter
TIMA	flat plate specimens 5–25 mm in length and 5 mm in thickness
SEM and EDS	small blocks with 10 mm × 10 mm × 3 mm in dimension
CMP	cylindrical coal cores 10 mm in height and 25 mm in diameter

Abbreviations: MIP = mercury intrusion porosimetry; LTNA = low-temperature nitrogen adsorption; LFNMR = low-field nuclear magnetic resonance; WS = water sensitivity; SS = salinity sensitivity; CMP = rate-controlled mercury penetration; SEM = scanning electron microscope; and EDS = energy dispersive spectrometer.

TIMA, SEM, and EDS

Mineral composition and distribution were characterized by TIMA, which can combine the back-scattered electron and energy-dispersive X-ray in conjunction with advanced image and pattern recognition analysis. In this study, samples were analyzed in the high vacuum mode using modal analysis under an accelerating voltage setting of 25 kV, a working distance setting of 15 mm, and a field size setting of 1500 μm. SEM and EDS tests were applied, and the samples were polished by argon ion polishing using the Hitachi IM4000 argon polishing device. SEM maps were observed, and EDS detections were conducted using a double-beam electron microscope system named FEI Helios 650.

WS and SS Evaluation

WS simulations using cylindrical cores were conducted according to the following steps: (1) samples were first vacuumed for 24 h after drying at 110 °C for 24 h, then they were saturated with formation water at 1 MPa for 24 h. The saturation time of 24 h was applied because the T2 signal of LF-NMR no longer increased after 10 h since the beginning of the saturating process; (2) initial permeability (kf) was measured by injecting formation water into the sample under a flow rate setting of 0.05 mL/min, which was far less than the critical velocity sensitivity flowing rates (2 mL/min) tested ourselves and 0.9–1.2 mL/min referenced by Tao et al.[35] of bituminous coals deposited in northwestern China. The confine pressure was set to 10 MPa for the purpose of simulating the in situ coal reservoir condition using an experimental setup displayed in Figure ; (3) after step (2), the experimental samples were subjected to the LFNMR test for the purpose of obtaining the T2 spectrum before WS; (4) work fluid with 75% formation water salinity (FWS) was injected into the core at a flow rate of 0.05 mL/min under the confine pressure of 10 MPa, then permeability was measured after the inlet pressure and the outlet pressure were steady; (5) permeability of the cylindrical cores during the injection of work fluids with 50% FWS, 25% FWS, and distilled water were measured subsequently; and (6) after step (5), the cylindrical cores were subjected to LFNMR again for the purpose of obtaining the T2 spectrum after WS to distinguish the pore space variation before and after WS simulation. Before conducting the whole experimental procedures, LFNMR of the same standard sample was tested three times and the T2 spectra of those three were the same, thus the instrument error of LFNMR can be eliminated.

Figure 2

Experimental setup of WS and SS simulations.

Work fluids with 100% FWS, 200% FWS, 400% FWS, and 800% FWS were used in SS simulations. Other procedures of SS were similar to those of WS as introduced above. Permeability can be calculated according to Darcy’s lawwhere, k is the fluid permeability, mD; Q is the flow rate, cm3/s; μ is the fluid viscosity, mPa·s; L is the core length, cm; A is the cross-sectional area of the core, cm2; and ΔP is the differential pressure on both ends of the coal sample, MPa.

Permeability damage ratios of WS and SS were calculated by eq where, k is the fluid permeability for each kind of work fluid and mD; kf is the fluid permeability under 100% FWS conditions, mD.

Pore-Size Distribution Characterization

MIP tests were conducted by using an AutoPore IV 9500 mercury injection apparatus. The maximum mercury injection pressure, mercury surface tension, and contact angle were 100 MPa, 485 dyne/cm, and 130°, respectively. A significant compression in the coal matrix can be observed as the mercury intrusion pressure increases during MIP measurements,[36] leading to the generation of an inaccurate pore-size distribution result. An effective MIP data correction method proposed and applied by Han et al.,[37] Shao et al.,[36] and Zhang et al.[38] was applied in this study, whose details can be referred to in Appendix A. An automatic specific surface and pore analyzer, Tristar II 3020, was applied for LTNA tests. After high-temperature drying and degassing processes, samples were applied for LTNA tests at 77.3 K with relative pressures ranging from 0.001 to 0.995. The pore-size distribution of the coal sample in this research was obtained by combing data results from the MIP test and LTNA test, whose details can be referred to in Appendix A.

