An Analysis of Multi-Coal Seam Mining Impacting Aquifer Water Based on Self-Organizational Maps

Abstract

The degradation of groundwater quality due to mining activities is a major public concern globally. This study employed a combination of methods (multivariate statistics, Self-organizing mapping, and PHREEQC hydrogeochemical simulation) to uncover the hydrochemical characteristics and processes of mine water in the Kailuan mining area. Self-organizing mapping (SOM) clustering divided the mine water into three groups, TDS values gradually increased from the first to the third group, and the hydrogeochemical type of mine water gradually changed from Na-HCO₃ and CaMg-HCO₃ to CaMg-SO₄, Na-Cl, and mixed types. Principal component analysis (PCA) revealed that water–rock action and evaporation concentration were major ion concentration factors. According to the molar ion concentration ratio method, the main ions in mine water in Kailuan mining area originate from silicate and sulfate, and a small amount from carbonate rock weathering, and they are influenced by cation exchange. As a result of the PHReactor EQuilibrium Code (PHREEQC) simulation results, it can be concluded that better hydrodynamic conditions in mines are primarily controlled by carbonate dissolution. Mine water with poorer hydrodynamic conditions is mainly controlled by sulfate and carbonate dissolution, with sulfate dissolution having a greater effect. The results of this study provide an important scientific basis for the safe mining of mines and the protection of groundwater resources.

1. Introduction

Groundwater resources, widely distributed and less susceptible to pollution, have been exploited by humans [1], playing a vital role in ecosystems and supplying water for agriculture, industry, and domestic use globally [2,3,4]. In recent years, a series of factors such as domestic sewage, industrial wastewater, mining, and pesticide pollution have led to severe groundwater contamination, resulting in a slight decline in groundwater quality [5,6]. With rapid economic development, the demand for coal, an important global energy source, has increased greatly. However, coal mining may change the groundwater system, groundwater flow paths, and the hydraulic connection between aquifers. The annual discharge of mine water in China is as high as 7.4 billion m³, and with a large amount of mine water evacuation, the balance and stability of the underground water-bearing system were affected. Due to the special water quality and complex composition of mine water, it is very easy to cause groundwater pollution [7,8,9]. In many cases, valuable mine water with a high quality is not efficiently or wisely used, leading to the squandering of groundwater and clean mine water resources [10]. Therefore, studying the properties and underlying reasons for groundwater chemistry in coal mining regions can offer a crucial scientific foundation for ensuring safe mining practices and protecting groundwater resources in mines.

Water chemistry analysis methods, such as the Piper diagram and Gibbs diagram, have been extensively researched and utilized in various studies. These diagrams offer visual insights into groundwater chemistry, components, water quality, and hydraulic connections between groundwater sources and aquifers [11,12]. Additionally, the application of multivariate mathematical statistics enables the extraction of crucial information from extensive water chemistry datasets, revealing inherent patterns among aquifer components [13]. The ion ratio coefficient method aids in determining groundwater characteristics by examining ion ratio patterns [14]. Furthermore, computer numerical simulation techniques, when integrated with GIS, can identify groundwater recharge sources and unveil spatial distribution patterns of chemical substances in water [15]. Qu et al. conducted hydrochemical and isotopic analyses of 61 groundwater samples from typical coal fields in northwestern China using Self-organizing mapping (SOM) and Entropy Weighted Quality Index (EWQI) methods. The general chemical and qualitative characteristics of sulfate were investigated, the potential health risks of sulfate were assessed, and the geological and anthropogenic drivers of sulfate pollution were elucidated [16]. Zhang Miao et al. analyzed water–rock interactions in the Huaibei Plain mining area using water chemistry analysis, multivariate statistical analysis methods, and PHReactor EQuilibrium Code (PHREEQC) numerical simulation and identified the main hydrogeochemical transport processes [14].

Kailuan mining area in North China has a mining history of more than 100 years. Many scholars have carried out a series of studies on the groundwater in the Kailuan mining area, but few studies have been conducted on the 12–14 aquifer water alone. The study of 12–14 aquifer water is particularly important because it affects both 12 and 14 coal mining.

