地球化学水害防治.pdf

JOURNAL OF COAL SCIENCE 2 analysis of controlling factor based on the chemical composition of ground water 3 pure sample chosen and 4 discrimination model establishment After the processes two functions and a diagram have been established for three aquifers the Quaternary Coal bearing and Taiyuan Fm from the Renlou coal mine in northern Anhui Province China The method can be applied in almost all of the coal mines and can be used for evaluating the contribu tion ratios if the water is collected from a mixing source Keywords hydro geochemistry source discrimination multivariate statistical groundwater Received 5 March 2012 Supported by the National Natural Science Foundation of China 41173016 Tel 86 557 2871038 E mail sunlinh Introduction There are two main types of energy in this world coal and oil Oil is the major source of energy in USA whereas coal accounted for 60 percent of China s pri mary energy during 2010 and this proportion will maintain more than 50 till 2050 Therefore the safety of coal mine is directly related to the develop ment of China s economy In order to meet the demand of energy consume more and more attention have been paid towards the deep environments with up to 1000m depth as the drying up of coals in shallow areas In this background controlling of water inrush in coal mines will become more serious as they can brought to us so many acci dences with great losses during mining activities Take an instance more than seventy water inrush acci dences had occurred from 2007 to 2009 and more than 300 people died As one of the major methods for water disaster con trolling hydro geochemistry had attracted a series of studies related to water source discrimination and a large number of ideas and methods have been pub lished in recent years Huang and Chen 2011 Wei et al 2011 Chen et al 2009 Chen et al 2008a Gui and Chen 2007 However these methods were con structed by very complex methods and thus hard to be applied or they were established in a special environ ment and can not be migrated to another place with the changing of condition In this paper we report a case study in Renlou coal mine northern Anhui Province China Fig 1 for con struction of the water source discrimination model by using multivariate statistical analysis of hy dro geochemistry major ions which is easy to be handled and can be applied in almost all of the coal mines The possibilities for this study include 1 the mineral as well as chemical compositions are different between the rocks of aquifers and 2 previous studies revealed that the chemical composition of groundwa ter is mainly controlled by the rocks they flowed over 1 Background data collection and methods 1 1 Hydrological and geological condition Previous studies revealed that the deep aquifer sys tems in Renlou coal mine can be subdivided into four parts according to their different hydro and geological conditions Gui and Chen 2007 the Quaternary aq uifer the Coal bearing aquifer the Taiyuan Fm aquifer and the Ordovician limestone aquifer The Quaternary aquifer is composed of mudstone sandstone and con glomerate the Coal bearing aquifer is mainly com posed of sandstone and coal seams whereas the Tai yuan Fm aquifer and the Ordovician limestone aquifer are mainly composed of limestones Chen et al 2008b had studied the hydrological condition of these aquifers by using isotopic method and suggested that the hydrodynamic conditions of these four aquifers are different Quaternary aquifer is open because it can be recharged by the Taiyuan Fm Aquifer the Coal bearing aquifer is relatively closed whereas the lower limestone aquifers including Tai yuan Fm and the Ordovician limestone aquifer are also closed systems with rapid flow Fig 1 The location of study area 1 2 Data description Data in this study were collected from Renlou coal mine and the analytical processes were taken place during the period between 2000 and 2010 A total of forty five samples have been collected fifteen for each aquifer including the Quaternary aquifer S the Coal bearing aquifer M the Taiyuan Fm aquifer T The data related to the Ordovician limestone aquifer have not been collected as they were really rare be cause the groundwater sample from this aquifer is hard to be obtained The average concentrations of the major ions from the three aquifers are listed in Table 1 and shown in Fig 2 As can be seen from the table and figure the Coal bearing aquifer possesses the highest concentra tion of K Na HCO3 and CO3 whereas the samples from the Taiyuan Fm aquifer are enriched in Ca Cl and SO4 relative to other aquifers These differences between