Determination of the Relationship between the T2 Value and Pore Diameter

There is a consistent one-to-one match between the relaxation time (T2) value and pore diameter, namely the greater the T2, the greater the corresponding pore diameter.[39] In this study, T2 spectra were detected using an LFNMR device named RecCore-2500. Additionally, the relationship between the T2 value and pore diameter was quantitatively determined by comparing the accumulated pore volume of pore-size distribution with an accumulated signal value of the T2 spectrum. The determination process of the relationship between T2 and the pore diameter can be referred to in Appendix B, and then the T2 spectrum can be regarded as a direct pore-size distribution after transferring the T2 value to the pore diameter.

CMP

CMP can distinguish pores and throats and calculate the pore-throat quantity ratio with extremely low-speed mercury intrusion.[40,41] A CMP device named APSE730 was used for the measurements. Samples were first dried for 24 h at the temperature of 120 °C, then mercury was injected into the samples with an extremely low speed of 0.00005 mL/min. The maximum injection pressure was 6.2057 MPa. The contact angle and surface tension between coal and mercury were 130° and 485 dyne/cm, respectively. The uniformity coefficient of the throat (a), throat sorting coefficient (RCth), and averaged pore-throat ratio (Rav-pt), which are parameters reflecting pore-throat structural complexities, can be calculated according to the following equations, namelywhere a is the distribution frequency of the normalized radius of a throat; r is the diameter of the throat, nm; rmax is the maximum throat diameter, nm; Rav-th is the average throat diameter value, nm; R is the pore-throat ratio; and b is the distribution frequency of the normalized pore-throat ratio.

Results

Mineral Compositions

Table and Figure demonstrate each mineral’s content percentage occupying the total coal and total mineral, respectively, detected by TIMA technology. Mineral contents of the two HVBC samples, namely HVBC1 and HVBC2, were merely 0.48 and 2.09% occupying the total coal, respectively, significantly lower than those of the MVBC1 sample (12.88%, Table ). Kaolinite, pyrite, and smectite take up 56.3, 16.7, and 10.4% of the total mineral content of the HVBC1 sample, respectively (Figure a). Calcite, ankerite, gorceixite, and albite could also be detected in the HVBC1 sample, although their contents were very low. The mineral content of the HVBC2 sample was mainly dominated by apatite and kaolinite, with a percentage of 45.4 and 43.1% occupying the total mineral content, respectively (Figure b). Contents of smectite, gibbsite, gorceixite, and siderite were relatively lower, compared with those of apatite and kaolinite. As for the MVBC1 sample, content percentages of kaolinite and calcite were relatively greater, which were 66.7 and 13.9%, respectively, compared with those of smectite, ankerite, pyrite, albite, and quartz (Figure c).

Table 3

Mineral Compositions of HVBC1 and HVBC2 as well as MVBC1 Samples

	content (%)
phase	HVBC1	HVBC2	MVBC1
organic component content	99.52	97.91	87.12
mineral content	0.48	2.09	12.88
kaolinite	0.27	0.90	8.60
calcite	0.01	n	1.79
smectite	0.08	0.01	0.91
apatite	n	0.95	0.01
ankerite	0.03	n	0.79
pyrite	0.05	n	0.46
gibbsite	0.01	0.08	0.01
albite	0.02	n	0.08
gorceixite	n	0.11	n
Mg_Ca_sulphate	n	n	0.10
quartz	n	n	0.07
iron oxides	n	0.03	0.01
siderite	n	0.01	n
other minerals	0.01	n	0.05

Figure 3

Each mineral’s content percentage occupying the total mineral: (a) HVBC1 sample; (b) HVBC2 sample; and (c) MVBC1 sample.

