The variation of stable carbon isotope (δ¹³C_DIC) in dissolved inorganic carbon (DIC) can reflect the geochemical behaviour and the biogeochemical characteristics of carbon. At present, it is commonly used to study the source of carbon in surface water systems and reservoir water bodies, and spatial and temporal evolution process of carbon, and to reveal quality characteristics and geological origin of water^[1,2]. There are four forms of DIC in water: dissolved CO₂, carbonate acid (H₂CO₃), bicarbonate (HCO₃^-), and carbonate ion (CO₃^2-). It is found that both surface water systems and reservoirs are open water bodies, which are mainly affected by photosynthesis, respiration, carbonate weathering and soil CO₂ export, and where the δ¹³C_DIC values are generally negative.

The DIC in formation water of deep coal measures mainly comes from the dissolution of carbonate minerals and CO₂ in coalbed methane (CBM). However, the δ¹³C_DIC in water samples from CBM wells or deep shale gas wells generally shows positive anomalies, and some water samples have δ¹³C_DIC of more than 10‰. This phenomenon is very common in China and abroad. In the United States, the produced water samples from Pennsylvanian bituminous coal in the Black Warrior Basin have δ¹³C_DIC values of 2.8-13.1‰^[3], the water samples from Tertiary semi-bituminous coal in the Powder River Basin have δ¹³C_DIC values between 12.0‰ and 22.0‰^[4], and the produced water samples from Carboniferous CBM wells in the Atlantic Rim have δ¹³C_DIC values between -3.6‰ and 22.8‰^[5]. Produced water samples from Devonian shale gas wells of the Monongahela River Basin have δ¹³C_DIC of more than 8.5‰ in general, and 21.0‰ at maximum^[6]. In Shizhuang block, Shanxi province, the produced water samples from CBM wells have the highest δ¹³C_DIC value of more than 20.0%^[7]. It is generally believed that the geological cause of this phenomenon is the reduction of methanogens^[5,6,7,8,9], but there is little direct evidence. Some domestic and foreign researchers have examined the internal relationship between δ¹³C_DIC in produced water of CBM well and CBM productivity preliminarily, and concluded that the positive anomalies of δ¹³C_DIC are beneficial to high productivity of CBM^{[4-5, 10]}.

From January 2017, we tested and tracked the δ¹³C_DIC values of produced water samples from some CBM wells in western Guizhou for a long term. The test results show that the δ¹³C_DIC values of some produced water samples from the CBM wells are much higher than 10‰. In this study, with 20 CBM wells in western Guizhou as examples, the δ¹³C_DIC characteristics of produced waters from the CBM wells are analyzed to find out the geological reasons behind the spatial and temporal differences in δ¹³C_DIC of water samples from commingling production wells in the Songhe GP well group. Combined with 16S rDNA analysis method, archaea in typical wells are tested to reveal the direct geological causes of δ¹³C_DIC positive anomalies. Finally, the relationship between δ¹³C_DIC of produced water from CBM well and CBM productivity is discussed, and the geological response model of δ¹³C_DIC is put forward for produced water from commingling production CBM wells, which provides a geochemical and micro-organism method for identifying stacking fluid systems and analyzing gas and water contributions of layers in CBM wells in commingled production.

The CBM resources in the Upper Permian in the western Guizhou have geological characteristics of multiple, thin coal seams, a wide range of coal ranks, high stress, low water cut, and complex coal structure^[11,12,13]. CBM development test wells in western Guizhou are mainly distributed in a number of synclines in western Guizhou and northern Guizhou, such as Tucheng, Zhuzang, Dahebian synclines, etc. (Table 1).

