Model analysis for combustion characteristics of RDF pellet

Vol.12 No. 3Trans. Nonferrous Met. Soc. ChinaJun.2002[ Article ID] 1003 - 6326( 2002 )03 - 0534 -06Model analysis for combustion characteristics of RDF pelletLIU Gui- qing' , H. Hakamada? , Y. Itaya' , S. Hatano' , s. Mori'( 1. Department of Chemical Engineering , Nagoya University , Japan ;2. Chemical & Environmental Engineering Department , Kawasaki Heavy Industries , Ltd. , Japan)[ Abstract ] Fundamental studies of the combustion characteristics and the de HCl behavior of a single refuse derived fuel( RDF ) pellet were carried out to explain the de-HCl phenomena of RDF during fluidized bed combustion and to providedata for the development of high efficiency power generation technology using RDF previously. For further interpretingthe devolatilization and the char combustion processes of RDF quantitatively , an unsteady combustion model for singleRDF pellet , involving reaction rates , heat transfer and oxygen diffusion in the RDF pellet , was developed. Comparisonsof simulation results with experimental data for mass loss of the RDF samples made from municipal solid waste ，woodchips and poly propylene when they were heated at 10K/ min or put into the furnace under 1 073 K show the verifiabilityof the model. Using this model ,the distributions of the temperature and the reaction ratio along the radius of RDF pelletduring the devolatilization process and the char combustion process were presented , and discussion about the inference ofheating rate on the combustion characteristics were performed.[ Key words ] refuse -derived fuel ; combustion characteristics ; de HCl ; modeling analysis[ CLC number ] X 705[ Document code ]A1 INTRODUCTIONflue gas and the release of dioxins is quite low aswelf2,3 Tests of RDF in fluidized bed combustionWith the development of industrialization and( FBC ) boiler have given unexpected high de _HCI effi-urbanization , proper treatment and recycling utilza-ciency( 60% ~ 85% ), resulting in very low concen-tion of wastes have become important tasks of our so-trations of HCl( below 10- 4 ) at the inlet of bag-ciety. The recycling approaches of solid wastes can behouse even under high furnace temperature ( 830~classified as material-recycle and energy-recovery. Al-850 C 2.4.5,1 , while the HCI removal efficiency inthough many wastes such as wood , paper and naturala stoker-type incinerator for MSW with lime injectionfiber can be recycled as material to reduce the con-was as low as4%~ 25 % , resulting in very high con-sumption of virgin materials , since the original mate-centration of HCl in the flue gas between5.51X 106rials of plastics , synthetic rubber and synthetic fibers~7.47X 10- 6come from petroleum ，thermal recycling of theseIn our previous studies，the combustion charac-wastes can not only treat the wastes effectively but al-teristics and the de-HCl effect of a single RDF pelletso reduce the consumption of petroleum used as fuel.have been investigated through a series of experiments .So the thermal recycling of wastes bcome an attrac-carried out in an electric furnace 6,71. Through thesetive approach for waste treatment and energy recov-fundamental studies ,it has been known that , 1 ) theery. Among the thermal treatment technologies ，usemass loss of RDF was mainly due to devolatilizationof refuse-derived fuel( RDF ) as starting materials forand the temperature gradient in the RDF pelletcombustion and gasification processes presents severalstrongly depended on the heating rate and the oxygenadvantages over direct use of municipal or other solidconcentration in the ambient gasos ;2 ) calcium com-wastes. RDF is generally produced by selecting thepounds added to RDF can capture more amount of Clcombustible fraction of municipal solid waste( MSW )released from the RDF under quick heating conditionby mechanical sorting and processing. This results inthan under the slow heating condition , and oxygena relatively constant composition and good transporta-concentration of the feed gas had a strong inference ontion and storage possibilities ,since putrescible compo-the capture of CI by the ash. The higher the oxygennents are