Minerals’ Occurrence State

For HVBC1 and HVBC2 samples, minerals are mainly demonstrated as a punctated state (Figure a,b) and fractures are not filled by minerals, showing that these minerals are distributed sparsely. As for MVBC1, which is a MVBC sample, minerals are distributed both in fractures and pores, and in addition, the fracture filling degree of MVBC1 is significantly severer than that of HVBC1 or HVBC2 (Figure c). It can be found that the kaolinite filling in the pores of MVBC1 is distributed as banded zonation, and almost all fractures of MVBC1 are filled by calcite (Figure c).

Figure 4

Minerals composition and distribution: (a) HVBC1 sample; (b) HVBC2 sample; and (c) MVBC1 sample.

A kaolinite particle with an irregular hexagon shape (4 μm in side length, approximately), located on the bedding surface rather than pores, is found in spot 1 of Figure a. Spot 2 and spot 3 in Figure b depict pores filled by apatite and kaolinite, respectively, and both the former and latter minerals are irregular. Figure c,d demonstrates pores filled by albite and quartz, showing that the contact between albite (or quartz) and the pore-wall surface is tighter than the contact between kaolinite and the bedding surface in Figure a.

Figure 5

Minerals’ occurrence states in the HVBC1 sample: (a) kaolinite particle with an irregular hexagon shape located on the bedding surface, noting that the observed layer is the bedding surface; (b) apatite and kaolinite filling in pores; (c) albite filling in pores; and (d) quartz filling in pores.

Kaolinites with irregular shapes displayed in Figure a are very fragile, and their aggregates are cluster-like and stacked on the bedding surface. Spot 2 and spot 3 in Figure b depict pores filled by irregular apatite, and it can be seen that the contact between apatite and the pore-wall surface is tight. Figure c shows pores filled by dimple-shaped irregular kaolinite. Spot 7 and spot 8 of Figure d display irregular quartz and kaolinite particles located in the margins of pore surfaces.

Figure 6

Minerals’ occurrence states in the HVBC2 sample: (a) kaolinite with an irregular shape is very fragile; (b) irregular apatite filling in pores; (c) dimple-shaped irregular kaolinite filling in pores; and (d) irregular quartz and kaolinite particles.

A direction-aligned kaolinite with a strip structure is found in Figure a. Figure b displays coal pores filled by kaolinite particles. Spot 3 and spot 4 as well as spot 5 in Figure c show quartz, kaolinite, and pyrite distributed on the bedding surface. Particles of kaolinite and quartz are found to be attached on the throat wall surface, namely spot 6 and spot 7 in Figure d, respectively.

Figure 7

Minerals’ occurrence states in the MVBC1 sample: (a) direction-aligned kaolinite with a strip structure; (b) kaolinite particles filling in coal pores; (c) quartz, kaolinite, and pyrite distributed on the bedding surface, noting that the observed layer is the bedding surface; and (d) kaolinite and quartz attached on the throat wall surface.

WS and SS Results

For the HVBC1 sample, the permeability tested under work fluid with 75% FWS was obviously lower than that tested under 100% FWS conditions (Figure a). As salinity decreased from 75% FWS to 25% FWS, permeability slightly fluctuated. As salinity decreased from 25% FWS to deionized water, permeability decreased sharply again. Correspondingly, permeability damage ratio values were stable as salinity decreased from 100% FWS to 25% FWS, and they doubled as salinity decreased from 25% FWS to deionized water. The final WS permeability damage ratio of the HVBC1 sample was 35.5%, falling into the category of medium WS based on the China oil & gas industry standard (SYT5358-2010) (Table ). The permeability of the HVBC2 sample decreased continually, and the permeability damage ratio of HVBC2 increased continually, as salinity decreased from 100% FWS to deionized water (Figure b). The final WS permeability damage ratio of the HVBC2 sample was 23.4%, falling into the category of weak WS. The permeability of the MVBC1 sample first decreased sharply and then steadily, and the permeability damage ratio of MVBC1 increased steadily as salinity decreased from 100% FWS to deionized water (Figure c). The final WS permeability damage ratio of the MVBC1 sample was 28.2%, falling into the category of weak WS.