The coal-bearing strata in the Songhe block of Tucheng synclines are Upper Permian Longtan Formation. The middle member of this formation is delta front facies, while the upper and lower members are lagoonal tidal-flat facies. The coal-bearing strata are 341.00 m thick on average. There are 18 mineable coal seams, mainly including No. 1+3, 4, 9, 12, 15, 16, and 17, with a total thickness of 11.68 m, followed by No. 5, 6, 10, 11, 18, 27, 29, with a total thickness of 13.34 m. Among them, the No. (1+3)-10 coal seams belong to the upper member of the Longtan Formation, the No. 12-18 coal seams belong to the middle member of the Longtan Formation, and the No. 24-29 coal seams belong to the lower member of the Longtan Formation. Made up of mainly coking coal, the coal seams have a high CBM content of 6.46-20.99 m³/t, gas saturation greater than 70%, and pressure coefficient of 1.08-1.40, indicating abnormal high-pressure^[12]. The Songhe block currently has 8 CBM development test wells, which form a well cluster. All the wells are developed by multistage fracturing and commingling production with production layers of more than 6 coal seams in general (Table 1). By the end of August 2018, the wells had a maximum daily CBM production of approximately 3000 m³/d, and a stable production of about 500 m³/d later. The wells had a cumulative water production of approximately 1400-3300 m³ and an average cumulative water production per well of over 2000 m³.

The Zhuzang syncline of Zhijin block has dozens of CBM wells, of which Well ZP-1 in Zhi2 well group is a horizontal well developing the No. 23 coal seam; Wells Z2, Z3, Z4, Z5 and Z6 in Zhi2 well group are vertical wells, developing the Ⅲ coal seam group (the main coal seams of No. 20, 23, 27 and 30), after staged fracturing and commingled production manner. By the end of August 2018, the wells had a highest daily gas production of around 3000 m³, and most wells had a stable CBM production of over 1000 m³/d. The wells had cumulative water production of 500-3000 m³, and an average cumulative water production of over 1000 m³.

Wells B1, N1, and N3, located in the Bide syncline, are developed by staged fracturing and commingled production of 2-3 layers. Well ZH1, located at the Dahebian syncline, is produced after staged fracturing manner of 2 layers. Well D2 located in the Huangnitang syncline is produced after staged fracturing manner from 3 layers. Well F2 located in the Changgang syncline in northern Guizhou is produced from 3 layers after staged fracturing. By the end of August 2018, these wells had a highest daily gas production of 1500 m³ and stable gas production of around 1000 m³/d later, and a cumulative water production of 500-3000 m³ (Table 1).

For 20 CBM development wells in the study area, water samples were collected and tested once every 2-3 months, starting from January 2017. All water samples were collected directly from the outlet of CBM wells with 2.5 L pure water bottles. The plastic bottles were washed 3 times with the collected water, and then filled fully with water to expel all the air in the bottle and sealed with a bottle cap. Finally, the bottles were checked for leakage, marked with sampling time and place, and sent to the State Key Laboratory of Environmental Geochemistry, Guiyang Institute of Geochemistry, Chinese Academy of Sciences for testing within 72 h. The stable carbon isotopes testing equipment was a gas isotope ratio mass spectrometer (MAT253, USA). The test followed the Atekwana and Krishnamurthy’s method^[14]. Part of the test data is listed in Table 1.

Gas samples from some CBM wells were tested for δ¹³C of methane and carbon dioxide. The gas samples were collected by water drainage and gas gathering method and sent to the Key Laboratory of Petroleum Resource Research, Chinese Academy of Sciences within 72 h for testing. The testing equipment was gas chromatograph (HP6890) and isotope ratio mass spectrometer (Delta plus XP). The testing followed the national standard (GB/T 6041-2002) strictly^[15].

The composition of gas samples from different layers of Well GC-1 was tested by gas chromatograph (GC5890A) in Guizhou Research Center of Shale Gas and CBM Engineering Technology. The test followed the national standard (GB/T 13640-2014) strictly^[16].

16S rDNA gene sequence of water samples from 6 CBM wells in the GP well group was analyzed: (1) Water samples of 500 mL each were collected from the outlet of CBM wells and sent to the laboratory under anaerobic low temperature condition. (2) The methanogens were cultured in methanogenic culture medium at 35 °C. The methanogenic culture medium (1.0 L) including: 1.0 g of NH₄Cl, 0.1 g of MgCl₂·6H₂O, 0.4 g of K₂HPO₄·3H₂O, 0.2 g of KH₂PO₄, 0.1 g of tryptone, 1.0 mL of resazurin (0.1%), 1.0 g of yeast extract, 2.0 g of sodium acetate, 2.0 g of sodium formate, 0.5 g of L-cysteine hydrochloride, 0.2 g of Na₂S·9H₂O, 2.0 g of NaHCO₃, and 10 mL of trace element solution. The trace element solution was prepared by nitrilotriacetic acid, MnSO₄, MgSO₄·7H₂O, FeSO₄, NaCl, CoCl, CaCl₂, CuSO₄, ZnSO₄, H₃BO₃, Al(SO₄)₂, NiCl₂, and NaMoO₄. (3) After culturing for 4-5 d, 30 mL of the cultured sample was poured into a centrifuge tube to remove oxygen and then sealed and sent for test.