This study focuses on identifying the factors that regulate and the hydrogeochemical processes that govern mine water in the Kailuan mining region. Water samples were collected from coal aquifers located at significant burial depths, and their chemical composition, formation mechanisms, and primary ion origins were analyzed.

2. Study Area

The Kailuan mining area is situated in Tangshan City, Hebei Province, China. This region is characterized by a Carboniferous Permian coalfield within the coal-rich North China region, housing several mines including Zhaogezhuang Mine, Tangjiazhuang Mine, Linxi Mine, Majiagou Mine, Lujiatuo Mine, Tangshan Mine, Fangezhuang Mine, among others. The mining site spans an area exceeding 670 square kilometers and is characterized by the presence of coal-bearing strata from the Carboniferous and Permian Systems. These strata have a combined thickness of approximately 150 m and consist of 15 to 20 layers of coal, notably including coal seam 7 and coal seam 9. The topography of the region is predominantly an alluvial plain situated at the foothills, with elevations ranging from +10 to +70 m and a gradual decline from north to south. The area is intersected by four seasonal rivers—Shahe River, Shiliu River, Douhe River, and Jiyun River—which ultimately discharge into the Bohai Sea.

In the vicinity of the Kailuan coalfield, a northeast-oriented syncline is present, characterized by uplift and a southwestward plunge. The structure exhibits an approximate 20-degree inclination on its southeast flank, with a slightly steeper angle observed on the northwest side. The syncline is delineated by the Late Permian stratum at its core, flanked by the Permian coal measures, and bordered by Ordovician limestone along its periphery. The entire syncline is overlain by the Quaternary system, with the overlying sediment thickening from the northeast to the southwest of the coalfield, exceeding 800 m in thickness. The depth of the upper surface of the Ordovician system at the central axis typically does not exceed 2000 m.

Since the Paleozoic coal-bearing strata were established in the research area, a tectonic framework has emerged featuring four coal-bearing synclines: the Kaiping syncline, Cheqianshan syncline, Jinggezhuang syncline, and Xigongyao syncline. This framework has been shaped by the overlay of various tectonic processes, resulting in the formation of secondary fold structures like the Dujunzhuang anticline, Lujiatuo anticline, Fangezhuang syncline, and Bigezhuang syncline in localized areas. These secondary structures intersect with the Kaiping syncline along its axial direction.

Multiple groundwater aquifers are present within the Kailuan mining field (Figure 1). Aquifers I, VI, and VII serve as indirect mine-filling aquifers, while aquifers II, III, IV, and V function as direct mine-filling aquifers.

3. Materials and Methods

3.1. Sampling and Testing

In total, 89 samples were selected for water sampling, including 33 in Donghuantuo coal mine (DHT), 9 in Zhaogezhuang coal mine (ZGZ), 20 in Qianjiaying coal mine (QJY), 13 in Tangshan coal mine (TS), and 9 in Lvjiatuo coal mine (LJT).

Water samples were mainly collected from the 12–14 aquifer. Before collecting water samples, the 550 mL plastic bottle and its cap were rinsed three to five times with sample water. Water samples were stored and processed at low temperatures to inhibit redox and biochemical reactions. Water sample chemistry tests included Na⁺, Ca²⁺, Mg²⁺, Cl⁻, SO₄²⁻, and HCO₃⁻. The concentration of HCO₃⁻ was determined by acid-base titration. Na⁺, Ca²⁺, Mg²⁺, Cl⁻, and SO₄²⁻ were determined by ion chromatography (ICS-600). All water samples maintained a charge balance, with error < 5%.

3.2. Multivariate Statistical Analysis

Self-organizing mapping (SOM) network model was proposed and updated by Kohonen (1982 and 2014), which is an unsupervised learning neuronal network model that can map the input n-dimensional spatial data to a lower dimensional output and can maintain the original topological logical relationships of the data. It identifies and interprets spatial patterns and similarities in groundwater quality datasets, as well as significant discrepancies between groups of samples [17,18].