eliminated with lime addition. Therefore ,concentration was , the lower the capture fraction ofRDF can be produced dispersively then transported toCl wa57].a large-scale treatment facility to accomplish efficientinformation on com-energy recovery and minimization of pollutior[1l. Es-bustic中国煤化工ICl mechanism duringpecially , some recent investigations have shown thatcomb:MYHCNMHGely,inthisstudy,ande-HCl behavior of the calcium compound in the RDFunsteady combustion model，involving reaction rate ,during the combustion of RDF in fluidized bed can ef-heat transfer and oxygen diffusion in RDF pellet , isfectively control the corrosion problem caused by thedeveloped. Using this model ，the distributions ofφ [Reeived.万京数据- 10- 16 ;[ Accepted date ]2002 - 01 -24Vol.12 No .3Model analysis for combustion characteristics of RDF pellet535.temperature and reaction ratio along the radius ofReaction rate of char combustion ( based onRDF pellet during the devolatilization process and theGrain Model):char combustion process are predicted , and discus-E2R( xr), T(r ))= A2exp(sions about the inference of heating rate on the com-RT( rbustion characteristics are performed.C"(rX1-x{r)P/3 (8)2 MODEL3EXPERIMENT AND DETERMINATION OFPHYSICAL PROPERTIES AND MODEL PA-The samples used in this study are of shape inRAMETERScylinder with diameter between 1.5~2.0 cm andlength between 3.0~ 6.0 cm. According to the re-3.1Combustion experimentsults obtained in the previous study , the combustionAn electric heating reactor was used for combus-process of RDF consists of two sequential processes :tion experiments on single RDF pellets. The experi-the devoltilization process and the char combustionmental system was fully described previously61, soprocess. So in the combustion model two-step reac-only the essential features were given herein. The re-tion kinetics is considered. It is assumed that the charactor was constructed by connecting a vertical quartzcombustion begins when the overall reaction of de-tube ( 53 mm in diameter ,1 000 mm in length ) work-volatilization completes 99 %. To simplify the calcula-ing as primary reactor and a horizontal quartz tubetion program，following assumptions are also accepted( 45 mm in diameter，300 mm in length ) working asin the formulation of present model : 1 ) Solid spatialsecondary combustion chamber. Two separated elec-dimensions are considered constant ;2 ) As the aspect :tric heaters with PID controllers were used to main-ratios of the samples are more than two , the end ef-tain the temperatures of two combustion chambers in-fects caused by the mass and heat transfer along thedependently. A quartz pan used as sample supporteraxis are negligible ; 3 ) The reaction order of de-was hung on a digital electric balance with a long thinvolatilization is one.Assumption 1 has been widely used in the model-Inconel alloy wire. When the experimental systeming of biomass pyrolysis. In the case of RDF，it iswas set to measure the temperatures in the RDF pel-based on the observation results obtained by pre-ex-let，the sample supporting system was replaced by aperiments，where no relevant shrinkage phenomenathin quartz pipe including the probes of thermocou-were observed. Assumption 3 has been verified withples. Purge gas was supplied from the bottom tothe experimental results in TG analysis using pulver-maintain the oxygen concentration constant duringized RDF sample 4].each run.' The oxygen concentration in the purge gasThe model equations based on the preceding as-was regulated from0 to 21% by changing the mixingratio between pure nitrogen and air. The total flowsumptions are listed as follows.rate of the purge gas was kept constant at 5X 10- 2Heat balanceT.. 1aT. q _ 1aTm*/min( at 298K and 101 kPa ) during all runs.ar2.rJrλaJt(1)Pure oxygen( > 90% ) was used as secondary air toReaction heatpromote the burn-out of organic substances in the flueq=R(x(r ),T(r ) -OH; )(2)gas in the secondary chamber.