The sequencing was completed at Sangon Biotech (Shanghai) Co., Ltd. The DNA was extracted with the kit (E.Z.N.ATM Mag-Bind Soil DNA Kit). Archaea detection was conducted by PCR (Polymerase Chain Reaction) with three rounds of amplification. In the first round, the M-340F GU1ST-1000R primers were used for amplification, and in the second round, the first round PCR products were used for amplification. The primers used were fused with V3-V4 universal primers of the Miseq sequencing platform, including the primers 341F: CCCTACACGA CGCTCTTCCGATCTG (barcode) CCTACGGGNGGC WGCAG and primers 805R: GACTGG AGTTCCTTGGCACCCGAGAATTCCAGACTA CHVG GGTATCTAATCC. In the third round, Illumina bridge PCR-compatible primers were introduced. After PCR, the products were detected by agarose electrophoresis.

The DIC in produced water of CBM wells mainly comes from CO₂ dissolution, carbonate mineral dissolution and microbial action etc.^[17]. Compared with natural water (surface water and groundwater), produced water from CBM well is often richer in δ¹³C_DIC. It is generally believed that the positive anomaly δ¹³C_DIC value more than 10‰ of CBM well water sampleis related to microbial reduction. This is because microorganisms can produce methane by acetic acid fermentation (equation (1)) and carbon dioxide reduction (equation (2))^[18,19], during these processes, methanogenic bacteria preferentially absorb ¹²C, and ¹³C gradually becomes richer. Moderate to negative range of δ¹³C_DIC (generally -7‰-0) is mainly related to CO₂ dissolution in CBM and carbonate mineral dissolution. Very low value of δ¹³C_DIC (-14‰--7‰) is related to oxidation^[17] of surface water^[4]. Moderate positive δ¹³C_DIC (generally 0-10‰) is mainly the combined result of dissolution of carbonate minerals in coal measures and slight microbial reduction.

Fig. 1 shows the distribution of δ¹³C_DIC values of produced water samples from CBM wells in western Guizhou in March 2017. The δ¹³C_DIC values of Wells GP-2 and N1 exceed 10‰, and the δ¹³C_DIC of Well N1 is the largest, reaching 27.2‰. Therefore, it is inferred that microbial reduction occurred in the coal measures of these two wells and produced secondary biogas. The δ¹³C_DIC value of river water sample in Tucheng block is -13.1‰, a very low negative value, which is related to oxidation. δ¹³C_DIC values of produced water samples from other CBM wells are relatively low. Water samples from Wells GP-1, GP-5, Z3, Z4, Z5, N3, B1 and D2 have moderately negative δ¹³C_DIC values, which are related to CO₂ dissolution. In contrast, water samples from Wells GP-3, GP-4, GP-6, GP-8 and Z6 have larger positive values, so microbial reduction effect can’t be ruled out in these wells.

The nine batches of data from January 2017 to July 2018 were further analysed (Fig. 2). From the distribution of R_o,max and δ¹³C_DIC, it can be seen that the samples with positive anomaly δ¹³C_DIC value of more than 10‰ mainly come from the GP well group in Songhe block with medium coal rank except Well N1. Well N1 has a drainage period of up to 2 years and flowback rate of more than 150%. According to the conventional ions and hydrogen and oxygen isotope characteristics, its produced water may be external water^[20]. It is generally believed that the secondary biogas in coal seam is mainly generated in the stage with R_o,max of 0.3%-1.5%^[19]. In CBM wells with a R_o,max of greater than 1.5%, positive anomalies of δ¹³C_DIC are rarely greater than 10‰. The R_o,max of GP well group is approximately 1.5%, indicating there are material conditions for generation of secondary biogas.

