Mechanism of groundwater inrush hazard caused by solution mining in a multilayered rock salt mining area : A case study in Tongbai , China

. School of Environmental Studies, China University of Geosciences, Wuhan 430074, Hubei, P.R.China. . Three Gorges Research Center for Geo-Hazard, Ministry of Education, Wuhan 430074, Hubei, P.R.China. . Department of Geological Engineering, Hubei Land Resources Vocational College. Wuhan 430074, P.R.China. . Hydrological Engineering Environment Technology Consulting Co. Ltd. Wuhan 430074, P.R.China.


Introduction
Solution mining is commonly used in salt mine exploitation, as salts are soluble in water.In this method, high-pressure and -temperature water with low salinity is injected into a mineral deposit through production wells to dissolve the mineral salts.After being drawn from the wells, the soluble salt is purified and processed further.However, the high-pressure and -temperature water used in this process not only dissolves minerals but also cause fractures in the strata, which usually results in hazards, such as brine leakage or groundwater inrush.In this situation, drinking groundwater for the public is normally polluted following groundwater inrush, thus creating a hazard and threatening the health of local residents.
The rock-salt-mining area in this study is located in Tongbai County, Henan Province, China.This mining area has the second largest trona reserves in the world, while its glauber salt reserves reach 45 million tons.Since trona and glauber salt were put into production in 1990 with singleand double-well convection mining as the main producing method, five inrush points appeared in the town of Anpeng, Tongbai County, from June 2011 to May 2013.Among these five inrush points, four (Y1-Y4) were long-term (longer than 2 years) inrush points with stable discharge, while one (Y5) was a sudden inrush point (as shown in Figs. 1 and 2).Almost Since the groundwater inrush hazard involved a large geographic area and the inrush source was quite hard to distinguish due to the multilayer distribution of the different ore bodies and the complexity of the water inrush component, a targeted treatment programme to stop the water inrush and mitigate the groundwater pollution were needed urgently in research region.Therefore, the source and channel of the water inrush were taken as the research focus in this study.Furthermore, this research can provide a valuable reference for avoiding and assessing groundwater inrush hazards in similar rock-salt-mining areas, which is advantageous for both groundwater quality protection and public health.
2 Geological and hydrogeological setting

Geological conditions
The mining area is located in north-western Tongbai County.The landscape is characterised by hollows and ridges, with an elevation ranging from 140 to 200 m above sea level.The main development period of the research area consists of strata from the Hetaoyuan, Liaozhuang and Fenghuangzhen formations, from the oldest to the youngest.The Hetaoyuan Formation from the Palaeogene consists mainly of dolomite, muddy dolomite, mudstone, dolomitic mudstone, sandy conglomerate and siltstone.The third segment in the Hetaoyuan Formation is composed of thick mudstone interlayered with sandy conglomerate as well as thin layers of shale, muddy dolomite and glauber salt.The second segment is composed of mudstone interlayered with muddy dolomite and dolomite as well as small amount of trona.The first segment consists of mudstone, muddy dolomite, dolomite, shale, siltstone and trona.The upper part of the Liaozhuang Formation from the Palaeogene consists of mudstone interlayered with gypsum, while the lower part consists of alternating layers of mudstone and sandy conglomerate.The Fenghuangzhen Formation from the Neogene and Quaternary periods consists of alternating layers of sandy conglomerate and sandy clay (Shi et al., 2013).Detailed infor-mation on strata, lithology, aquifer and the position of different ore beds in the research area is shown in Fig. 3.
According to geologic references and field investigation, in the north-eastern mining area, a hidden east-west-oriented fault develops at the bottom of the first segment of the Hetaoyuan Formation and another four, hidden, south-northoriented faults develop at the bottom of the second segment of the Hetaoyuan Formation.These five faults are outside the scope of the trona mine, so they have little effect on the ore bed.A few small-scale hidden faults develop at the bottom of the third segment of the Hetaoyuan Formation, although within the scope of the glauber salt mine, they have little effect on the glauber salt ore bed which is distributed at the top of the first segment of the Hetaoyuan Formation.A hidden east-west-oriented fault is developed at the bottom of the Liaozhuang Formation in the range of the glauber salt mine, but it has little effect on the glauber salt mine because of its small scale.

