Modelling of the relationship between trace elements and three species of sulfur in coal

环境科学2E980212ENVIRONMENTAL SCIENCES科技期刊环境科学1998年6月第10卷第2期Modelling of the relationship between trace elementsand three species of sulfur in coal"eLu Xiaohua. Zeng hancaiNational Key Laboratory of Coal Combustion, Huazhong Universityof Science and Technology, Wuhan 430074, ChinaWei Lu-xianCenter of microanalysis wulan University of automobile Industry Euhan, chinaAbstract--Based on the determination of several trace elements incoal particulate of different specific gravity, three-variablesregression equations relate the contents of trace elements Be, CdCo, Pb, Cu, Ni, As and three species of sulfur in coal were wellestablished. For elements Cd. Co Pb Cu and ni. the regressioequations were successfully used for prediction of these traceelements in individual part of coal with different specificgravity. Factor analysis was also used to analyze the data setsThe results showed that a three factor model can interpret the datasets reasonably. Trace elements Cd, Co, Pb, Cu, Ni, inorganicsulfide and total sulfur in coal are high correlated with the firstfactor. Trace elements Be, sulfate and organic sulfide are highcorrelated with the second factor and trace elements cr and as arehigh correlated with the third factor The factors can beinterpreted by the chemical properties of these elementsKeywords: trace elements; coal; regression analysis; factor analysis1 IntroductionCoal combustion produces not only large amount of gaseouspollutants but also large quantity of particulate matter. Theparticulate matter emitted through the stack is the most importantsource of trace heavy metals in atmosphere because of theenvironmental impact of heavy metals, there is a growing interestin the study of trace elements in coal(Solari, 1989; Evans, 1990;Skeen, 1990). From mine to combustion, various trace elements incoal are released into environment and most of them are potentiallytoxic to plants and animals under certain conditions andconcentrationsThe concentration and distribution of trace elements in coalare important for prediction of their behavior on release intoatmosphere when coal is burnt. Different trace elements behavedifferently in coal combustion. Generally trace elements associated中国煤化工with the organic matter and inorganic sulfides volatilize morereadily in coal combustion and are enriched in the submicrometerCNMHGfly ash by volatilization and condensation mechanisms (Linak, 1986Flagan. 198fle∥/ F yqikan_ htm抽取200 before/kjgk(2010)/jee/980212htm(第1/9页)200912-312256:31环境科学2E980212Depending on their chemical properties and the physicochemicalprocesses which occur during the coal formation, trace elements canexist in the mineral matter, in the organic matter or in both(Nichols, 1968; Raask, 1985). There have been several studies onthe concentration and distribution of trace elements in coalGluskoter et al.(Gluskoter, 1974) proposed an empirical method todetermine the affinity of trace elements with the organic coalmatter. Regression analysis and factor analysis are multivariablestatistical methods commonly used in studying the correlationrelationship between variables. Solari et al. (Solari, 1989)developed a multivariable regression model for the partialconcentration of trace elements in the organic and inorganicmatter Using Solaris model, Pires and Teixeira(Pires, 1992)studied the concentration and association of these elements in theorganic, sulfide and non sulfide fractions of Leao coal Factoranalysis was used to study the distribution of trace elements inseven Chinese coals and the mathematical interpretation was provedby the experimental results (lu, 1995). The results of thesestatistical method showed the relationship between trace elementsand the organic and inorganic substances in coal. Up to now,noresearch work on the relationship between the contents of traceelements and three species of sulfur in coal was reported. In orderto study the relationship