The δ¹³C₁ of produced gas samples from some CBM wells in the GP well group was measured. The tested δ¹³C₁ values of gas samples collected in March 2017 are -43.6--41.9‰, with an average value of -42.8‰. δ¹³C₁ of -55‰ is generally considered as the boundary value between biogenic gas and thermogenic gas^[21]. According to this standard, the gas in the study area is mainly thermogenic gas. But according to the thermal simulation regression formula (equation (3)) of primary CBM coal and rock without obvious secondary reform effects proposed by Liu and Xu^[22], at the R_{o, max} of 1.3%-1.7%, the calculated δ¹³C₁ range is -32.25‰ to -29.63‰, which is much higher than the average value of -42.8‰, and thus indicates that there is slight secondary reform of CBM in this area, rather than pure thermogenic gas. The δ¹³C of CO₂ in produced gas from Well GP-3 was tested at 10.7‰ in the later stage, which indicates the CO₂ in this well is of biogenic origin^[8]. This result also proves that there is microbial reduction in this area.

Taking the GP well group as an example, the dynamic variation characteristics of δ¹³C_DIC data of 9 batches of water samples collected from January 2017 to July 2018 were analysed (Fig. 3a). On the whole, the δ¹³C_DIC values of most wells increase slowly with the drainage time, which is mainly due to the increasing proportion of produced water from the middle member of Longtan Formation. In January 2017, only the δ¹³C_DIC of Well GP-2 was more than 10‰, then in July 2018, that of Wells GP-2, GP-3, GP-4 and GP-5 were all more than 10‰. The Mann-Kendall method^[23] was used to test the variation trend of δ¹³C_DIC with time, and the results are shown in Table 2. A positive value of Z indicates increasing trend, while a negative value indicates decreasing trend. The absolute values of Z greater than or equal to 1.28, 1.64 and 2.32 means passing the tests of 90%, 95% and 99% significance, respectively. The trend test results show that the δ¹³C_DIC values of Wells GP-2, GP-3, GP-4, GP-5, GP-6 and GP-8 show an increasing trend, and the increasing trend of Wells GP-3, GP-5, GP-6 and GP-8 is very obvious while the GP-2 and GP-4 wells show an unobvious increasing trend. The Well GP-1 shows an obvious decreasing trend, while the Well GP-7 shows an unobvious descending trend.

In August 2017 and January 2018, some wells showed two obvious drops of δ¹³C_DIC. It is speculated that the drop in August 2017 is related to the rainy season precipitation in the study area, which occurs from May to September every year in Guizhou Province. The recharge of atmospheric precipitation will cause a decrease in the δ¹³C_DIC value in some wells^[5]. In January 2018, the decrease of δ¹³C_DIC value in the GP-1 and GP-2 wells was due to engineering reasons. The 1+3 coal seams in Well GP-1 were re-fractured and later the 1+3 coal seams were produced separately. Meanwhile, Well GP-2 was shut down to cooperate with the fracturing in Well GP-1. As a result, the gas production of the two wells was 0 in January and the δ¹³C_DIC value greatly reduced, before returning to a normal level in March 2018. Research shows that the surface water has a very low δ¹³C_DIC due to oxidation^[19]. As the fracturing fluid is often prepared with river water, and the δ¹³C_DIC value of the river water sample near the GP well group is -13.1‰, this further proves that the main reasons for the decrease of the δ¹³C_DIC value are the atmospheric precipitation and the fracturing fluid invasion at the early stage of drainage. The δ¹³C_DIC values of other wells vary stably with time (Fig. 3b). Except δ¹³C_DIC value of Well N1 of higher than 20‰ and that of Z4 well of less than -5‰, the values of the other wells are all in the range of -5‰-5‰. This may be related to the facts that the CBM wells have few development layers, generally 1-4 layers and stable gas source, and are in stable production period.