Hydrogeological conditions
The groundwater in the mining area can be divided into pore water in the loose rock mass and bedrock fissure water according to the lithology and hydrogeological features.In the upper part of the Liaozhuang Formation, a mudstone interbedded with gypsum is considered a relatively weak permeable stratum, especially under conditions of high-pressure and -temperature water injection during the mining period.The shallow aquifer contains unconsolidated pore water above this weak permeable stratum, while the deep aquifer contains a bedrock fissure beneath this weak permeable stratum.
The flow direction of the shallow groundwater is controlled by the regional terrain.Taking the underground watershed as the boundary, the groundwater on the southern side of the watershed mainly flows from north-east to southwest with the Yanhong River as the drainage base, while the groundwater on the northern side of the watershed mainly flows from south to north with the Xia River as the drainage base.The deep groundwater is in relatively closed burial con-ditions, with slow velocity and nearly the same flowing direction as the shallow groundwater.The water inflow of a single well with poor water content is approximately 100 m 3 d −1 , but it can reach 1000-2000 m 3 d −1 if it has rich water content.The annual amplitude of the groundwater level is from 2 to 4 m, while the depth is stable at 2.3-4 m.Residents in Anpeng use groundwater as their drinking water, which comes from wells in the porous aquifer.
Gypsum mainly occurs at the top of the Liaozhuang Formation, glauber salt occurs in the third member of the Hetaoyuan Formation, and the trona occurs at the bottom of the second member of the Hetaoyuan Formation as well as on top of the first member of the Hetaoyuan Formation (Fig. 3).The surrounding rocks of every mineral layer, including mudstone, shale, sandy conglomerate, psammitic rock and dolomite, have sufficient thickness and good water resistance.Therefore, the effect of groundwater on the mineral deposit is minimal in the mining area.

Distribution and characteristics of the ore body
The three ore bodies overlap in plane distribution, as shown in Fig. 4. The vertical distribution of the ore bodies from deep to shallow is trona (buried depth: 1560.92-2929.53m), glauber salt (buried depth: 1003.66-1397.58m) and gypsum (buried depth: 134-338 m).The trona and glauber salt bodies are at least 250 m apart from each other vertically.
The trona has 11 horizontal layers, with an average thickness of 2.11 m.The chemical composition of trona is mainly NaHCO 3 (average of 77.06 %) and Na 2 CO 3 (average of 16.33 %) (Wang, 1987).The glauber salt has four layers, with an average thickness of 8.93 m.The dip angle of the ore bed layer is less than 10 • .The average mineral grade is 60.14 %.The main composition of the glauber salt is Na 2 SO 4 (> 90 %) with a small amount of NaCl.

Methods
Based on the field investigation results the source of the water inrush was determined by chemical analysis of the water inrush at different sites and times and by analysis of the physical and chemical reaction principles for the different brines combined with the PHREEQC simulation method.

Sampling and testing
The five groundwater inrush points (Y1∼ Y5) and some shallow groundwater points (resident wells: SY1 ∼ SY6) near the accident site were chosen as groundwater quality sampling points, as shown in Fig. 4. Water from each point was sampled on 9 March 2013.
Water samples were filtered using a 0.45 µm millipore filtration membrane in the field and then filled with a polyethylene bottle which had been soaked in acid and washed with deionised water.Filtered water samples were acidified until they reached pH < 2 by addition of ultrapure HNO 3 for the determination of cations; water samples for the determination of anions were not treated.
Elements tested in the laboratory included 26 cations (K + , Na + , Ca 2+ , Mg 2+ , Sr 2+ , etc.) and 5 anions ( The instrument used for the determination of cations was an inductively coupled plasma atomic emission spectrometer (Agilent ICP-OES 5100), with minimum detection limit at 0.0001 mg L 1− .The instrument used for the determination of anions was an ion chromatograph (ICS-1100), and the minimum detection limit was 0.001 mg L 1− .CO 2− 3 and HCO − 3 were tested according to the "Groundwater quality test method: Determination of carbonate and bicarbonate by hydroxide titration (DZ/T 0064.49-93)", with a minimum detection limit at 0.01 mg L 1− .
In addition, from March to April 2013, at the Y5 and Y3 sites, three water quality automatic recorders (Levelogger gold, Canada) were arranged for water inrush monitor-ing.Monitoring indicators were temperature, water level and electrical conductivity.The purpose of the monitoring was to fully understand the water quality of the inrush throughout the accident, especially in the process of repairing the well.

