Chemometric differentiation of natural gas types in the northwestern Junggar Basin, NW China

In recent years, the natural gas has displayed a growing significance in oil and gas exploration in the northwestern Junggar Basin (NWJB), although oil has been the main focus of exploration in the basin. Here, we systematically discuss the classification and origin of the natural gases from the NWJB based on the natural gas geochemistry and chemometric methods. The natural gases collected from the NWJB were chemometrically classified into three groups. Group A gases, defined as coal-derived gases, were likely generated from the mixing of the Jiamuhe Formation and Carboniferous strata. Group B gases, defined as the mixing of coal-derived and oil-associated gases, were restricted to the source rocks of group A and C gases. Group C gases, defined as oil-associated gases, were likely derived from both the Fengcheng and Wuerhe Formations, with a higher contribution from the latter strata. The result of this study suggests that the potential of oil generation in the Wuerhe Formation has been underestimated in the past. This is in accordance with geochemical and geological evidence. This study provides an effective chemometric method of natural gas classification and evaluation of hydrocarbon generation potential. This contributes to a better understanding of the origin of gases and distribution of oil and gas, assisting in exploration deployment in the basin.


Introduction
Chemometrics is a useful tool for the analysis of multivariate data. With noise being identified and removed from the data, useful information can be extracted using chemometrics to show affinities and variations of samples (Kramer, 1988;Peters et al., 2005). Therefore, chemometric methods have been widely used in many disciplines (Bevilacqua et al., 2017;Chabukdhara and Nema, 2012;Madsen et al., 2010), especially in petroleum geochemistry, they have been used for a long time (Kvalheim et al., 1985;Øygard et al., 1984;Peters et al., 1986;Zumberge, 1987). In addition, principal component analysis (PCA) has been frequently applied to reveal the genetic families of crude oils (He et al., 2012;Peters et al., 2013Peters et al., , 2016 in order to make a detailed oil-source rock correlation (Mashhadi and Rabbani, 2015;Wang et al., 2014Wang et al., , 2018. Additionally, multidimensional scaling (MDS) and alternating least squares (ALS) are reliable tools for oil-oil correlation (Wang et al., , 2018 and the quantitative evaluation of the relative contribution from each source to the mixed oil in basins with multiple source rocks (Peters et al., 2008;Zhan et al., 2016aZhan et al., , 2016b. Moreover, PCA has been confirmed to be an effective method for natural gas classification on the basis of chemical composition and stable carbon isotope of gases in previous studies (Wang et al., 2019).
Recently, the abundant oil and gas reserves in the northwestern Junggar Basin (NWJB) have received increasing attention of scholars (Chen et al., 2014;Tao et al., 2016), but information about the genetic types and origins of natural gases in this area remains controversial. Chen et al. (2014) recognized four different types of gases in the Zhongguai Uplift of the NWJB. These types of gases include sapropel-type gas sourced from the Fengcheng Formation (P 1 f) mudstone, humic-type gas derived from the Wuerhe Formation (P 2 w) mudstone, mixed sources of sapropel-and humic-type gases, and mixed sources of sapropel-type gas and deep inorganic gas. Gao et al. (2016) performed an analysis on the genetic types of natural gases from the Wuxia Fault Zone of the NWJB. The result suggests the existence of one gas type that is derived from the Fengcheng Formation (P 1 f). Chen et al. (2016b) proposed that the NWJB gases can be genetically classified into oil-associated gas, coal-derived gas, and mixed gas. The oil-associated gas may have been originated from the Fengcheng Formation (P 1 f), and the coal-derived gas may have been mainly sourced from the Jiamuhe Formation (P 1 j) while part of the contributions may have been derived from the high-maturity source rock of the Wuerhe Formation (P 2 w). The mixed gas may have been mainly derived from the Wuerhe Formation (P 2 w), mixed with the contributions of the highly mature source rock from the Fengcheng Formation (P 1 f). Tao et al. (2016) conducted a detailed research on the geochemistry and origin of natural gases in the Mahu Sag of the NWJB and concluded that three genetic types of natural gases exist in the study area: an oilassociated gas sourced from the Fengcheng Formation (P 1 f), a coal-derived gas sourced from Carboniferous strata and the Jiamuhe Formation (P 1 j), and a coal-derived gas sourced from the Wuerhe Formation (P 2 w). However, systematic studies on the natural gas classification and origin of the NWJB are available only to a limited extent (Chen et al., 2016b).