3.3.1. Identification of produced water source in multi-layer commingled production

A positive anomaly value of δ¹³C_DIC greater than 10‰ is considered an important indication of existence of microbial reduction and secondary biogas^{[3,4,5,6,7,8,9]}. Yang et al.^[24] examined the δ¹³C_DIC dynamic characteristics of the water samples produced from the CBM wells in the Songhe GP well group from January 2017 to November 2017 and tried to find out the geological reasons for the differences between the wells. They argued that the high δ¹³C_DIC value of the produced water from Well GP-2 is because its developed layer is in the middle member of the Longtan Formation. The middle member of Longtan Formation is delta front sedimentary facies with fine sandstone, while the upper and lower members of Longtan Formation are lagoon-tidal flat facies, with siltstone and argillaceous siltstone. The difference in the sedimentary facies and lithologic assemblage leads to higher permeability and water cut of the middle member of the Longtan Formation, meanwhile, the upper member of the Longtan Formation also has a relatively good permeability and higher water content because of shallow depth, so three sets of superimposed fluid systems exist matching sedimentary facies^[11]. The coal seams have a R_o,max range from 1.3% to 1.7%, over-pressure^[12], and temperatures of approximately 30-40 °C, suitable for generation and survival of microbes. Therefore, the middle member and the upper member of the Longtan Formation are suitable for survival and preservation of microbes. The vertical distributions of CO₂ and C₂H₆+C₃H₈ in Well GC-1 of GP well group have segmentation feature. The upper member (No. 1-10 coal seams) and middle member (No. 12-18 coal seams) of the Longtan Formation have higher CO₂ concentration and lower concentration of C₂H₆+C₃H₈, while the lower member (No. 24-29 coal seams) of the Longtan Formation has lower CO₂ concentration and higher C₂H₆+C₃H₈ concentration. It is speculated that the reason of the lower concentration of C₂H₆+C₃H₈ and higher concentration of CO₂ in the upper and middle members is that bacteria reduced organic matter and heavy hydrocarbon in the coal seams to produce CO₂ and CH₄^[25] (Fig. 4). At present, the liquid level of other wells that produce from all the three members of Longtan Formation is in the middle and lower members, and the upper member has been exposed. The contribution capacity of exposed reservoir is small. Therefore, the production layers of most wells are mainly the middle and lower members of Longtan Formation, and the lower member in some wells. For example, the liquid level of Well GP-7 is now at the top of the No. 27 coal seam, which means that the upper and middle members of Longtan Formation have been exposed, so its main production layers are the No. 27 and 29 coal seams, so the δ¹³C_DIC value of the produced water is relatively low.

The average values of 9 batches of data from January 2017 to July 2018 were plotted as a histogram (Fig. 5). Compared with the average values from January 2017 to November 2017: (1) The average value of δ¹³C_DIC of Well GP-2 is still the largest. (2) The average values of δ¹³C_DIC of Wells GP-1 and GP-5 have larger changes, the average value of Well GP-1 becomes smaller, while the average value of Well GP-5 becomes larger. The reason is that 1+3 coal seams in Well GP-1 were re-fractured in January 2018 and were changed into separate drainage, and the flowback duration of the fracturing fluid was short, part of the fracturing fluid remained in the formation, causing drop of the δ¹³C_DIC average value of the produced water from this well. Well GP-5 is adjacent to Well GP-1 and is in the updip direction, so it is prone to indirect enhanced stimulation. It can be seen from the production curves of Well GP-5 (Fig. 6) that: (1) In January 2018, the dynamic liquid level of Well GP-5 increased significantly from 110 m to more than 200 m. Before 2018, the dynamic liquid level was above the No. 16 coal seam, and affected by the re-fracturing of Well GP-1, the liquid level rose to the vicinity of the 1+3 coal seam. (2) The daily water production increased in 2018, while the daily gas production decreased from 351 m³/d in November 2017 to 96 m³/d in January 2018, and the δ¹³C_DIC increased significantly in January 2018. The reason is that the No. 29 coal seam stopped producing gas due to the increase of liquid level after January 2018, and the production layers were mainly in the middle and upper members of Longtan Formation, and consequently, the δ¹³C_DIC value increased considerably.