between trace elements and differentspecies of sulfur and to assess the environmental effects of traceelements produced from coal combustion, in this work, based on thedetermination of some trace elements in particulate of coal withdifferent specific gravity, regression analysis and factor analysisare used for modelling the relationship between the contents oftrace elements and three species of sulfur in coal particular ofdifferent specific gravity in this paper. The results of this paperare of significance for the investigation of decreasing theheavymetal pollution of atmosphere caused by coal combustion2 Experimental2.1 Classification of coal into different specific gravityA coal sample of Qingshan bituminous was crushed to an averageparticle size of 50-60 F m. Six solvents of different specificgravity(Table 1)were used for sequential classification of coaland seven samples of different specific gravity were obtained. Ineach step, the coal powder on the upper part of the solvent wascollected and the bottom part was used in the next stepTable 1 Preparation of solutions with different specific gravity6CompositionCohsOHHoOCH,CI,CCICCl4+Ch3IChaIc gravity071001.30160中国煤化工CNMHG2.2 Determination of trace elements and three species of sulfur inThe seven samples of different specific gravity were digestedfle∥/ F yqikan_ htm抽取2000 efore/kjgk(2010)/jee/980212htm(第2/9页)200912-312256:31环境科学2E980212and then the trace elements be. cd co pb. cu ni cr and as weredetermined by ICP-aeS and gFAAs. The sulfate(sOn) and total sulfur(TS)were determined by BasOa gravimetry and inorganic sulfur (Is)was determined by k2 Cr2O7 oxidation-reduction titration(Bai1990)The organic sulfide(os) was calculated by subtraction2.3 ComputationThe computer programs of regression analysis and factoranalysis in the advanced statistical package SPSS/PC+(Norusis,1986)were used to analyze the data sets3 Results and discussionsThe concentrations of trace elements and three species ofsulfur in the coal samples of different specific gravity are listedin Table 2Table 2 Concentrations of trace elements and three speciesof sulfur in coal with different specific gravity56ccific gravity<0.760.76-1.01.0-1.31.3-1.61.6-1.921.92-2.282.280.6750.7210.8730.882l.873SOa, 90.0460.0540.0630.0700.0730.0750.1910.5960.6571.7940.4980.4190.2140.1870.1530,003Ag/K0.13<0.01<0.01<0.0l<0.01<0,01<0.012.502.305.7010.26.617.48Pb, pg/40.349.345.954,4111.4187.9Cu, ug/g20.521.221.833.810.510.118.312.415.928.419.217.52.620.118.624.729.3As, pg/g24.712.910.127.912.021.939.83.1 Regression modelsSolari proposed a regression model to study the distributionorganic components and k kinds of inorganic components in coalthen the concentration of a trace element c in coal isC W.+>Ciwwhere Coi and cik are the contents of the trace element in organic中国煤化工CNMHGcomponent j and inorganic component k in coal respectively. Woi andwik are the percentages of organic component j and inorganicfle∥/ F yqikan_ htm抽取200 before/kjgk(2010)/jee/980212htm(第3/9页)200912-312256:31环境科学2E980212component k respectively. If the woi and wik of all kinds oforganIcand inorganic components are known, the partial concentration ofthe trace element in all components can be obtained by regressionanalysis. In fact it is impossible to get all the Ws, in this paperwe assume that the trace elements exist in sulfates, inorganicsulfides, organic sulfides and other unknown components. So themodel followed can be used to fit the data of trace elements threespecies of sulfur and other unknown componentsC=CW。