Figure 8

Permeability and permeability damage ratio (abbreviated as PDR in this figure) variation as salinity of the experimental fluid decreased [(a) HVBC1 sample; (b) HVBC2 sample; and (c) MVBC1 sample] and increased [(d) HVBC1 sample; (e) HVBC2 sample; and (f) MVBC1 sample].

Table 4

WS and SS Damage Degree Evaluation Indexes, According to China Oil & Gas Industry Standard (SYT5358-2010)

WS or SS permeability damage ratio (%)	damage degree
permeability damage ratio ≤ 5	none
5< permeability damage ratio ≤ 30	weak
30< permeability damage ratio ≤ 50	weak to moderate
50< permeability damage ratio ≤ 70	moderate to strong
70< permeability damage ratio ≤ 90	strong
permeability damage ratio > 90	extremely strong

Permeability of the HVBC1 sample decreased continually, but the decreasing degree was slight, and the permeability damage ratio of the HVBC1 sample increased slightly as well, as work fluid salinity increased from 100% FWS to 800% FWS (Figure d). The final SS permeability damage ratio of the HVBC1 sample was 10.8%, falling into the category of weak SS. As salinity increased from 100% FWS to 800% FWS, permeability of the HVBC2 sample slightly fluctuated and remained almost unchanged. Correspondingly, the permeability damage ratio values were very small. The final SS permeability damage ratio of the HVBC2 sample was merely 1.7%, falling into the category of no SS (Figure e). The permeability of the MVBC1 sample decreased slightly as well, and the permeability damage ratio of the MVBC1 sample increased slightly, as salinity increased from 100% FWS to 800% FWS (Figure f). The final SS permeability damage ratio of the MVBC1 sample was 9.8%, falling into the category of weak SS as well. It could be observed from Figure that WS of HVBC and MVBC samples is more obvious than SS of those samples, and samples with a relatively higher WS permeability damage degree also have a relatively higher SS permeability damage degree.

Discussion

Pore-Size Distribution Results and the Relationship between T2 and the Pore Diameter

The incremental transition pore volume derived from corrected MIP data was close to that derived from LTNA data because the uncorrected MIP data exaggerated the transition pore volume, as displayed in Figure a–c demonstrating the pore-size distribution of the three samples. Compared with using uncorrected MIP data singly, the pore-size distribution combined with corrected MIP data and LTNA data together could be regarded as a more correct characterization method. The pore-size distribution of the HVBC1 sample was close to that of the HVBC2 sample, manifesting as meso-pore volume or transition pore volume, which was the greatest, followed by macro-pore and super macro-pore volume, and micro-pore volume was the lowest (Figure a,b). The pore-size distribution of the MVBC1 sample, which was MVBC, was different from that of the two HVBC samples, demonstrating that macro-pore and super macro-pore volumes were the greatest, followed by meso-pore volume and transition pore volume, and micro-pore volume was the lowest (Figure c). The total pore volume of the MVBC1 sample (0.010 cm3/g) was significantly lower than that of HVBC1 and HVBC2 samples, which were 0.040 and 0.042 cm3/g, respectively. Volumes of macro-pores, super macro-pores, meso-pores, and transition pores of the MVBC1 sample were significantly lower than those of the two HVBC samples. The micro-pore volume of MVBC1 was greater than that of the HVBC2 sample but lower than that of the HVBC1 sample, as marked by the red dashed line in Figure .

Figure 9

Pore-size distributions (abbreviated as PSD in this figure) derived from the combination of corrected MIP data and LTNA data: (a) HVBC1 sample; (b) HVBC2 sample; and (c) MVBC1 sample; and the cumulative pore-volume frequency curve derived from pore-size distribution data and LFNMR data: (d) HVBC1 sample; (e) HVBC2 sample; and (f) MVBC1 sample. (Abbreviations: MIP = mercury intrusion porosimetry; LTNA = low-temperature nitrogen adsorption; and LFNMR = low-field nuclear magnetic resonance).