Burial depth also has a great influence on δ¹³C_DIC (Fig. 7). For Wells GP-3 and GP-4, and GP-6 and GP-8 which have similar developed layers, and similar production layers under dynamic fluid level constraints, it can be seen that the two groups of wells show similar regularity, which is the deeper the burial depth, the greater the value of δ¹³C_DIC is. The Zhi2 well group (Wells Z2, Z3, Z4, Z5 and Z6) shows similar regularity. The 5 wells are same in production layer, the deeper the burial depth, the larger the δ¹³C_DIC value is. This is because CBM wells in shallow layers are more likely to receive recharge from atmospheric precipitation, which would cause drop of the δ¹³C_DIC value. The produced water samples from the Carboniferous CBM wells in the Atlantic Rim, USA show similar regularity. The produced water samples from deeper CBM wells have larger δ¹³C_DIC values than those from shallower CBM wells^[5].

3.3.2. Common response of microorganisms

In recent years, 16S rDNA amplification and sequencing technology has been widely used in studying the microbial structure of coal mines or coal-bearing basins. The research of microbial sequencing has been carried out in many regions abroad^{[26,27,28,29,30,31]} (such as Yubai in Japan, the Surat basin in Australia, Illinois basin and Powder River basin in US) and in China^{[7, 32-33]} (such as Ordos Basin and Qianqiu mine in Henan).

To fully verify the microbial origin of the δ¹³C_DIC, the 16S rDNA amplification and sequencing test was done on produced water samples from Wells GP-1, GP-2, GP-3, GP-5, GP-7 and GP-8 for the first time in January 2019. The results show that there are a large number of methanogens in the produced water samples from the six wells. The classes include Methanobacteria, Methanomicrobia, Methanococci etc., and the former two are predominant, accounting for 60.58% and 37.29%, respectively. A total of more than 15 species of genus are included. Among them, Methanobacterium is dominant, followed by Methanothrix (Fig. 8). Methanobacterium is the main hydrogenotrophic methanogen, which can metabolize H₂ and CO₂ into CH₄ as reaction in equation (2). Methanothrix is a methanogen of the acetic acid type, which can produce CH₄ and CO₂ through anaerobic metabolism as in equation (1) without H₂ and CO₂.

In addition, hydrogenotrophic genus Methanocorpusculum, Methanoregula, Methanospirillum and Methanoculleus, methyl- trophic genus Methanomassiliicoccus and Methanolobus, mixed genus (hydrogenotrophic and acetic acid-trophic) Methanococcus and other genus were all detected in the produced water. Although these methanogens are small in proportion in the water samples, it can be determined that there are three types of methanogens in the well group of the study area: hydrogenotrophic, acetic acid-trophic and methyl-trophic. They produce methane in a variety of ways. Methanobacterium, a hydrogenotrophic bacterium, is the main genus of methanogens. This is consistent with the conclusion that most biogas is produced by hydrogenotrophic methanogens through CO₂ reduction^[34]. The correlation coefficient between the sequence number of genus Methanobacterium and δ¹³C_DIC is very high (Fig. 9) (coefficient of 0.885 4). This shows that during the reduction of methanogens, especially hydrogenotrophic methanogens in the reaction of equation (2), CO₂ metabolism produces CH₄ and preferentially absorbs ¹²C, which is the main reason of the ¹³C enrichment. Meanwhile, it also shows that the decomposition of heavy hydrocarbons into CH₄ may be mainly accomplished with the participation of hydrogenotrophic methanogens.

According to the abundance distribution of archaea at the genus level shown in Fig. 8, the six wells can be divided into two types apparently, of which GP-1 and GP-7 wells are one type, and the other wells are roughly the other type. This is consistent with the difference in the average values of δ¹³C_DIC in 9 batches of data previously mentioned. The feature further corroborates that the DIC is closely related to methanogens, and that different superimposed fluid systems formed by different lithological and physical properties of multi-coal seam gas-producing intervals have different genera of bacteria.