SOM consists of a single-layer neural network containing only input and output layers, and the output layer is also called the computational or competitive layer. SOM learning consists of a competitive process, a collaborative process, and an adaptation process. The steps are as follows.

(1) Initialize the SOM network. The weight $W_j$ of each node in the output layer is randomly assigned an initial value, and defined as the training end condition.

(2) Select an input vector $X_i$ from the input samples randomly and calculate the connection weight vector with the minimum distance from $W_j$ in $X_i$, as shown in Formula (1).

$$‖X_i-W_g‖=\underset{j}{\min }‖X_i-W_j‖$$

(1)

where ‖ ‖ is the distance function, and the European distance is used here.

(3) Define g as the winning unit, and N_g $\left(t\right)$ is the adjacent area of the winning unit. For the cells in the adjacent area, adjust the weight according to Formula (2) to make them close to $X_i$

$$W_j\left(t+1\right)=W_j\left(t\right)+\alpha \left(t\right)h_{gj}\left(t\right)[X_i\left(t\right)-W_j(t\left)\right]$$

(2)

where $t$ is the number of studies, $\alpha \left(t\right)$ is the learning rate of the $t$ time, and $h_{gj}\left(t\right)$ is the neighborhood function of g.

(4) With the increase in learning times, repeat steps (2) and (3) and stop the training when the training conditions are reached.

(5) Output specific cluster number and cluster center.

Component analysis is one of the multivariate statistical methods, abbreviated as PCA, which uses the idea of dimension reduction to transform multiple original related variables into several uncorrelated and orthogonal synthetic variables (i.e., principal components). Each synthetic variable obtained after transformation can reflect most of the information of the original variable.

In the process of analyzing the source of groundwater chemical composition, the PCA method can extract factors with similar sources and find out the relationship between each factor [19,20]. Principal component analysis includes five steps: (1) data standardization; (2) calculate the correlation coefficient matrix; (3) calculate the characteristic value; (4) select the main components; and (5) calculate the principal component score.

3.3. Modeling of Hydrogeochemical Reactions

PHREEQC is a powerful and widely used program for modeling hydrogeochemical processes. PHREEQC can perform the following calculations: (i) component morphology and saturation indices; (ii) transport processes for intermittent reactions and one-dimensional reversible and non-reversible reactions; and (iii) inverse hydrogeochemical simulation. In this study, the program was utilized to calculate the saturation index (SI) for 89 groundwater samples and to simulate the hydrogeochemical processes occurring along the flow path of the aquifer [21,22].

4. Results and Discussions

4.1. General Physical and Chemical Statistical Characteristics

The statistical results of the main chemical composition of mine water in each coal mine in Kailuan mining area are shown in Figure 2.

The medication concentrations in all mine water except Lvjiatuo were Na⁺ > Ca²⁺ > Mg²⁺ in the order of Na⁺ > Mg²⁺ > Ca²⁺. The order of anions is HCO₃⁻ > SO₄²⁻ > Cl⁻, except for Qianjiaying and Lvjiatuo (SO₄²⁻ > HCO₃⁻ > Cl⁻ and HCO₃⁻ > Cl⁻ > SO₄²⁻). The average pH value of mine water in most coal mines is higher than 7, indicating that the mine water is weakly alkaline.

Comparing the average values of the main chemical components of the coal mines (Figure 2 and Table S1), all coal mines have an average value of Na⁺ greater than 130 mg/L except for the Donghuantuo mine. The highest average values of Na⁺, Ca²⁺, and TDS are 193.73 mg/L, 126.00 mg/L, and 1140.54 mg/L for the Qianjiaying mine. The highest average value of Mg²⁺ was 108.70 mg/L, while the average value of other coal mines was less than 50 mg/L. The highest average value of Cl⁻ was 34.08 mg/L. The average value of SO₄²⁻ was greater than 100 mg/L in Qianjiaying, Tangshan, and Linxi mines, with the highest average value of 557.81 mg/L in Qianjiaying mine. The average value of HCO₃⁻ was greater than 300 mg/L, with the highest value of 606.69 mg/L in Zhaozizhuang mine.