Analysis of the physical and chemical reaction principles in different brine mixing conditions
During the accident, the brine leakage of the trona (2000 m below the ground) or glauber salt (1000 m below the ground) might flow through the gypsum deposit (200-400 m below the ground), which is comprised primarily of CaSO 4 and cause physical and chemical reactions while it rushes out of the ground.Thus, the formation of the chemistry component in water inrush might be from glauber brine, trona brine or a mixture of the two flowing through the gypsum layer with accompanying physical and chemical reactions.To provide a basis for further analysis of the water inrush source, the physical solubility of the gypsum and the reaction were analysed when the glauber salt brine, the trona brine or a mixture of the two flowed through the gypsum deposits.

The physical solubility of gypsum (CaSO 4 )
Gypsum is slightly soluble; when in water, its acidity is apparent.Equation ( 1) provides the dissolution rate equation of gypsum in water: where R Gypsum is the dissolution rate of gypsum, k 1 is the rate constant, A g is the surface area of gypsum, V is the liquid volume in contact with the gypsum surface, IAP is the product of ion activity, and K is the ion solubility product.
IAP K Gypsum is affected by the temperature, as is the case for R Gypsum The solubility of gypsum in water reaches a maximum of 0.2097 g/100 g at 40 • .The solubility decreases when the temperature is below or above 40 (2) same ion is placed into an electrolyte-saturated solution.
Thus, the solubility of gypsum will be reduced when glauber salt brine flows through and dissolves the gypsum deposits; the gypsum will be even harder to dissolve in this situation.Thus, if the glauber salt brine flows through the gypsum deposits, the brine characteristic would apparently not change.

3.2.3
The reaction of trona brine or a mixture of trona and glauber salt brine with gypsum The HCO − 3 and CO 2− 3 contents in trona brine or in mixed brine are very high, as are the solution alkalinity and pH.If the reaction kinetics is not taken into account, the pH has little influence on the dissolution of gypsum (Yang, 2003;Xu and Li, 2011).The reaction occurs when the brine with high concentrations of HCO − 3 and CO 2− 3 flows through the gypsum deposits.The main chemical reactions are as follows: In Eq. ( 4), CaSO 4 is slightly soluble, while CaCO 3 is insoluble.The reaction easily occurs when an insoluble substance is produced by a slight soluble substance, and the ionic equation is as follows: The Gibbs free energy (G) is −22.7 kJ mol −1 under the standard state.When G is negative, the reaction, which is endothermic, occurs freely.The reaction is faster at higher temperatures.Equation ( 5) shows that G is 2102 kJ mol −1 under the standard state.When G is positive, the reaction will not freely occur.Thus, the reaction shown in Eq. (5) will not occur, but the chemical reaction will still proceed as shown in Eq. ( 4), when trona brine or mixed brine flow through the gypsum deposits.

The carbonate equilibrium effect during the reaction of different brine
The carbonate equilibrium in the trona brine or in the mixed brine is affected by pH.The carbonate in groundwater exists in three forms: free carbonic acid, bicarbonate and carbonic acid.
In the trona brine (pH > 10), the concentration of HCO − 3 is 5-20 times that of the CO 2− 3 concentration, and CO 2− 3 in the brine is dominant in this case.When the trona brine flows through the gypsum, CaSO 4 reacts with CO 2− 3 and CaCO 3 precipitates.If the concentration of CO 2− 3 in the brine decreases, a reversible reaction will take place and drive the equilibrium to the right.Thus, the reverse reaction will occur when the trona brine flows through the gypsum as follows: The circular reactions as shown in Eqs. ( 7) and (8) will occur when mixed brine flows through the gypsum because it has similar properties to the trona brine.Thus, taking the carbonate equilibrium effect into account, the concentrations of HCO − 3 and CO 2− 3 will decrease, while SO 2− 4 increases after CaCO 3 precipitates.