Moreover, only few studies have reported on natural gas classification using chemometric methods (Wang et al., 2019), and the applications of MDS and ALS in natural gas exploration have not been reported. Here, taking the NWJB as an example, we further study on the natural gas classification and origin of the NWJB using chemometric methods.
Geological setting and geochemical characteristics of potential source rocks The Junggar Basin is in the northern part of Xinjiang Uygur autonomous region, northwestern China (Figure 1(a)). It is divided into four oil-gas zones, namely northwestern, central depression, eastern, and southern parts (Figure 1(b)), based on structural units and the relative concentration of oil and gas. At present, the major oil reservoirs in this basin, such as the Karamay Oilfield in the Mahu Sag, are in the NWJB (Tao et al., 2016). Exploration of natural gas shows promising potential in the NWJB with a total reserve of $2.1 Â 10 12 m 3 estimated by the latest third-round assessment of petroleum resources, although the current focus of exploration is primarily on oil (Chen et al., 2014).
The Permian and Carboniferous strata are essential source rocks for crude oils and natural gases in the NWJB (Cao et al., 2005;Graham et al., 1990). The Permian source rock is mainly composed of a set of widespread mudstone and siltstone in various depositional sags of the basin (Wang et al., 2013). The Permian source rocks of the NWJB exist in the lower Wuerhe (P 2 w), the Fengcheng (P 1 f), and the Jiamuhe (P 1 j) Formations (Cao et al., 2005;Pan et al., 2003). The Carboniferous source rock in the NWJB is mainly dark volcanic mudstones in marine, swamp, and lagoon environments. All these sets of source rocks have the potential for hydrocarbon generation, but the kerogen type and maturity level are slightly different. The P 2 w and P 1 f have great potential to generate oil with a relatively low maturity, while the P 1 j and Carboniferous strata are source rocks of gas with a relatively high maturity (Chen et al., 2016a;Gao et al., 2016;Meng and Parhati, 1999;Tao et al., 2016).

Samples and methods
In this study, we used 71 gas samples published in previous studies (Chen et al., 2014;Tao et al., 2016;Yang et al., 2008) to classify types of natural gas in the NWJB (supplementary Appendix A; Figure 1(c)). The gas samples from previous publications were screened as  follows before the data set was determined: (1) gas samples with abnormal values of their geochemical composition were excluded and (2) the mean values of proxies obtained from the same well and depth were used for chemometric analysis. After data collection, PCA and MDS were applied to identify the types of natural gas in the NWJB, and ALS was introduced to de-convolute the mixed oil-associated gases. The chemometric analysis includes six geochemical parameters: C 1 (%), C 2 (%), C 3 (%), d 13 C 1 , d 13 C 2 , and d 13 C 3 , which are similar to the parameters used in a previous study (Wang et al., 2019). The PCA was performed using commercial software (Pirouette 4.5, Infometrix Inc.), and the MDS was performed with in-house software . The PCA settings were programmed as follows: preprocessing ¼ range scale, maximum factor ¼ 3, validation method ¼ none, and row-¼ none. The MDS settings were the same as that for the PCA, but the Bray-Curtis distance was used for the measurement of similarity . The ALS settings were programmed as follows: preprocessing ¼ range scale, maximum sources ¼ 2, and initial estimates from ¼ rows.