aquifers provide probability for discriminate them by using their chemical compositions Table 1 Average concentration mg L of major ions in dif ferent aquifers K Na Mg2 Ca2 Cl SO42 HCO3 CO32 S 15 320 78 2 174 628 345 231 7 68 M 15 731 22 3 17 9 566 72 2 915 41 0 T 15 385 78 1 297 945 359 244 3 44 Fig 2 Average concentrations of major ions in different aq uifers 1 3 The processes The technological process for establishment of wa ter source discrimination model in this study is shown in Fig 3 Firstly all of the data including analytical results of groundwater from different aquifers other hydro and geological data were collected from the coal mine Secondly the hydrological and geological data except for analytical results of groundwater were analyzed for understanding the relationship between aquifer e g recharge run off and excretion and the rock mineral composition respectively Thirdly the analytical results of groundwater were processed by using multivariate statistical analysis correlation and R mode cluster for determine the relationship be tween major ions and then principle component analysis were applied for evaluating the controlling factor of the chemical compositions of the groundwa ter in combination with the geological interpretation Q mode cluster analysis was then processed for chosen the relative pure samples That s because some of the groundwater collected from one aquifer might have been recharged by another aquifer Then the pure samples were analyzed by discriminant analysis for construction the function and diagram for water source discrimination Fig 3 Technological process for establishment of water source discrimination model SUN Linhua et al Establishment of water source discrimination model in coal mine 3 2 Controlling factor of the groundwater chemistry To reveal the controlling factor of groundwater chemistry correlation analysis was firstly applied for evaluating the relationship between major ions be cause a close relationship of elements means that they may originated from similar source or controlled by similar factors e g adsorbed by similar particles the method had long been used for evaluating the source of contaminants such as heavy metal and any other elements in soils dusts and groundwater in combina tion with R mode cluster analysis Yalcin et al 2010 Zhou et al 2008 Cloutier et al 2008 Well correlation are observed between K Na HCO3 and CO3 Mg Ca Cl and SO4 Table 2 and they suggest the following information 1 K Na HCO3 and CO3 must be originated from similar sources or conditions because they are also enriched in the Coal bearing aquifer Table 1 and Fig 2 Consid ering about the composition of the aquifer K and Na must be originated from the dissolution of K feldspar and plagioclase in the sandstones whereas HCO3 and CO3 must be originated from the coals 2 Mg Ca Cl and SO4 must be originated from the dissolution of calcite dolomite chloride and sulfate With reference to the R mode cluster analytical re sult Fig 4 Mg Ca CO3 and SO4 form the first clus ter whereas K Na and HCO3 are well correlated with Cl as a separate This result imply that K Na and HCO3 are originated from dissolution of sand stones and coals respectively whereas the CO3 in the groundwater is from the dissolution of carbonates in combination with Mg Ca and also Cl and SO4 are produced by dissolution of chloride and sulfate in ma rine facies with carbonates dominant Table 2 Correlation analytical results K Na Mg2 Ca2 Cl SO42 HCO3 Mg2 0 47 Ca2 0 49 0 68 Cl 0 02 0 51 0 75 SO42 0 42 0 80 0 76 0 42 HCO3 0 77 0 44 0 59 0 42 0 45 CO32 0 54 0 48 0 53 0 37 0 49 0 61 Principle component analysis was applied to the analytical data and varimax rotation was applied si multaneously for quantitatively evaluating the cluster behavior Two factors had been extracted with ei gen value higher than 1 and a total explanation of they is 77 Table 3 and Fig 4 The first pc is responsible for 40 7 of the total variance and is represented by Mg Ca Cl and SO4 after varimax rotation VF1 whereas PC2 explains 36 3 of the total variance and is participated by K Na HCO3 and CO3 VF2 The result is consis tent with correlation analysis Table 2 The pc score plot of the samples is shown in Fig 5 As can be seen from the figure samples from the Coal bearing aquifer are characterized by high PC2 scores and isolated from the Quaternary and Taiyuan Fm aquifer samples with low PC2 scores except for one sample However some of samples from the Quaternary aquifer are in tersected with the samples from the Taiyuan Fm aqui fer although the former and latter tend to have low and high PC1 scores respectively