Simulation of groundwater inrush source
For further quantitative analysis of the water inrush source and component, the international hydrological and geochemical simulation software PHREEQC was used to simulate the water-rock interaction.The PHREEQC software was developed by the US Geological Survey and is able to calculate geochemical action within a temperature range of 0∼ 300 • (Wei, 2010).
Based on the deduction that the main water inrush source around Anpeng was trona brine leakage, the simulation method PHREEQC was used and combined with the possible channel of water inrush to establish a conceptual model.Then, the hydrogeochemical simulation of the water-rock interaction was conducted.Subsequently, the mixed ratio of groundwater inrush and shallow groundwater around Anpeng were quantified to better verify the source of the water inrush.

Conceptual model
Around Anpeng, the trona brine leakage flowed through the specified mineral assemblages and mixed with shallow groundwater in different proportions.

Initial data input
The parameters of the trona brine were taken from the enterprise's production testing data.The parameters of the shallow groundwater were taken from the same aquifer but outside the study area and can basically represent groundwater background values.The specific parameters are shown in Table 1.

Setting of stratum and mineral
The formations from the bottom to the top during the process of the brine leakage flowing into the shallow groundwater and then flowing out of the ground were as follows: the third member of the Hetaoyuan Formation from the Palaeogene, the Liaozhuang Formation and the Fenghuang Formation from the Neogene and Quaternary periods.To simplify the mining area according to the thickness of the rock stratum and the proportion of mineral composition, it can be assumed that the layer through which the trona brine flowed contains , potash mica is aluminium silicate as K, Al, Mg, Fe and Li.

Results and discussion
On 9 March 2013, in Anpeng, water samples from five groundwater inrush points and six surrounding water quality monitoring points (resident well) were tested.The results of water chemical composition are shown in Table 2, and the distribution of the sampling points is shown in Fig. 4.
According to the water quality analysis, the brine inrush had a relatively high salinity, with some water inrush samples containing SO 4 -Na and some containing HCO 3 -Na.The crystals mainly consisted of NaSO 4 , Na 2 CO 3 and NaHCO 3 .The composition of the water inrush and the crystals was the same as that of the high-concentrated ions in the trona brine (Na 2 CO 3 , NaHCO 3 , etc.) and in the glauber salt brine (Na 2 SO 4 ).

The source of the water inrush
An automatic water quality recorder was set up at the Y5 inrush point on 4 March 2013.The monitoring lasted from 5 March to 20 March 2013.Thus, the relationship between the inrush points and the S02 well can be assessed according to the correlation of the changes between temperature/electrical conductivity and the concentration of brine during the S02 production well repair period (5 March to 14 March 2013).
The production of glauber ceased during the investigation (2 March to 15 March 2013), so it could be determined how glauber mining affects the water inrush hazard based on a dynamic water quality situation.

The source of water inrush at the Y5 point
After successful repair of the S02 well, the conductivity and temperature of the water inrush decreased significantly.The CO 2− 3 concentration remained at 0 and the HCO − 3 concentra-tion decreased to 500 meq L −1 , while the SO 2− 4 concentration increased to 600 meq L −1 .Subsequently, the concentrations of these three ions were in a state of dynamic balance.The analysis shows that the source of the water inrush at the Y5 point is closely related to the S02 trona well.
In order to ensure whether the glauber brine exists at this point as part of an inrush source, further analysis was performed.The depth of the trona production well rupture was 234 m, and the gypsum deposit was developed at the depth of 134-338 m, so while the leakage of the trona brine flowed through the gypsum deposit, reactions would occur as shown in Eqs. ( 7) and ( 8).
According to the ion milliequivalent concentrations (Ca 2+ : 0.61; CO 2− 3 : 905.3;HCO − 3 : 1332.94;Cl − : 107.43; and SO 2− 4 : 267.89 meq L −1 ) at the Y5 point, the concentration of Ca 2+ was negligible compared to the other main ions.Only the reaction between CO 2− 3 and CaSO 4 had to be taken into account because of the large number of CO 2− 3 , high velocity, the short contact time with gypsum and the high temperature.The reaction of CO 2− 3 and CaSO 4 would take place at a ratio of 1 : 1 according to Eq. ( 7), and three types of water inrush sources could be assumed under this precondition as follows.
The water inrush source was only from the trona brine.The CO 2− 3 and CaSO 4 in the brine reacted at a ratio of 1 : 1, and the SO 2− when the contribution ratio of the trona brine was equal to 1, the contribution ratio of the glauber brine was equal to 1.6 × 10 −5 , small enough that it can be ignored.Thus, it could be confirmed that the water inrush source at Y5 was exclusively the leakage of trona brine from the broken S02 well.