4.2. SOM Algorithm Results

The color depth of each neuron represents the component value of the chemical parameters of the water sample point. The graph intuitively presents the distribution of neuron distance and corresponding color depth to explain the information and qualitative relationship between various hydrochemical parameters. SOM results (Figure 3a) show that TDS values have an obvious correlation with SO₄²⁻, Ca²⁺, Na⁺, and Mg²⁺ plasma in water samples, indicating that they are all major contributors to TDS. Na⁺ and Ca²⁺ are positively correlated with SO₄²⁻, indicating the dissolution of sulfate minerals (gypsum and anhydrite).

Na⁺ and Cl⁻ in the sample are positively correlated ions. This shows that the dissolution of rock salt/potassium salt minerals affects the mine water chemistry [23].

According to the SOM clustering results (Figure 3b) and Table S1, group 1 is located in the upper left yellow area of the label map, accounting for 67.4% of the total number of samples. The second group is located in the lower right green area of the label graph, accounting for 23.6% of the total number of samples. The hydrogeochemical characteristics of the first group and the second group are that the main ion concentration and TDS value are low, indicating that the hydrodynamic conditions of the two groups of mines are good and the exchange capacity is strong. The third group (9%) is located in the blue area at the lower left of the label map. Except for HCO₃⁻, the concentration of Ca²⁺, Mg²⁺, Na⁺, and other major ions in this group are higher than those in the first and second groups. It shows that the hydrodynamic conditions of this group of mines are poor, and the ion concentration increases due to water–rock interaction and evaporation concentration.

4.3. Hydrochemical Types

The Durov diagram and Piper diagram of three groups of mine water are obtained by using Origin2021 water chemical analysis software. The results showed that the TDS value increased from group 1 (120.91 mg/L) to group 3 (4205.02 mg/L). The hydrogeochemical type of mine water gradually changes from Na-HCO₃ and CaMg-HCO₃ to CaMg-SO₄, Na-Cl, and mixed type (Figure 4).

4.4. PCA and Correlation Analysis

The TDS concentrations of the three groups of mine water were significantly different, as well as the correlation between TDS values and each ion, and the source of the ions and the cause of the TDS variation can be explained by analyzing the correlation between each ion and TDS [24].

Using Origin software, PCA and correlation analysis were performed on TDS and conventional ions of three groups of mine water (Figure 5). Three principal components (PCs) were identified, which explained 80.5% of the total variance.

PC1 explains 46.4% of the total variance, which is highly correlated with TDS, Ca²⁺ and SO₄²⁻. It can be inferred from the Gibbs diagram that PC1 represents the water–rock interaction (dissolution/precipitation of sulfate) and the natural evolution of groundwater (evaporation and concentration). PC2 explains 23.4% of the total variance, which is highly correlated with Na⁺ and HCO₃⁻. Combining the Gibbs diagram, it can be inferred that PC2 may be related to salt rock dissolution, evaporation, and concentration. PC3 explains 10.6% of the total variance and has an obvious correlation with Mg²⁺ and Cl⁻, indicating the ion exchange between sulfate minerals and carbonate minerals.

It can be seen from the figure that the first group of TDS has a positive correlation with Ca²⁺, Mg²⁺ and SO₄²⁻, the second group of TDS has a positive correlation with Na⁺, Ca²⁺, SO₄²⁻ and HCO₃⁻, the third group of TDS has a positive correlation with Ca²⁺, Na⁺, SO₄²⁻ and HCO₃⁻, and the correlation coefficients of TDS with Ca²⁺ and SO₄²⁻ are large. The TDS of mine water in this area is mainly affected by Ca²⁺ and SO₄²⁻, followed by Na⁺, Mg²⁺ and HCO₃^-, indicating that there was sulfate carbonate leaching. The correlation between TDS and Ca²⁺ is significantly greater than that between Na⁺ and Mg²⁺; Ca²⁺ and Mg²⁺ are positively correlated, and Ca²⁺ and Na⁺ are weakly negatively correlated, which indicates that there is also a cation exchange effect leading to an increase in the content of Na⁺ components and a decrease in the content of Ca²⁺ and Mg²⁺ components, so the early mine water type changed from Na-HCO₃ and CaMg-HCO₃ to CaMg-SO₄, Na-Cl and mixed type.