Predecessors did a lot of work on the relationship between the T2 value and pore diameter, using uncorrected MIP data,[42−44] LTNA data,[45] or the T2 cutoff method.[46] LFNMR is an efficient tool for quantifying full-scale pore-size distribution.[47] The determination of the relationship between T2 and the pore diameter using pore-size distribution of this study can be more accurate, compared with the traditional methods using solo uncorrected MIP data or LTNA data. The quantitative relationships between the T2 value and pore diameter of the three samples, which are revealed according to the calculation method introduced in Appendix B, are demonstrated by each equation in Figure d–f, respectively. The cumulative pore volume frequency curves derived from pore-size distributions were close to those derived from LFNMR on the whole, for both of the HVBC and MVBC samples (Figure d–f). Key parameters of B-4 were determined by applying the least-square method. Values of n for the three samples were very close, and the value of C for the HVBC2 sample was almost two times and ten times that of the HVBC1 sample and MVBC1 sample, respectively (Figure d–f), showing that the relationship between T2 and the pore diameter varies with different samples, even for the two HVBC samples with close maturity (Table ).

Pore Space Variation Before and after WS and SS

After determining the relationship between T2 and the pore diameter, volume variations of pores with different diameters can be determined by comparing T2 spectra obtained before and after WS or SS. Figure depicts the diameter-transferred T2 spectra before and after WS or SS. It can be found that, for the HVBC1 sample after WS, volumes of macro-pores (marked by R3) and meso-pores (marked by R2) decreased slightly while those of transition pores (marked by R1) increased slightly (Figure a). As for the HVBC2 sample, volumes of macro-pores and super macro-pores (marked by R5) decreased overall, while those of meso-pores (marked by R4) increased slightly (Figure b). Moreover, for the MVBC1 sample, volumes of super macro-pores (marked by R9) and meso-pores as well as transition pores (marked by R7) decreased in general, while those of macro-pores (marked by dashed R8) increased dramatically, and those of micro-pores (marked by R6) increased slightly (Figure c). For the HVBC1 sample after SS, volumes of macro-pores and super macro-pores (marked by S2 in Figure d) decreased slightly, while those of meso-pores (marked by S1) increased slightly (Figure d). As for the HVBC2 sample, volumes of macro-pores and super macro-pores (marked by S4) decreased overall, while those of meso-pores and transition pores (marked by S3) increased generally after SS (Figure e). As for the MVBC1 sample, volumes of super macro-pores (marked by S9) and meso-pores (marked by S7) as well as micro-pores (marked by S5) decreased, while those of transition pores (marked by S6) and macro-pores (marked by S8) increased (Figure f).

Figure 10

Pore-volume variation before and after WS simulation: (a) HVBC1 sample; (b) HVBC2 sample; and (c) MVBC1 sample; and pore-volume variation before and after SS simulation: (d) HVBC1 sample; (e) HVBC2 sample; and (f) MVBC1 sample.

Migration Potential and Migration Pathway of Minerals During WS and SS Processes

For the two HVBC samples, non-clay minerals such as apatite, albite, and quartz are filled in pores, and the contact between those minerals and pore-wall surfaces is tighter than that between clay minerals and bedding surfaces (Figures a–d and 6a,b). Additionally, kaolinite particles have poor adhesion to the coal skeleton and weak adhesion between wafers. Therefore, under the action of shear, the kaolinite aggregate falls off from the base of skeleton particles and may also be easily broken into fragments, which can promote its migration.[48] Hence, compared with non-clay minerals, kaolinites in this study are easy to migrate during WS and SS simulation processes. Some quartz and pyrite are found to be distributed on the bedding surface or attached to the throat wall surface in the MVBC1 sample (Figure b,d). These quartz and pyrite may migrate during WS and SS simulation processes.

As for clay minerals, only kaolinite and smectite are found in the three samples (Table ). Kaolinite in the three samples is mainly displayed as particle, attaching on the bedding surface (Figures a, 6a, and 7a) or partially filling in meso-pores or transition pores (Figures d and 7b,d). Although kaolinite has no swelling ability, it is characterized by easily breaking apart and migrating, then concentrating at the pore-throat, leading to severe plugging and loss of permeability.[10,49−51] Hence, those kaolinite particles may result in a permeability decrease during WS and SS simulations. Smectite is 100% expandable, causing loss of porosity and a decrease in permeability,[19] hence a permeability decrease may also be attributed to smectite expanding during WS simulation, particularly for the MVBC1 sample with the smectite content greater than that of the two HVBC samples.

