Preparation of solid medium for use in separation with gas-solid fluidized beds

Availableonlineatwww.sciencedirect.comMININGo Science directSCIENCE ANDTECHNOLOGYELSEVIERMining Science and Technology 20(2010)0743-0746www.elsevier.com/locate/jcumtPreparation of solid medium for use in separationwith gas-solid fluidized bedsLUO Zhenfu*, ZUO Weil, TANG Ligang, ZHAO Yuemin, FAN Maoming'School of Chemical Engineering and Technology, Key Laboratory of Coal Processing and Efficient Utilization ofMinistry of Education, China University of Mining Technology, Xuzhou 221116, China2CPT, Eriez Manufacturing Co., Erie, PA 16514, USAAbstract: The highly-efficient dry separation technique using a gas-solid fluidized bed is very beneficial for increasing coal gradea way that operates at low cost and high efficiency. Grinding experiments were performed using a planetary balla frequency converter. The effect of fed mass, rotation frequency of the mill, grinding time and the ball-size ratio on grindingce was investigated. The grinding parameters were optimized by numerical calculations using Artificial Neural NetworkMatlab. A regression equation for predicting the yield of the desired product(i.e, 0.3-0.15 mm magnetite powder) isThe maximum yield of 0.3-0.15 mm particles was 47. 24%. This lays a foundation for the industrial-scale production ofedium required for separation with a magnetite-powder fluidized bed.medium solids; magnetite powder; grinding; numerical calculation1 IntroductionHowever, only a relatively narrow size range of themedium is suitable for fluidized bed separation. It isThere is a serious coal shortage in China. Coal complex and costly to prepare a large mass ofaccounts for nearly 70% of China's energy consum- particles having a narrow size range. The study ofption. A highly efficient dry separation technique for low cost and high efficiency preparation methods tocoals located in regions where water resources are in prepare flotation medium is urgently requishort supply is urgently needed. The efficient utiliza-tion of high-carbon energy sources also requires2 Experimentalthis!.highly efficient dry coal beneficiationtechniques using gas-solid fluidized bed technologyThe raw magnetite ore was crushed by a jawhave been proposed. Many scientists and engineers in crusher. The crushed product was then screened intodifferent countries have contributed to the develo-sIze ranges>0.3mm,0.3to0.15,0.15to0.125,0.125pment of gas-solid fluidized bed dry coal beneficia- to 0.074 and <0.074 mm. The >0.3 mm fraction wastion2-6. In this technique the feedstock is separated then ground in a planetary ball mill powered with aformed from a fluidizing gas and a solid medium(7-10] frequency converter. The rotation system of the millaccording to the bed density within a fluidized bedis shown in Fig. 1-12. The factors studied during theObviously, the particle size distribution of themedium is one of the key factors that influence frequency grinding time and ball size ratio. The yieldfluidization performance and separation qualitof desired product was used as the response variableBecause of its high saturation magnetization and Additionally parameter optimization and numericalhardness magnetite powder is a typical medium for calculations were carried out using Artificial Neuralusein the separatibed. When the fineness of the Network in Matlabmedium is 90% to 95% less than 0.074 mm most ofthe individual minerals can be liberated from each 3 Results and discussionother. This is beneficial to the subsequent separation3.1中国煤化工 ing performanceived 10 January 2010; accepted 20 May 2010Corresponding author. Tel: 86 516 83995283THsCNMHG carried out toE-mail address: zhuo@ cumt. edu.awMn actors on grindingdoi:10.1016/s16745264090274Xce and TechnologyFig. 