This result indicates that these two aquifers must have been mixed with or recharged by each other and it is demonstrated by the truth that a serious water inrush accident had occurred in the coal mine in 1996 and the Quaternary aquifer was recharged by the Taiyuan Fm aquifer Fig 4 R mode cluster analysis of major ions Table 3 Results of principle component analysis Before rotation After rota tion ComponentTotal Var Cum Total Var Cum 1 4 2 59 7 59 7 2 9 40 7 40 7 2 1 2 17 3 77 0 2 5 36 3 77 0 PC1PC2 VF1 VF2 K Na 0 68 0 68 K Na 0 05 0 96 Mg2 0 82 0 19 Mg2 0 74 0 40 Ca2 0 90 0 26 Ca2 0 84 0 41 Cl 0 64 0 63 Cl 0 89 0 04 SO42 0 82 0 19 SO42 0 74 0 41 HCO3 0 79 0 39 HCO3 0 32 0 82 CO32 0 74 0 24 CO32 0 39 0 67 3 Establishment of source discrimination model 3 1 Sample preparing Because of the mixing between aquifers relative pure samples without assimilation from other aqui fers must be chosen firstly before model construction Considering about the rock compositions of the aqui fers and the score plot of factors Fig 5 the following method have been applied for sample chosen Firstly two samples from the Coal bearing aqui fer with PC1 score higher than one have been deleted because the Coal bearing aquifer is mainly composed of sandstones the dissolution of plagioclase can not release too much Ca and Mg into the water Secondly four samples from the Taiyuan Fm aquifer with PC1 score lower than 0 have been deleted because the rocks of Taiyuan Fm aquifer is mainly composed of carbonates with very small amount of clastic compo nents Lastly six samples from the Quaternary aquifer with PC1 score higher than 0 are also deleted because there are limited carbonate rocks occurred in the aqui fer and thus these samples must be assimilated by the groundwater from the Taiyuan Fm aquifer Fig 5 Component and score plot of the first two PCs 3 2 Model construction Discriminant analysis is useful for building a predictive model of group membership based on ob served characteristics of each case The procedure generates a discriminant function or a set of functions based on linear combinations of the predictor variables that provide the best discrimination between the groups The functions are generated from a sample of cases for which group membership is known the functions can then be applied to new cases with meas urements for the predictor variables but unknown group membership In this study samples from the Quaternary Coal bearing and Taiyuan Fm aquifers are assigned as 1 2 and 3 respectively And then the discriminant analysis has been processed by using SPSS and the data after preparing After processing two functions and one discriminant diagram Fig 6 have been generated Function 1 0 009 Na K 0 035 Mg 0 031 Ca 0 007 Cl 0 006 SO4 0 008 HCO3 0 003 CO3 0 024 1 Function 2 0 008 Na K 0 008 Mg 0 002 Ca 0 008 Cl 0 001 SO4 0 009 HCO3 0 01 CO3 6 009 2 If a sample we do not know its source aquifer then the result of Function 1 and 2 can be calculated fol lowing the above functions and then the source aqui fer can be identified by plotting the result in Fig 6 Another important thing can be obtained from dis criminant analysis that it can produce the average functional concentrations of the chemical composi tions of different aquifers the S M and T in Fig 6 which can be used for evaluating the contribution ra tios of different aquifers by using simple calculation if the water is collected from a mixing source In this study the centroids for the Quaternary Coal bearing and Taiyuan Fm aquifers are 0 66 3 16 4 81 0 99 and 5 14 1 42 respectively Fig 6 Diagram of discriminant analysis 4 Conclusions Multivariate statistical analysis of chemical compo sitions major ions of groundwater in deep seated en vironment in Renlou coal mine northern Anhui Prov ince China were processed for constructing the source 4 Journal of Coal Science 2 The processes for constructing the water source discrimination model can be subdivided into three steps Controlling factor analysis pure sample cho sen and model construction 3 Multivariate statistical analysis suggests that the groundwater chemistry in Renlou coal mine are con trolled by dissolution of plagioclase K feldspar and carbonates with different degrees 4 Two functions and one diagram have been es tablished for water source discrimination in Renlou coal mine which can also be applied for evaluating the contribution ratios of sources if the water is col lected from a mixing source References Chen H J Li X B Liu A H and Peng S Q 2009 Identifying of mine water inrush sources by fisher discriminant analysis method Journal or Central South University Science and technolo