+CWp+CW,+∑cnwere C. C and C are the concentrations of trace element inorganic sulfides, inorganic sulfides and sulfates in coalrespectively. Wo, Wp and Ws are percentages of organic sulfides,inorganic sulfides and sulfates in coal respectively. Cin and winare the concentrations of trace elements and percentages of otherunknown components in coal. Based on model(2), a linear regressionmodel with three independent variables was used to fit the datasets in Table 2 for every trace elementY =bo+ b1X1+b2X2+ b3x3(3)where X, X, and X3 are percentages of sulfates, inorganic sulfidesand organic sulfides in coal samples of different specific gravity,Y is the concentration(f g/g)of trace element in correspondingsamples. In this paper the samples No. 1, 2, 3, 4, and 6 are used toestablish the regression models and the sample no 5 is used fororediction of these equations. For the eight trace elements thelinear regression equations and correlation coefficients are shownin Table 3Table 3 Regression models between trace elements and threespecies of sulfurRegression equationsCorrelation coefficientsY=0.145-2.76X1+0.0389X2+0.122X30.7760Y=-0.952+11.27X1+3.23X2+3.23Xy0.9966Y=649-1243X1+2.38x2+1.03X30.9671Y=-5147+7913X1+57.50X2+67.93Xy0.9900Y=-7.99+248.8X1+12.84X2+23.25x30.9970Y=-37.98+58.6X1+5.33X2+41.69X中国煤化工Y=81.24-716.1X:-1.49X2-58.25X3CNMHGY=-75.11+925.3X1+15.19X2+98.94X0.3934The regression equations in table 3 show that for tracefle∥/ F yqikan_ htm抽取200 before/kjgk(2010/jee/980212htm(第4/9页)200912-312256:31环境科学2E980212lements Cd, Co, Pb, Cu and Ni, there are good linear relationshipsbetween the concentrations of trace elements and the percentages ofsulfates, inorganic sulfides and organic sulfides. The largecorrelation coefficients of these equations show that there arehigh correlation between these elements and three species of sulfurin coal. trace elements be and Cr also show certain correlationwith three species of sulfur in coal, while trace element As showslow correlation with three species of sulfur in coal. The resultsof regression analysis can be interpreted by the association ofthese trace elements with three species of sulfur. Many researchwork showed that trace elements Cd. Pb. Co Cu and ni exist in coalmainly with inorganic sulfides and trace element Be generallyassociates with organic sulfides or some times with sulfates( Pires,1992;Lu,1995Using these regression equations, the concentrations of traceelements in sample No 5 is predicted and shown in Table 4. It showsthat for most of these trace elements the results predicted bythese equations are in good agreement with those determinedTable 4 Prediction of sample no. 5 by the regression modelsCPredicted2,487.3054.422.214.919.117.6Determined<0.011.812.418.63.2 Factor modelsIn this work the original data set was used in factoranalysis. table 5 is the correlation matrix of twelve variables. Infactor analysis if the correlation coefficients between variableare small, it is unlikely that the variables are appropriate forthe factor model. Table 5 shows that most correlation coefficientsare large. All variables have large correlation with others. So thmatrix is appropriate for factor analysisTable 5 Correlation matrix of twelve variables中国煤化工CNMHGfle∥/ F yqikan_ htm抽取200 before/kjgk(2010)/jee/980212htm(第5/9页)200912-312256:31环境科学2E980212ISOSCCuAsT1.000SO40.6331.0000.9880.7381.0000.781-0.965-0.8671.000Be-0.331-0.745-0.4110.6791.000Cd0.9970.6390.986-0.785-0.3321.0000,9730.5960,953-0.747-0.4040.9811.000Pb0.9820.7360.988-0.854-0.4910.9760.9571,000Cu0.9900.7210.993-0.840-0.4390.9890.9670.9951.000Cr0.4270.5390.470-0.505-0.1450.4870.4640,4000.4711.000N0.8180.5010.831-0.705-0.2270.8260.8240.7780.7790.3131.000As0.166-0,0290.1410.0130.4210.2270.2040.0440.1410.8120.1371.000The principal component analysis was used to estimate theinitial factors The initial statistics from principal componentle correlation matrix are listed in Table 6. It showsthat more than 95% of the total variance is attributable to thefirst three factors The remaining nine factors together accountfor only less than 5 of the variance. So a model with threefactors may be adequate to represent the data setTable 6 initial statisticsFactorEigenvaluePercent of varianceCumulate percent8.3913069.91.8905415.81.186259,90.318002.798.20,213901.80.000000.0120.0From the estimation of factor number of 3. the initial factormatrix and the varimax rotated factor matrix were obtained andshown in Table 7. In the