3.3.3. Indicative significance of CBM productivity

Fig. 10 shows the relationship of the δ¹³C_DIC values of produced water samples from wells in the study area and the corresponding daily production of CBM of the wells in stable production period (March 2017). The horizontal well ZP1 is not considered in the analysis because its high productivity is much more related to engineering factors. The relationship between daily production of CBM and δ¹³C_DIC value is not obvious, rather the daily production of CBM tends to decrease with the increase of δ¹³C_DIC value (Fig. 10a). At present, Well GP-7 mainly produces CBM from the No. 27 and No. 29 coal seams in the lower member of the Longtan Formation^[24] and shows similar regularity (Fig. 10b), that is with the increase of δ¹³C_DIC, its daily gas production decreases. In contrast, Well ZH1 shows an increase of daily production of CBM with the increase of δ¹³C_DIC value. Therefore, the relationship between δ¹³C_DIC value and CBM production is complex. The main reason is that the positive anomaly of the δ¹³C_DIC value is closely related to coal rank and microbial activity. However, when the secondary biogas produced by microbial activity is not absolutely dominant and the gas produced is still mainly thermogenic gas, the sensitivity of the value of δ¹³C_DIC to the CBM productivity response will drop.

Based on the above analysis and the effective indication of carbon source and microbe by δ¹³C_DIC, the geological response mode of δ¹³C_DIC for produced water from medium rank multi- seam commingling production CBM wells has been put forward (Fig. 11). The positive anomaly δ¹³C_DIC is mainly caused by the reduction of methanogens (mainly hydrogenotrophic methanogens). Sedimentary facies and lithology segmentation of coal-measure strata form different superimposed fluid systems, which have different δ¹³C_DIC values of produced water and different archaea communities correspondingly. Under the geological condition that the coal measure strata are over-pressured and medium rank, the produced water from the middle member (middle fluid system) with better permeability and higher water content would have more positive abnormal δ¹³C_DIC, the produced water from the upper member (upper fluid system) has less significant positive anomaly of δ¹³C_DIC than the middle member and mainly Methanobacterium. Whereas the produced water from the lower member (lower fluid system) with lower permeability and water content has lower δ¹³C_DIC value and weaker microbial action. At the same time, the strata near the shallow outcrop are more likely to be recharged by atmospheric precipitation, so the δ¹³C_DIC value of produced water is smaller. Based on the material and physical properties of multi-coal seam sedimentary background, this model considers the methane bacteria participation and mixing of atmospheric precipitation, and reveals the geological mechanism and microbial mechanism of the δ¹³C_DIC difference in produced water from CBM wells in multi-layer commingling production. The model objectively provides effective geochemical evidence for the superimposed fluid systems controlled by sedimentary facies. It can also be used to analyse the gas and water production contribution of layers in CBM wells produced in commingled manner. But the sources of δ¹³C_DIC controlled by sedimentary facies and the mechanism of microbial action need to be studied further.

The research shows that positive anomalies of δ¹³C_DIC are common in produced water from medium rank coal seams. There are more than 15 species of methanogens in the produced water, in which hydrogenotrophic Methanobacterium is the dominant genus, followed by acetic acid-trophic Methanothrix. There is a significant positive correlation between δ¹³C_DIC value and the sequence number of the dominant methanogens in produced water. The positive anomalies of δ¹³C_DIC are caused by reduction of methanogens, especially hydrogenotrophic methanogens.

Segmentation of sedimentary facies and lithology in multi-coal seam measures results in permeability and water content segmentation, which in turn causes the segmentation of δ¹³C_DIC and archaea communities in produced water. Under the geological background of over-pressure and medium rank of coal measure, the produced water from the upper and middle members with higher permeability and water content have positive anomalies of δ¹³C_DIC and mainly Methanobacterium. Produced water from the lower member with lower permeability and water content have lower δ¹³C_DIC value and weak microbial action. The shallow coal seams near the outcrop are easily recharged by atmospheric precipitation, and the value of δ¹³C_DIC is small in produced water.

The geological response model of the δ¹³C_DIC of produced water from medium coal rank CBM wells in multi-layer commingled production reveals the geological and microbial mechanisms of the δ¹³C_DIC difference. It provides effective geochemical evidence for the identification of superimposed fluid systems controlled by sedimentary facies, and can also be used for the analysis of gas and water production contribution of different layers in CBM wells produced in commingled manner.

R_o,max—the maximum reflectivity of vitrinite, %;

Z—Mann-Kendall test factor, dimensionless;

δ¹³C₁—stable carbon isotopic composition of methane in CBM, ‰;

δ¹³C_DIC—stable isotopic composition of dissolved inorganic carbon, ‰.