In group 2, it can be seen that the correlation between Na⁺ and Cl^- is weak, while the correlation coefficient between Na⁺ and Ca²⁺, and Mg²⁺ is close to 1, and the correlation between TDS and SO₄²⁻ is strong, indicating that the mine water chemistry in this group is mainly controlled by ion exchange rather than the dissolution of salt rock and sodium feldspar. The properties of group 1 are similar to those of group 2.

In addition, the correlation coefficients of SO₄²⁻, Ca²⁺ and Mg²⁺ in the first group are 0.84 and 0.95, respectively, which are slightly higher than those in the second group, indicating that the chemical composition of the two groups of mine water may depend on the dissolution of gypsum and sulfate, and the degree of dissolution tends to increase in the later period. In addition, the correlation between Na⁺ and HCO₃^- reaches 0.85, indicating the contribution of carbonate rocks.

The third group showed that Ca²⁺ and SO₄²⁻, Cl⁻ were positively correlated, and the correlation coefficients of TDS and SO₄²⁻ were close to 1, indicating that the chemical properties of mine water in this group were mainly controlled by the dissolution of sulfate.

Overall, water–rock action and evaporative concentration are the main factors affecting the concentration of major ions.

4.5. Analysis of Ion Source Based on Ion Combination and Ratio Method

Gibb’s plots were initially used to determine the evolution of surface water chemistry, but it has also been used to analyze the formation mechanism, such as evaporation concentration control type, rock weathering control type and precipitation control type [25].

As shown in Figure 6, water samples of group 3 were mostly distributed in the rock weathering area and had high TDS values, indicating that the main ion formation effects of this group of water samples are all controlled by rock weathering. Group 1 and group 2 were mainly distributed in the rock weathering area, indicating that the group of water samples are mainly affected by rock weathering. Group 3 and group 1 are distributed in the evaporation concentration area, indicating that the water samples were affected by the combined effect of rock weathering and evaporation concentration. All water sample points are a long way from the atmospheric precipitation area, indicating that atmospheric precipitation has little influence on the chemical properties of mine water. In addition, a portion of the water sample points are distributed outside the region, which may be linked to human activities (mining). However, due to the wide range and complexity of groundwater ion sources, other methods need to be selected for further analysis of ion sources.

In group 2, the values of Na⁺/(Na⁺ + Ca²⁺) ranged from 0.7 to 1.0 (Figure 6a). This indicates that the principal reaction occurring in group 2 is the dissolution of rock salt and carbonate. The degree of evaporation and rock salt dissolution determines the change in Cl⁻/(Cl⁻ + HCO₃⁻) (range 0.1 to 0.2) in agreement with the results of the saturation index analysis.

During the chemical evolution of groundwater, different ions show different contributions and different ion ratios have different meanings [26,27]. In order to explore the main types and chemical compositions of weathered rocks, three groups of water samples were analyzed by ion ratio analysis [28,29,30].

Combined with the ion proportional relationship analysis, the Origin 2018 software was used to draw Figure 6b. The results show that all groups of water samples are mainly distributed between the silicate rock and evaporite control end elements, and are closer to the silicate rock control end elements, indicating that the mine water in the study area is mainly controlled by the weathering dissolution of the silicate rock and evaporite. Most of the water sample points are far from the carbonate control end element, indicating that the mine water in the study area is less controlled by the weathering and dissolution of carbonate rocks.