rotated factor matrix the first factor Flshows high correlation with variables ts. IS. Cd. Co. Pb. Cu and niin coal. It can be interpreted by the similar chemical propertiesof these elements, i. e, these trace elements have a strongtendency to associate with inorganic sulfides in coal. It iscoincide with the results of regression analysis. The second factorF2 show high correlation with variables SO4, Os, and Be. It means中国煤化工that Be is generally associated with organic substances andCNMHGsulfates in coalfle∥/ F yqikan_ htm抽取200 before/kjgk(2010)/jee/980212htm(第6/9页)200912-312256:31环境科学2-E980212Table 7 Factor matrix of factor analysisInitial factor matrixRotated factor matrixFFF0.96290.06800.23770.96070,21770.13480.79930,49550.43960.85200.192300.01540.12420.92430.33860.153700.22150.27980.64170.72050.14390.5076065540.46270.17100.91020.20890,96990.11770.19550.95020.22400.95240.08390.19670.93300.23370.97880.08620.13330.91250.38300.06200.98340.01010.11710.91610.34440.15180.5400,61940.56760.24680.30830.91770.821400960.87140.10340.07070.16220.95840,22450.10700.28790.9491According to the three factor model, factor score werecalculated and shown in Table 8. The factor score on FI and F2 wereused to classify the twelve variable in Ffa coordinate systemclassification of the variables was shown in fig. 1. It shows thatvariables Ts, Is, Pb, Cu, Co, Cd, Ni locate closely in FrFcoordinate system It also means the similarities of theseTable8 Factor score coefficient matrixFactor 1Factor 2Factor 3TS0.19930,11220.04800.12840.41750.11320.15390.02740.02200.0278-0.27990.0440Be0.16110.49360.18500.10070.18230.09260.02360.00420.07120.15020.02180.02150.21360.16310.08130,03540.1312YHCNMHGfle∥/ F yqikan_ htm抽取200 before/kjgk(2010/jee/980212htm(第7/9页)200912-312256:31环境科学2E9802124Pb0.25Coc-NiAsOSFig. 1 The classification of the variables4 ConclusionsThe factor model and regression model of trace elements inQingshan bituminous of different specific gravity show that traceelements Cd, Co, Cu, Pb and Ni associate with inorganic sulfideswhile trace element Be associates with organic sulfide or sulfatesin coal. These results are coincide with those of our experimentalresearch work (Lu, 1995). Because the trace elements associatedwith organic and inorganic sulfides vaporized easily and enrichedin fly ash ( linak, 1986; Flagen, 1981), so it can be predicted thatin the process of coal combustion, trace elements Cd, Co, Pb, Cuand Ni are more readily vaporized and enriched in submicrometer flash particulate and result in harmful impact to the atmosphericenvironment than other trace elementsReferencesBai jR, Liu fQ. Coal analysis. Beijing: Coal Industry Press. 1990Evan JB, Sellers GA, Johnson RG. Energy Fuels, 1990; 4: 440Flagen RC, Taylor DD. Eighteenth symposium(international)oncombustion. Pittsburgh: The Combustion Institute. 1981: 1227Gluskoter HJ, Ruch RR, Shimp NF Occurrence and contribution ofpotential volatile trace elements in coal, Urbana: IllinoisState Geological Survey. 1974Linak WP, Peterson TW. Twenty-first symposium (international)oncombustion(circular 499). Pittsburgh: The combustionInstitute. 1986: 399Linak WP, peterson TW. Twenty-first symposium (international)oncombustion. Pittsburgh: The Combustion Institute. 1986: 399Lu XH, Zeng HC. Huanjing Huaxue, 1995;14(6): 494Nichols BD Coal bearing strata(Eds by Murchison D and Weston TS)Edinburgh: Oliver and boyd. 1968: 269Norusis m. advanced statistics SPSS/PC+, SPss Inc. 1986Pires m. teixeira eC. Fuel. 1992 :71: 1093Raask e Prog Energy Combust Sci, 1985;11: 97Skeen CJ, Libby BJ, Crandell WB. Energy Fuels, 1990: 20: 63中国煤化工Solari Ja. Fiedler h. Schneider CL. Fuel. 1989: 68: 536CNMHGk: This work was supported by the Chinese National Foundation forproject of Pandeng and the Environmental Science Foundation of HUSTfle∥/ F yqikan_ htm抽取200 before/kjgk(2010)/jee/980212htm(第8/9页)200912-312256:31环境科学2E980212Received May 26, 1997)中国煤化工CNMHGfle∥/ F yqikan_ htm抽取200 before/kjgk(2010)/jee/980212htm(第9/9页)200912-312256:31