As introduced in Section , for HVBC samples, transition pore and meso-pore volumes increased overall after WS or SS, while macro-pore or super macro-pore volumes decreased after WS or SS. Combining with that, kaolinite of the two HVBC samples mainly displayed as particles, attaching on the bedding surface or partially filling in meso-pores or transition pores with high potential for detaching and migrating, as introduced in Section , it can be concluded that a part of kaolinite distributed in meso-pores or transition pores first detached and then migrated to other types of pores during the WS and SS processes, leading to an overall volume increase in transition pores and meso-pores. The migrated kaolinite agglomerated continually during migration and finally deposited in the narrow throat of macro-pores or super macro-pores, leading to a decrease in the macro-pore or super macro-pore volumes and loss of permeability. Figure demonstrates the sketch map of pore-volume variations and kaolinite migration pathways of the two HVBC samples during WS or SS.

Figure 11

Sketch map of pore-volume variations and kaolinite migration pathways of the two HVBC samples during WS or SS (transition pores or meso-pores in the magnified figure was enlarged for seeing clearly) (abbreviations: WS = water sensitivity; SS = salinity sensitivity).

As for the MVBC1 sample with high contents of kaolinite and smectite, pore-volume variations after WS or SS were quite complicated, as displayed in Figure c,f. Overall, transition pore and meso-pore volumes of the MVBC1 sample decreased after WS, which were different from those variations of the two HVBC samples. It could also be found that the macro-pore volume increased, while the super macro-pore volume decreased for the MVBC1 sample after WS. This phenomenon could be explained as follows: (1) kaolinite was severely filled in macro-pores of the MVBC1 sample, as displayed in Figure c, and those kaolinites were detached, migrated, and finally deposited in super macro-pores, leading to a macro-pore volume increase and a super macro-pore volume decrease; and (2) smectite, which was detected by TIMA but not observed by SEM, may have filled in transition pores and meso-pores. The smectites swelled after WS, leading to decreased volumes of transition pores and meso-pores.

The macro-pore volume of the MVBC1 sample increased while the super macro-pore volume decreased as well after SS. This phenomenon could also be explained by the kaolinite filled in macro-pores was detached, migrated, and finally deposited in super macro-pores. Some small coal fines and kaolinite in transition pores may block micro-pores and meso-pores after migration in the SS process of the MVBC1 sample, leading to a volume increase in transition pores and a volume decrease in MP and meso-pores, as displayed in Figure f.

Gong reported that SS damage degrees of anthracite samples are weak,[52] which are the same as those of the bituminous coals in this study (Table ). Because research on the SS of coal is scarce, further discussions are limited. Zhao,[53] Hu et al.,[54] Tao et al.,[10] Gong,[52] Zuo et al.,[55] Tian and Wu[56] applied anthracite samples to study WS; Wang et al.[21] applied lignite samples to study WS; and Gao et al.,[57] Geng et al.[58] and we applied bituminous coal samples to study WS. Table displays the results of these studies. It can be seen that WS permeability damage ratios vary tremendously even for coals with the same ranks. For instance, Tao et al.[10] reported WS permeability damage ratios of anthracite, which vary from 5.21 to 52.08%, and all data of anthracite from Zhao,[53] Hu et al.,[54] Tao et al.,[10] Gong,[52] Zuo et al.,[55] and Tian and Wu[56] show that WS permeability damage ratios of anthracite vary from 5.21 to 83.06%, indicating that there is no close relationship between the WS damage degree and coal rank.