( for devolatilization and char combustion process ,i=1 ,2 respectively )3.2 RDF samples and chemical propertiesMass balance :Experimental runs were performed with three8x;kinds of RDF :a commercial RDF made from MSW,psf.at = R(x(r),T(r))(3)a model RDF made of wood chip and polypropyleneDegree of conversion :pellet. Their chemical features are given in Table 1.x; =wo,i - Ws,i(4)3.3 Physical properties and model parametersWg0,iDiffusion of oxygen :The necessary physical properties and model pa-3c=D.((72+72;)+Rf xKr))5+ 1?c. (5)rameters were determined through pre- experimentalwork中国煤化工.ties of RDF( λo ) andConsumption of oxygen :theirwith a Thermal Con-Wxodx2ductivMYHc N M H Gature ,then the valuesRf x6(r))= -2M dt(6)were calibrated for the operating temperature withReaction rate of devolatilization :the Kuni & Smith Equation. The thermal conduc-R(x(r),T(r)= P0Arex(-Ei,)tivity for the sample during the devolatilization andRT(r))char combustion process was evaluated according to a(1-x(r ))(7)proportional relation between the thermal conductivi-536 .Trans. Nonferrous Met. Soc. ChinaJun.2002ty and the density of the sample.The physical properties and the model parame-The reaction heat( - -△H; ) and the specific heatters obtained are given in Table 2 and Table 3 , re-capacity( Cp ) of RDF were measured by means ofspectively.differential scanning calorimeter ( DSC ). The datapresented in Refs.[ 8 ,9 ]were adopted for the specif-4 RESULTS AND DISCUSSION .ic heat capacities of char and volatiles.The activation energy( E1 ) and the pre-expo-4.1Comparison between experimental data andnential factor( A1 ) for the devolatilization were ob-calculation results for mass loss curvetained through TG analysis by using of the sampleThe calculation results by present model and thepulverized from RDF under an inert atmosphere+ J ,experimental results for RDF-m are shown in Fig. 1.and the activation energy( E2 ) and the pre- exponen-In the case of slow heating( 10 K/min )in air( CA=tial factor( A2 ) for the char combustion were deter-21% ),as shown in Fig. 1( a ), excepting the resultsmined by TGA under different oxygen concentrationsfor the initial period of the process , the mass lossusing pulverized char sample obtained by pyrolysis ex-curves are well agreeable in the overall process. Theperiment under 1 073 K(11]difference that occurred in the initial period is believ-The effective diffusion cofficient of the oxygenable to be the result of the vaporization of moisture inwas evaluated by the method presented in Ref.[ 12 ]the sample.Table 1 Properties of tested RDF ( Dry base except for moisture )Proximate analysis/%Utimate analysis/ %PelletheatingSamplevalueMoistureVolatileFixedNOSCCadensity/(kg m-/k} kg-1 X as received )carbonRDF-wd 20 1108.181.217.8 1.045.2 5.5 0.9 47.4 0.02 Trace <0.11 540RDF-pp 45 850 .<0.199.3 0.1 0.684.415.0 0.34 <0.1 0.01 <0.01 3.0RDF-m 15 2806.369.510.1 20.435.2 6.0 0.7 32.3 0.10 1.20 9.21 560Table 2 Physical properties of RDFsDiameterDensityThermal conductivity，Specific heat capacity，Diffusion cofficient ,/cmw:/(kg m7 3) λ;/(W m~↓K-1 )cp(J kg~'k-1)D。/(m^s 1 )RDF-mRDF2.01 1540.291097+3.44x10-2T-Char1.66980.101 003 + 2. 09T{10]Ash5810.068.0X10-6RDF-wd RDF1 0450.201 320+ 3.44X10-2T1.33601 003+ 2.09T10]121.610-RDF-pp RDF1.02000一4.2x10-2+Volatiles3.8x10-+T-1 10-1117X 108τi12]Table 3 Kinetic parameters for devolatilization and char combustion of RDFsDevolatilizationChar combustionPre- exponentialActivationReactionHeat ofentislenergyreactionfactor/s-1/(kj mol-1) order /(kJ mo中国煤化工order《F emo-1)1.7x 10362.01.0x 10YHCNMHG122.4X 103RDF-wd7.0x 10472.011.0x10-33.1X 10420.020.0x 103RDF-pp .6.3X 109140. 11.0x10-2Vol.12 No.3.Model analysis for combustion characteristics of RDF pellet537 .In the case of quick heating where the samplethe progress of reaction is homogeneous within thewas inserted to the isothermal atmosphere( 1 073 K )RDF and accelerated with heating time. During theshown in Fig. 1( b), the results are well consistentchar combustion process( as shown in Fig. 2( b )),throughout the devolatilization process. But little de-with the progress of reaction , a peak temperature isviation presented in the char combustion process ofshifted from the surface to center along the radial di-RDF-m and RDF-wd. This phenomenon is a reflec-rection. From the reaction ratio curves it is found thattion of the complexity for reaction during char