4.1.2
The sources of water inrush at the Y4, Y3, Y2 and Y1 points The water inrush quantity and the dynamic variation of the concentration of SO 2− 4 and HCO − 3 at points Y1-Y4 were not obvious when the S02 well was under repair and all the glauber wells were shut down (from 2 to 15 March).This result shows that the sources of these water inrush points were not due to the underground mining activities of the glauber brine or the rupture of the S02 well but rather to the brine leakage from other trona wells.

Components and mixed proportions of the water inrush
The PHREEQC simulation conditions were assumed to be as follows: (1) the trona brine did not mix with shallow groundwater after flowing through the mineral layer or (2) the trona brine mixed with shallow groundwater in a ratio of 1 : 2, 1 : 10, 1 : 100, 1 : 200, 1 : 500, 1 : 1000 and 1 : 5000 after flowing through the mineral layer.The simulation results are shown in Table 3. Table 3 shows that when the trona brine flowed through the bedrock of Hetaoyuan, Liaozhuang, Fenghuangzhen Formation and shallow groundwater successively, the concentra- tions of Na + , Cl − and SO 2− 4 decreased while the HCO − 3 concentration increased with an increasing proportion of shallow groundwater.The Ca 2+ concentration decreased at first and then increased.
The ion concentrations at Y5, except for SO 2− 4 , were similar to the ion concentrations in the trona brine.However, at the same time, the HCO − 3 concentration was nearly 0 meq L −1 .When the trona brine flowed through the layer, it reacted rapidly and poured out of the ground directly because of the high velocity of the water inrush at Y5.Meanwhile, the trona brine was not continuously provided in the simulation.Thus, the concentration of HCO − 3 would be near to the concentration of trona brine in reality.Therefore, the trona brine must have a rapid inrush and almost not mix with the shallow groundwater.
The PHREEQC simulation results show that (1) the water inrush source of Y5 was the trona brine almost all from the ruptured S02 well, (2) the water inrush source of Y3 was a mixture of trona brine and groundwater in a ratio of 1 : 10-1 : 100, and (3) the water inrush sources of Y4, Y2 and Y1 were a mixture of trona brine and groundwater with ratio of 1 : 200.