Table 5

Results on WS and SS Permeability Damage Ratios from Predecessors’ Published Work

predecessors’ publication	coal type	permeability damage ratios of WS or SS	total mineral content (%)
Zhao[53]	anthracite	two samples with WS permeability damage ratios of 80.81 and 83.06%, respectively, with an average of 81.93%	7.50% (clay mineral content)
Hu et al.[54]	anthracite	one sample with the WS permeability damage ratio of 54.74%	none correlated data published
Tao et al.[10]	anthracite	Six samples with WS permeability damage ratios of 52.08, 5.21, 6.94, 32.47, 10.59, and 14.14%, respectively, with an average of 20.23%	nine samples with the mineral content of 9.86, 3.50, 11.96, 7.41, 11.63, 2.25, 7.91, 14.89, and 5.70%, respectively, with an average of 8.35%
Wang et al.[21]	lignite	one sample with the WS permeability damage ratio of 54.18%	3.82%
Gong[52]	anthracite	four samples with WS permeability damage ratios of 28.71, 43.31, 26.21, and 22.48%, respectively, with an average of 30.18%	none correlated data published
Gao et al.[57]	bituminous coal	two samples with WS permeability damage ratios of 90.66 and 95.07%, respectively, with an average of 92.87%	none correlated data published
Geng et al.[58]	bituminous coal	five samples with WS permeability damage ratios of 54.03, 40.26, 47.57, 45.84, and 29.92%, respectively, with an average of 43.53%	five samples with the mineral content of 11.10, 5.70, 8.30,5.00, and 3.60%, respectively, with an average of 6.74%
Zuo et al.[55]	anthracite	nine samples with WS permeability damage ratios of 63.63, 52.70, 32.61, 14.23, 14.30, 12.02, 11.13, 7.10, and 5.60%, respectively, with an average of 23.72%	none correlated data published
Tian and Wu[56]	anthracite	five samples with WS permeability damage ratios of 36.93, 9.91, 32.61, 15.17, 21.79%, respectively, with an average of 21.05%	none correlated data published
Gong[52]	anthracite	four samples with SS permeability damage ratios of 19.39, 18.64, 19.33, 18.64%, respectively, with an average of 18.99%	none correlated data published

Influence of the Pore Structure on WS and SS

Parameters of a, RCth, and Rav-pt were derived by CMP using eqs –5. The greater the a, the diameters of throats of a sample are closer to the largest throat of the same sample, and the throat diameter is more evenly distributed. The smaller the RCth, the diameters of throats of a sample are going to be closer to the mean diameter of the same sample and the better the sorting property of the throats of that sample. The greater the Rav-pt, the more complicated the pore-throat structure of the sample. In summary, the greater the a or the smaller the RCth or the smaller the Rav-pt, the more homogeneous the structures of macro-pores and super macro-pores of a sample. It could be seen in Table that value a of the HVBC2 sample was the greatest, followed by that of the MVBC1 sample, while that of the HVBC1 sample was the lowest. Values of RCth and Rav-pt of the HVBC1 sample were the greatest, followed by those of the MVBC1 sample, and those of the HVBC2 sample were the lowest. These results indicated that macro-pore and super macro-pore structural complexities of the HVBC1 sample were the greatest, followed by those of the MVBC1 sample, and those of HVBC2 samples were the lowest.

Table 6

Values of Macro-Pore Parameters and Permeability Damage Ratiosa

sample	WS permeability damage ratio (%)	SS permeability damage ratio (%)	a	RCth	Rav-pt	sum volume of macro-pores and super macro-pores (cm3/g)	kaolinite content (%)	smectite content (%)
HVBC1	35.5	10.8	0.16	0.84	201.04	0.0067	0.27	0.08
HVBC2	23.4	1.74	0.51	0.45	144.08	0.0103	0.90	0.01
MVBC1	28.2	9.8	0.19	0.71	214.42	0.0038	8.60	0.91

(Abbreviations: WS = water sensitivity; SS = salinity sensitivity; a = uniformity coefficient of throat; RCth = throat sorting coefficient; and Rav-pt = average pore-throat ratio).