Analysis of the ratio relationships among groundwater components can further reflect the water chemistry processes under natural water–rock interactions [31,32]. Na⁺ in water bodies is mainly derived from atmospheric precipitation, silicate minerals, and evaporated salt mineral dissolution. Under natural conditions, salt rock dissolution is the main source of Na⁺ and Cl⁻ in groundwater, and the milligram equivalent concentration ratio relationship is generally around 1. As can be seen in Figure 7a, the ratio of Na⁺ to Cl⁻ in all mine water samples from Kailuan mining is much greater than 1, indicating that the Cl⁻ in mine water is not sufficient to balance Na⁺ and that Na⁺ may have other sources, and may also originate from weathering of silicate rocks or cation exchange [33]. Generally, Ca²⁺, Mg²⁺, HCO₃⁻ and SO₄²⁻ in groundwater mainly come from the dissolution of carbonate or silicate and evaporites (such as sulfate), such as calcite (CaCO₃), dolomite (CaMg (CO₃)₂) and gypsum (CaSO₄·2H₂O).

The Ca²⁺/SO₄²⁻ or Mg²⁺/SO₄²⁻ values are close to 1, which usually indicates that Ca²⁺, Mg²⁺, and SO₄²⁻ in groundwater mainly originate from the dissolution of sulfate minerals, such as gypsum. In Figure 7b, the SO₄²⁻/Ca²⁺ values of group 1 and group 3 are all close to 1, indicating that the early mine water chemistry is mainly influenced by the dissolution of sulfate minerals.

(Ca²⁺ + Mg²⁺)/(HCO₃⁻) can reflect the contribution of weathering of carbonate rocks to Ca²⁺, Mg²⁺, and HCO₃⁻ in groundwater [9]. In Figure 7c, the water sample points of group 3 are mainly located below the 1:1 line, indicating that the water samples of this group are less controlled by carbonate dissolution. On the contrary, the water sample points in group 2 are mainly located above the 1:1 line, which indicates that the water samples in this group are more controlled by carbonate dissolution. The water sample points from group 2-group 1-group 3 gradually tend to be below the 1:1 line, indicating that the water samples are less and less controlled by carbonate dissolution as the mineralization increases, which may involve other hydrogeochemical processes.

(Ca²⁺ + Mg²⁺)/( HCO₃⁻ + SO₄²⁻) can further determine the main source of Ca²⁺, and Mg²⁺, and if the ratio is close to 1, it means that Ca²⁺ and Mg²⁺ were mainly from the dissolution of carbonate and gypsum [34]. A ratio greater than 1 indicates that Ca²⁺ and Mg²⁺ mainly originate from the dissolution of silicates and sulfates, while the opposite indicates that Ca²⁺ and Mg²⁺ originate from the dissolution of carbonate rocks.

In Figure 7d, the sample points of both group 1 and group 3 are distributed below the 1:1 relationship line, and nearly all water samples have less Ca²⁺ + Mg²⁺ than HCO₃⁻ + SO₄²⁻, indicating that Ca²⁺ and Mg²⁺ mainly originate from the weathering of silicate and sulfate, and most water samples of group 1 are distributed near the 1:1 line, indicating that Ca²⁺, Mg²⁺ mainly originate from the dissolution of carbonate and gypsum.

The milligram equivalent ratio of (Ca²⁺ + Mg²⁺)-(SO₄²⁻ + HCO₃⁻)/(Na⁺-Cl⁻) usually reflects whether cation replacement occurs in the groundwater of the study area. When the correlation is negative, cation replacement occurs and vice versa. As can be seen from Figure 8, cation replacement occurs, and the concentration of Ca²⁺ and Mg²⁺ ions decreases with the increase in Na⁺ ions during the dissolution of part of the rock salt [35].

In summary, the main ions in the mine water of Kailuan mining are derived from sulfate and a small portion from carbonate rock weathering dissolution and are influenced by Ca²⁺, Mg²⁺, and Na⁺ cation exchange.

4.6. Hydrogeochemical Simulation Results

The factors affecting the water chemistry of mine water quality are complex and diverse. In this study, the amount of mineral precipitation precipitated or dissolved in solution was determined by inverse simulation (INVERSE_MODELING) calculations using PHREEQC 2.5 software to comprehensively assess the mine water chemistry processes in Kailuan mining and analyze water–rock interactions to provide quantitative analysis results.