Causes of the brine leakage
Trona is produced by either a single well or a double/multiple well convection mining method that is water-soluble (Lin, 1987).The main mining unit consists of a salt cavity and production well.Thus, the instability of the salt cavity and the rupture of the production well are the main possible causes of brine leakage.
(1) Analysis of salt cavity stability The possibility of salt cavity collapse: Trona is distributed at the bottom of the second member of the Hetaoyuan Formation and in the upper part of the first member of the Hetaoyuan Formation, with developed dolomite strata at the roof and floor.The thick and hard surrounding rock structure determined that the cavity is produced by hydrofracture but it is hard to fill with large-scale fractured channels and can remain intact and stable.
The development of a roof fracture: When a mineral is under exploitation, the surrounding rock in the cavity is under pressure from the inner brine.This pressure is equal to the sum of the water injection pressure and the water column pressure in the production well.The water injection pressure of the trona production well is approximately 10-20 MPa, while the 1560.92-2929.53m (mineral buried depth) water column pressure is approximately 15.3-28.71MPa.Thus, the greatest water pressure on the surrounding rock in the cavity is 48.71MPa.The main lithology of the surrounding rock is dolomite (500 m in thickness and 142.66 MPa in compressive strength), which is nearly 3 times that of the greatest possible water pressure.Therefore, large-scale fractures in the surrounding rock of the trona mineral would be difficult to develop under the effect of sustained water pressure.
(2) Analysis of production well rupture The phenomenon of brine leakage caused by the S02 well rupture in Anpeng indicates that production well damage is a major cause of brine leakage.The depth of the S02 well rupture is 234 m underground, i.e. in the gypsum deposit, which is strongly hygroscopic.The pressure caused by the water swelling is approximately 0.15 MPa (Li and Zhou, 1996), which may damage the production well and induce brine leakage.The high concentration of SO 2− 4 (> 250 mg L −1 ) generated by the reaction of brine leakage and gypsum can also corrode the production well and lead to groundwater inrush.

Analysis of water-conducting channel
According to our analysis, the most probable reason for brine leakage in trona is the production well rupture.The leaking brine flows along the water-conducting channel into the shallow aquifer and even pour out of the ground.However, the geological structure in the mining area shows no waterconducting fault development.Thus, the water-conducting channel, that the brine leakage flows along, is probably the structure fissure zone or the abandoned gypsum exploitation well.
Structural fissure is the main type of fissure that occurs in groundwater inrush hazards when using the solution-mining method.The structural fissure is determined by the maximum horizontal principal stress, which is controlled by the tectonic stress field in the mining area.The connection direction of the S02 well and the other water inrush points is NW-SE, the same as that of the structural fissure zone development direction.This indicates that the main water-conducting channel in Anpeng is controlled by the structural fissure zone.
The inrush points in Anpeng are all at the abandoned gypsum exploitation wells, which were not closed properly.Thus, high-pressure cavity water or brine leakage can flow along the structural fissure zone, finally connect with these wells and then pour out of the ground through boreholes.Therefore, the abandoned gypsum exploitation wells are the main channels through which the shallow polluted groundwater flowed out of the ground, as shown in Fig. 5.

Conclusions
This study aimed to investigate the source and channel of the water inrush in a multilayer rock-salt-mining area.To achieve the set objectives, an analysis of geological and hydrogeological conditions, an analysis of physical and chemical reaction principles of different brine, the PHREEQC simulation method, and an analysis of geological and artificial causes of the conducting channel where brine leakage flowed from the damage depth out to the ground were combined.
Long-term solution mining with high-pressure andtemperature water not only dissolves minerals, but also may cause rupture of strata and damage of the production well, which usually results in brine leakage or groundwater inrush.Geological and hydrogeological conditions are the basis which determines the total risk of the groundwater inrush hazard.Physical and chemical reaction principle analysis of different brine and hydrogeochemical simulation of water-rock interaction in different assumed conditions using the PHREEQC simulation method can determine the exact source of the brine leakage as well as identify the mixed proportion of water inrush while the brine flows through the mineral layer.Other than geological reasons, mining techniques such as pressure control of injection water and groundwater quality monitoring of exploitation wells may also determine the risk of a groundwater inrush hazard in a multilayer rock-salt-mining area.

Figure 2 .
Figure2.The sudden groundwater inrush point (Y5).The hightemperature inrush groundwater was being pumped after the ground was broken.

Figure 4 .
Figure 4. Sketched map of hydrogeological conditions and the distribution of groundwater inrush points in the mining area.

Figure 5 .
Figure 5. Schematic diagram of the source and channel of the groundwater inrush hazard in the multilayered rock-salt-mining area in Tongbai County.
• .The content of SO 2− 4 and Ca 2+ obtained by physical dissolution is very low.

Table 1 .
Initial data on trona brine and background value of groundwater for the PHREEQC simulation.

Table 2 .
Chemical composition of groundwater from the inrush hazard points and surrounding resident wells.

Table 3 .
Simulation results for a mixed proportion of trona brine inrush using the PHREEQC method.