As discussed in Section , the migrated kaolinite was deposited in the narrow throats of macro-pores or super macro-pores, leading to a permeability decrease; thus, structural complexities of super macro-pores and macro-pores may play a key role in determining the WS or SS damage degree. Table shows that the WS and SS permeability damage ratios of the HVBC1 sample were the greatest, followed by those of the MVBC1 sample and those of the HVBC2 sample were the lowest. The WS and SS damaged degree orders of the three samples were the same as the macro-pore and super macro-pore structural complexity orders of those three, indicating that WS and SS permeability damage ratios were closely related to structural complexities of macro-pores and super macro-pores. Meanwhile, samples with high clay mineral contents and low values of sum volumes of macro-pores and super macro-pores (taking MVBC1, for instance) were not the severest in WS and SS damage. Therefore, it could be summarized that damage degrees of WS or SS of the three samples were mainly dominated by macro-pore and super macro-pore structural complexities rather than the clay mineral content or pore volume. Hence, the main WS and SS mechanisms of MVBC and HVBC in this study can be summarized as the detachment, migration, and deposition of kaolinite contributing to a macro-pore or super macro-pore volume decrease, and a more complex pore structure with a greater pore-throat ratio and stronger heterogeneity leading to severer plugging of the throat induced by those deposited kaolinites, leading to more permeability loss.

Influence of Minerals on WS

Tao et al.[10] and Geng et al.[58] reported the relationship between WS permeability damage ratios and the mineral content of anthracite and bituminous coal, respectively, and they found that WS permeability damage ratios increase as the mineral content increases, with only two exceptions marked by the blue dashed line in Figure a. In particular, an obvious increasing trend of the WS permeability damage ratio can be found as the clay mineral content increases, as shown in Figure b, which can be evidence that the clay mineral induces the generation of WS. However, as the carbonate mineral content increases, the WS permeability damage ratio generally decreases (Figure c). This phenomenon can be attributed to the fact that the contact between the coal matrix and carbonate mineral is so tight that the carbonate mineral is not easily detached from the coal surface. The two exceptions in Figure c, marked by the red dashed line, are both high in the carbonate mineral content and clay mineral content, and the latter leads to their high WS permeability damage ratios. The relationship between the WS permeability damage ratio and mineral content of results studied by this paper is not obvious. This is because the WS damage degree of bituminous coal samples of this study is closely related to the pore structure, as discussed in Section . Figure d manifests a negative relationship between the WS permeability damage ratio and 100% FWS permeability; this phenomenon can be attributed to the fact that coal samples with high permeability are usually low in clay mineral content, and these samples are wide in the aperture of the coal fracture. Hence, the detachment, swelling, migration, and deposition of clay minerals have little influence on permeability, and correspondingly, the WS permeability damage ratios of these coals are low. Overall, the clay mineral content, permeability, and pore structure complexity are key factors influencing the WS damage degree.

Figure 12

Relationships between WS PRD and (a) mineral content, (b) clay mineral content, (c) carbonate mineral content, and (d) 100FWS permeability, respectively.

In the future, the following aspects can be given attention to enhance the scientific understanding of WS and SS, with the purpose of better instructing the actual CBM production: (1) studies on WS and SS must be related to the actual work fluid composition, particularly the conditions of drilling and fracturing as well as workover fluids; (2) comparative studies between coal and different ranks, mineral contents, mineral types, and mineral occurrences using large amounts of samples need to be conducted; and (3) advanced transparent geological analysis technology such as using micro-CT during WS and SS simulations can be conducted to observe the dynamics of reservoirs and minerals during the experimental process.

Conclusions

This work studied and discussed the control of mineral occurrence and pore structural complexity on the WS and SS of bituminous coal deposited in northwestern China; the main conclusions are summarized as follows:

HVBC samples of this study were characterized as medium to weak WS (with the highest permeability damage ratio of 35.5%) and weak SS (with the highest permeability damage ratio of 10.8%) and MVBC samples of this study were characterized as weak WS (with the highest permeability damage ratio of 28.2%) and weak SS (with the highest permeability damage ratio of 9.8%).

Kaolinite, distributed as irregular particles or fragile aggregates attached on the bedding surface or filled in pores, is the dominant clay mineral in HVBC and MVBC of this study. For HVBC samples of this study, the detachment, migration, and deposition of kaolinite contributed to the macro-pore or super macro-pore volume decrease after WS or SS. Kaolinites filled in the macro-pores of MVBC of this study detached, migrated, and finally deposited in super macro-pores after WS and SS, leading to losses of super macro-pore volume and permeability;

More complex pore structure with a greater pore-throat ratio and stronger heterogeneity can lead to severer plugging of the throat induced by deposited kaolinite after WS or SS, leading to more permeability loss.