According to the results of SOM clustering and the Durov plot, TDS from group 2-group 1-group 3 is getting higher and higher. In combination with the hydrogeological conditions of the mining area, two simulation paths are set up using Qianjiaying mine water samples. In addition, the uncertainty limit is set to 0.05.

The changes in the hydrochemical components of the solution at the beginning and end of the inversion path were shown in Figure 9.

The saturation indices and concentration variations of the starting and ending solutions of the two inversion paths are shown in Figure 9a. Both calcite and dolomite are supersaturated, indicating that calcite and dolomite are the main minerals in the coal aquifer of Kailuan mining. Gypsum and halite both show negative values, indicating their dissolution (Figure 9b).

According to the hydrogeological situation of Kailuan mining and the results of the previous analysis, the following reactive minerals were selected for the “possible reactive mineral phases” in the groundwater hydrogeochemistry: calcite, dolomite, gypsum, halite, CO₂, CaX₂, NaX, MgX₂ and Cl⁻, SO₄²⁻.

On the simulated paths S1 and S2, calcite precipitates initially and subsequently dissolves (Table S2). Under the combined action of calcite and cation exchange, calcium ions increased first and then decreased. Gypsum was dissolved continuously, and the dissolution tended to increase in the later stage. Under the influence of the continuous dissolution of gypsum, the sulfate concentration increased, and the increase was greater in the later stage. Gypsum dissolves in mine water in different simulated paths and is accompanied by cation exchange. The water of mines with better hydrodynamic conditions is mainly controlled by carbonate dissolution, and the water of mines with poor hydrodynamic conditions is mainly controlled by sulfate and carbonate dissolution, and the sulfate dissolution control is stronger, which is consistent with the results of multivariate statistical analysis and hydrochemical analysis. The hydrogeochemical reactions of groundwater in different periods have obvious differences, which is also the result of the comprehensive influence of geological complexity.

5. Conclusions

Studying the chemical properties and potential causes of groundwater in coal mining areas can provide an important scientific basis for ensuring safe mining and protecting groundwater resources. This study aimed to reveal the hydrochemical characteristics and processes of mine water. The following conclusions were obtained.

(1) SOM clustering divided the mine water into three groups, and the TDS values increased from the first group (120.91 mg/L) to the third group (4205.02 mg/L). The hydrogeochemical type of mine water gradually changed from Na-HCO₃, CaMg-HCO₃ to CaMg-SO₄, Na-Cl, and mixed type.

(2) PCA combined with personal correlation analysis obtained three principal components: PC1 explained 46.4% of the total variance, with a higher correlation with TDS, Ca²⁺, and SO₄²⁻, which may represent water–rock interaction (dissolution/precipitation of sulfate) and natural evolution of groundwater (evaporation concentration); PC2 explained 46.4% of the total variance, with a higher correlation with Na⁺ and HCO₃⁻, which may be related to the dissolution of salt rock, evaporation concentration of groundwater; PC2 explains 23.4% of the total variance and correlates with Na⁺ and HCO₃⁻, which may be related to the dissolution of salt rock and evaporation concentration; PC3 explains 10.6% of the total variance and correlates significantly with Mg²⁺ and Cl⁻, indicating the ion exchange between sulfate minerals and the ion exchange effect of carbonate minerals.

(3) The molar ion concentration ratio method was used to show that the main ions in mine water in Kailuan mining area originated from silicate and sulfate, and a small portion from carbonate rock weathering dissolution and was influenced by Ca²⁺, Mg²⁺, and Na⁺ cation exchange.

(4) PHREEQC simulation results show that gypsum and calcium exchange exists in the dissolved form on path S1 and S2. The water of mines with better hydrodynamic conditions is mainly controlled by carbonate dissolution, and the water of mines with poor hydrodynamic conditions is mainly controlled by sulfate and carbonate dissolution, and the sulfate dissolution control is stronger, which is consistent with the results of multivariate statistical analysis and hydrochemical analysis.

In future research, long-term monitoring should be conducted and more water chemistry ions should be tested to conduct more in-depth research on the long-term hydrogeochemical evolution of mine water.