4 presents the yield variation withhanges in the grinding time. As can be seen in Fig. 4a grinding time of 3 to 4 minutes results in arelatively low yield of 0.3-0.074 mm particles: yieldfalls within a range of 59. 6%-77 5%. This is becausethe grinding time is too short. when the grinding timeincreases to 5 minutes the yield reaches a relativelyhigh value of 86.6%. As the grito 7 minutes the yield was 83.2. This small decreasein yield results from over grinding. Furthermore, theyield of 0.3-0. 15 mm particles first increases withFig. 1 Schematic of the rotation of the planetary ball milltime to a maximum value of 41.7% at a grinding timeThe variation in ground product yield with feed of 5 minutes. But after this point additional timeload is shown in Fig. 2. The yield of >0.3 mm causes the yield of 0.3-0.15 mm particles to decreaseparticles initially increases with increased load. Thisindicates partial grinding. The yield of the desired0.3-0. 15 mm particles reaches a maximum of 40.8%with a 400 g load and then decreases gradually. Inaddition, the yields of 0. 15-0. 125, 0. 125-0.074 and<0.074 mm particles varied in a nearly linear way as8:5013mnthe feed load varied. A feed load of 400 g is, therefore,optimum for the grinding process5035404.55.05.56rinding time(mm)Fig. 4 Variation of the grourThe yield variation as a function of ball size ratio15illustrated in Fig. 5. Note that the yield in >0.3 mm10particles increases with an increase in the ball sizeratio, which indicates partial grinding. The yield of0.3-0. 15 mm particles reaches a maximum of 41.9%Feeding capacity (gwith a ball size ratio of 0.2 and then decreasesFig. 2 Variation of the ground product yield with fed mass linearly after this point. Furthermore, as the ball sizeratio increases the yield of <0. 15 mm particles tendsYield variation as a function of rotation frequency to decrease linearly. a ball size ratio of 0.2 isis presented in Fig. 3. As shown there when the therefore, most suitable for the grinding processrotation frequency is 25 Hz the yield of desired,0.3-0.15 mm particles reached the maximum value of37.7%. The yield then decreases linearly as frequencycontinues to increase. The yield of <0.15 mm20.074mmparticles tends to increase linearly with rotationfrequency. This indicates that when the rotationequency was more than 25 Hz the feedstock wasfreover ground. So the optimal value for rotationfrequency during grinding is 25 HZ0160200240280.3270Ball size ratioFig 5 Variation of the ground product yieldwith ball size ratioA four-factor four-level orthogonal experiment wascarried out to investigate the combined influence ofntal results were18202224262830323436中国煤化工ware. The expmenRotation frequency(Hz)CNMHGicles in the 0.3 toFig 3 Variation of the ground product yieldrespectively. As shown in Table 2, a maximum yieldwith rotation frequencyof 46.03% was achieved with a feed of 583 g, a 22 Hzo Zhenfu et alrotation frequency, a 4 min grinding time and a 0.2Yield of 0.3-0. 15 mm particlesball size ratio=4895156+054275A0.80950BTable 1 Experimental results for yield of-606750C+25.71976D0.3-0. 15 mm particleswhere A is the fed mass; B the rotation frequency; CFactor A, rotation Factor C, Factor D, Yield ofthe grinding time; d the ball size ratio.uency time(min)ratioparticles(%6)3.2 Numerical calculation of optimized experimental parameters033234An Artificial Neural Network(ANN) is a com-05855050005000.33mutational model that tries to simulate the structuraland functional aspects of biological neural networks5600033The aNn function of Matlab was used to optimizethe grinding parameters. The program for analyzing9the experimental data is shown in Table 3. Thetraining time and the precision of the network were95000.171000 and 0.0001, respectively. A feed mass of 563. 2 g,10400a rotation frequency of 23.8 Hz, a grinding time of3844. 