Natural gas classification using PCA and MDS
The results of the PCA of the six selected proxies of chemical and carbon isotopic compositions showed three genetically distinct natural gas groups: A, B, and C (Figure 3(a)). These inferences are similar to the results of a previous study that a d 13 C 2 value is used to determine the genetic types of natural gases (Dai et al., 2005). Similar results are obtained from the MDS plot, indicating that this method is also effective for natural gas classification (Figure 3(b)). Gases of groups A and B occurred mainly in the Zhongguai Uplift, whereas those of group C occurred mainly in the Mahu Sag and adjacent uplifts (e.g. Ke-Bai and Wu-Xia Fault Zones). Distinct gas source rocks of each group can be inferred from different ranges of chemical and carbon isotopic compositions (Table 1).
Group A gases were composed of 12 samples, collected from the Permian reservoir in the P 2 w and P 1 j formations in the Zhongguai Uplift. The gases in this group were characterized by a relatively high content of methane, ranging from 90.51% to 95.63%, with a mean value of 93.61%. The d 13 C 2 and d 13 C 3 values of group A gases were the highest among the three groups, ranging from -27.3& to -23& and -26.6& to -20.1&, respectively. Thus, group A gases were defined as coal-derived gases because the d 13 C 2 values of such gases are commonly higher than À27.5& (Dai et al., 2005).
Group B gases consisted of 11 samples collected from the Triassic and Permian reservoirs in the T 1 b, P 1 f, P 2 w, and P 1 j formations in the Zhongguai Uplift, Wu-Xia Fault Zone, and Mahu Sag. In terms of the mean values of chemical and carbon isotopic compositions, group B gases had a moderate content of methane and d 13 C 2 and d 13 C 3 values similar to those of the gases in group A and C. This probably indicates a mixed source of the gases in group B (Dai et al., 2005), which is also supported by results of the PCA and MDS plots, indicating that the gas samples of this group take an intermediate position between the gases of group A and group C.
Group C gases, the most common types of gases in the NWJB, contained 48 samples. The gases of this group were distributed in the Triassic, Permian, and Carboniferous strata of the entire study area. The gas was characterized by a low content of methane from 64.68% to 95.28%, with a mean value of 82.03%. In addition, the d 13 C 2 and d 13 C 3 values of group C gases were relatively lower in comparison with those of the gases in groups A and B, and therefore, these gases are oil-associated (Dai et al., 2005).

Possible gas-source correlations
Overall, three groups of natural gases showed a wide range of chemical and carbon isotopic compositions, suggesting that they were likely derived from multiple sources. Notably, another significant characteristic of natural gases in the NWJB is the partial reversal in stable carbon isotopes between propane and butane ( Figure 4). This can be well explained by mixed sources for the three group gases. Dai et al. (2004) proposed that partially reversed  (6) d 13 C orders for natural gas alkanes may have been caused by (1) bacterial oxidation, (2) mixing of biogenic and thermogenic gases, (3) mixing of gases from two source rock intervals or one source rock unit with different maturity, and (4) mixing of sapropelic and humic gases. The d 13 C 1 values for all of the gases in the NWJB were in the range of -54.4& to -26.6&, indicating that the natural gases were unlikely that they originated from a significant mixing of bacterial oxidation or biogenic and thermogenic gases. Therefore, the most plausible cause for the partial isotope reversal in the NWJB gases is the mixing of gases from two source rock intervals or one source rock unit with different maturity or mixing of sapropelic and humic gases.
In accordance with the results of Tao et al. (2016), geochemical characteristics of group A gases were similar to those of the source rocks in the Jiamuhe Formation and Carboniferous strata based on biomarker proxies of the condensates from wells (e.g. the Ke75 and Ke77). This is also supported by the kerogen types of source rocks in the Jiamuhe Formation and Carboniferous strata (Figure 2). Moreover, the carbon isotopic composition of kerogens in different source rock intervals indicates a predominance of Type-II kerogen in source rocks of the Fengcheng Formation and Type-III kerogens in source rocks of the Jiamuhe Formation and Carboniferous strata ( Figure 5). Therefore, group A gases were likely derived from the Jiamuhe Formation and Carboniferous strata. However, the contribution of source rocks in the Wuerhe Formation to group A gases cannot be inferred and needs further research.