1 min and a ball size ratio of 0.21 results in a12600200.25295predicted maximum yield, of 0.3-0. 15 mm particles,43.6of 47. 2386%, which is consistent with the experimen1460035tal value of 46.03%1530038.14 ConclusionsTable 2 Optimized experimental results for1)During single-factor experiments a feed mass of0.3-0. 15 mm particle yield400 g, a rotation frequency of 25 Hz, a grinding timeof 5 min and a ball size ratio of 0.2 gave optimumles(%) bitA&.yield of the desired 0.3-0. 15 mm sized magnetitepowder of40.8%,37.7%,41.7%and41.9%,0.274348respectively25032502844.762) The regression equation for predicting 0.3-0.15mm particle yield is: Yield=4895156+0.54275A-220.80950B606750C+2571976D. With a583 g feed5342230274497a 22 Hz rotation frequency, a 4 min grinding time and6522290.2641.50a 0.2 ball size ratio the experimental yield reaches7455220.184562maximum value of 46.03%3)The experimental parameters were optimized02545.34using ann in Matlab. The calculated value of104460.1730.19maximum yield for the 0.3-0. 15 mm sized particleswas 47.2386% which is consistent with the observedThe regression equation for predicting yield can be experimental value of 46.03%written in terms of the coded factors as:Table 3 Artificial Neural Network(ANN)for optimizing the experimental parameters6006006006000170200250330.200.170.330.250.170.33033020025t=[2227740116940.228.23483342953863953819.7t=newff(minmax()[20. 1). (taeLtrainParam goal=00001; Precision of ne9 Obtained networkb=80:1.5:70[A, B, C, D]ndgrid (a, b, c d ):B(),A(Y=sim(neL p2);[Ymax,r中国煤化工CNMHG(A(r), B().Cr),D(r)IScience and TechnoloVol 20 No5Acknowledgements6Houwelingen J A, de Jong T P R Dry cleaning ofoRicaThis work was financially supported by thebelgica,2004,7(3/4):335-343National Natural Science Foundation of China [7]Luo Z F Fan MM, Zhao Y M, Tao XX, Chen QR.Chen ZQ. Density-dependent separation of dry fine coalNos.50921002 and 90510002)and the National Highin a vibrated fluidized bed. Powder Technology, 2008,Technology Research and Development Program of187(2):119123China(No 2007 18).[8] Luo Z F, Zhu J F, Tang L G Zhao Y M, Guo J, Zuo W.Chen S L Fluidization characteristics of magnetiteReferencesder after hydrophobic surface modification, Interna.ional Joumal of Mineral Processing, 2010, 94(4):166-17naxx,[9] Tang L G Zhao Y M, Luo Z F, Liang CC Chen Z Q,Zhou n x. Effect of chemicalXing H B. The effect of fine coal particles on theiboelectrification of coal and mineperformance of gas-solid fluidized beds. internationalScience and Technology, 2010. 20(3Journal of Coal Preparation and Utilization, 2009, 29(5)[2] Dong X, Beeckmans J M. Separation of particulate sol-ids in a pneumatically driven counter-current fluidized[10] Luo Z F, Zhu J F, Fan M. Zhao Y M, Tao XX Lowcascade, Powder Technology, 1990, 62(3): 261-267sity dry coal beneficiation using an air dense medium[3] Mak C, Choung J, Beauchamp R, Kelly D J A, Xu Zfluidized bed. Journal of China University of miningTechnology,2007,17(3):306-309mineral matter for mercury rejection from alberta [11] Tang L G Zhao Y M, Luo Z E. Chen Z Q, Liang CC,Xing H B Fluidization characteristics of medium solidsPreparation and Utilization, 2008, 28(2): 115-132with a wide size range. Journal of China Universiry of[4] Zhao Y M, Luo Z F Chen Q R. Fundamental andMining Technology, 2009, 38 (4): 509-514(In Chial developments of dry coal cleareview.Coal Preparation Society of America(CPSA) [12] Chen SZLiWX, Yin Z M. Research on workingJournal,2004,3(3):14-l8neary high-energy ball mill. Mining[5] Sahu A k, Biswal, S K, Parida A. Development of airMetallurgical Engineering, 1997, 17 (4): 62-65.(Indense medium fluidized bed technology for dry benefi-ciation of coal-a review. Intemational Journal of CoalPreparation and Utilization, 2009, 29(4): 216-241中国煤化工CNMHG