Group B gases were most likely derived from the mixing of sapropelic and humic gases because of the moderate d 13 C 2 values , ranging from -29.1& to -26.8& (Dai et al., 2005). In addition, group B gases were mainly distributed in the middle of group A and C gases as shown in Figures 3 and 4, suggesting that group B gases were probably derived from a mixture of source rocks from group A and C gases.  As discussed above, group C gases are oil-associated gases. Group C gases have a wide range of the chemical and carbon isotopic composition as well as partially reversed d 13 C order between propane and butane, as a result of two periods of accumulations in the Fengcheng Formation (Tao et al., 2016). However, previous studies have confirmed that the two sets of source rocks in the Fengcheng and Wuerhe Formations have great potential to generate oil in the NWJB, and currently discovered oils are mainly derived from the Fengcheng Formation (Chen et al., 2016bYu et al., 2017). Nevertheless, some other studies proposed that the contribution of source rocks in the Wuerhe Formation to the oil generation in the NWJB was underestimated in the past (Chen et al., 2016c(Chen et al., , 2016d. This can be attributed to a limited understanding of source rock distribution as the source rocks of the Wuerhe Formation revealed by drilling nowadays are mainly located at the margin of the basin with little in the center of the basin (Chen et al., 2016b). Briefly, group C gases likely consisted of mixed oil-associated gases from either the Fengcheng and Wuerhe Formations (P 2 w) or the two periods of accumulation in the Fengcheng Formation (P 1 f). To answer these inferences, we introduced ALS to de-convolute group C gases. Since the sample number for chemometrics should usually be more than 30 , only group C gases were discussed in this paper. In the light of the ALS results (Table 2), group C gases had two possible sources (source 1 and source 2), and the contribution of source 2 to the gases in group C was higher than that of source 1 (Figure 6). Table 3 shows the stable carbon isotopic composition of the gaseous hydrocarbons (C 1 -C 3 ) from the Fengcheng Formation (P 1 f) under thermal simulation at different heating temperatures. The carbon isotopic composition of source 1 was similar to that of the Fengcheng Formation (P 1 f) at 350 C, while that of the source 2 seemed to be independent of that from the Fengcheng Formation (P 1 f). Thus, source 1 was probably derived from mature source rocks of the Fengcheng Formation (P 1 f), while source 2 was most likely derived from the Wuerhe Formation (P 2 w). This conclusion is also supported by similar ranges of the carbon isotopic composition between the P 2 w and P 1 f extracts and crude oils in the NWJB (Figure 7). Also, these results are in a good agreement with the three events of petroleum generation in the NWJB during the Middle-Late Permian, Late Triassic, and Early Cretaceous, corresponding to the source rocks in the Jiamuhe, Fengcheng, and Wuerhe Formations, respectively (Cao et al., 2005). In summary, group C gases were likely derived from the Fengcheng and Wuerhe Formations, with a higher contribution from the latter strata. Therefore, the oil generation potential of the Wuerhe Formation in the NWJB was probably underestimated in the past, and more attention should be paid in future exploration.

Conclusions
The results of two chemometric methods based on the PCA and MDS of six proxies of chemical compositions and isotopes signatures in natural gases in the NWJB show that the natural gases can be divided into three genetic groups: A, B, and C. Group A, B, C gases were determined as coal-derived gas, mixing of coal-derived and oil-associated gas, and oilassociated gas, respectively. All three groups of natural gases showed a wide range of their chemical and carbon isotopic composition, and there was a partial reversal of d 13 C values between propane and butane, indicating that these gases were derived from diverse sources. Group A gases may have been mainly derived from the mixing of sources in the Jiamuhe Formation and Carboniferous strata, while group C gases were probably derived from the Fengcheng and Wuerhe Formations. Noticeably, in accordance with the ALS results, the contribution of source rocks in the Wuerhe Formation to group C gases was higher than that of the Fengcheng Formation. This study provides an effective chemometric method of natural gas classification and evaluation of hydrocarbon generation potential. This contributes to a better understanding of origin of gases and distribution of oil and gas, assisting in exploration deployment in the basin.