com-there exists a clear boundary between reacted part andbustion process. By the comparisons , the reliability ofunreacted part along the radius , indicating that thethe model calculation is verified.reaction occurs on the surface of a shrinking unreactedCOr4.2 Model prediction of temperature and reaction4.2.2 In case of isothermal condition at 1 073 Kdistribution for RDF-mWhen the sample was inserted to the hot zone4.2.1 In case of slow heating at heating rate of 10( T= 1073K , CA= 10% )of the furnace instanta-K/min in airneously ,it was heated quickly from the surface to theThe calculation results for the temperature distri-center. The temperature distribution and the reactionbution and the reaction ratio of RDF-m with time areratio of RDF-m with time are shown in Fig.3.shown in Fig.2.The features both for the temperature and reac-During the devolatilization process ( as shown intion ratio distribution during the devolatilization pro-Fig.2( a )), the temperature increases with heatingcess are obviously different from that in slow heatingtime almost evenly along the radial direction ，whilecase.While the temperature close to the surface of1.0(a)1.0(b)AtmosphereRDF-m: airRDF-wd: air0.85RDF-pp: N2●RDF-wd(expt. )， RDF-m( expt.-Calc.0.6RDF-pp( expt. ).0.6个“台RDF-m(calc. )RDF-pp(calc.)0.4-_RDF-m .0.2-中0o00......0.2RDF-wdRDF-pp200400 600 800 10001200o---十Time/sTime/ 10'sFig.1 Comparison of calculated results with experimental data(a)- -Slow heating( heating rate= 10 K/ min );( b)- -Isothermal heating furnace temperature= 1073K)1300(6)302 400s一44 800s1 200750-3600 s0.82200号4200s:出11005400s7002400s|3 000 s1 000650- 2200s2000510.42700s日602000s550F1800s1 8005中国煤化工--1 600 s1 600s500产0.204060810TYHCN MH Go.8 1.0Position in radius r/RFig.2 Temperature and reaction ratio distribution in RDF-m heated at 10 K/min in air( a)-Devolatilization process ;( b )-Char combustion process-Temperature ----Reaction ratioTrans. Nonferrous Met. Soc. ChinaJun.20021200-240s1 300[150Qsi1 000-1 Furnare temp0.81 200F1800s1 200j900 s+0.8240s600sL300 s800180s0.61100.2100si120s906000.4官100060840030 s8| si40.2900-0.2200o 0.2 0.4 0.6 0.8 1.0oPosition in radius r/R(a)(b)Fig.3 Temperature and reaction ratio distribution in RDF -m inserted in furnaceat 1073K ,CA= 10%(a)- -Devolatilization process ;( b)- -Char combustion processTemperature ----Reaction ratioRDF is increased dramatically to near the furnaceThe results also show that char combustion pro-temperature , the temperature inside the RDF keepscess is not influenced by the heating rate. During charquite low. Therefore ,a significant temperature gradi-combustion , it keeps a clear boundary between theent forms along the radius at first and then fades withreacted part and unreacted part along the radial direc-ime. A reaction boundary is formed in the earlytion, so a shrinking unreacted core is formed , andperiod , but it disappears gradually accompanying thecombustion seems to be controlled by the diffusion offading of temperature gradient. This phenomenon in-oxygen through the reacted layer.dicates that the reaction is shifted from thermal trans-fer controlled process to chemical reaction controlledNomenclatureprocess.c-Concentration of oxygen in the char , mol/During the char combustion process , the featuresof temperature and reaction ratio are similar to thoseCλ- -Vclume fraction of oxygen in purge gas ,in the case of slow heating , where a clear boundaryexists between reacted part and unreacted part alongD。- -Effctive diffusion cofficient , m2/s ;the radius , and the reaction occurs on the surface of afi- -Fraction of volatile ;shrinking unreacted core. However ，the temperatureM- -Average molecular mass of combustiblepeak inside the RDF is not so sharp as in the case ofslow heating. This is attributed to the moderate com-component in char ,g/mol ;q- -Heat generated unit volume , W/m2 ;bustion in low oxygen concentration atmosphere.R- -Radius of RDF pellet ,m ;Therefore , the char combustion rate seems to be con-trolled by the diffusion of oxygen through the reactedRg- -Oxygen consumption rate ,mol( m' s);layer.r- -Radial distance of RDF pellet ,m ;T,Te, T,- -Temperature parameter ,K ;5 CONCLUSIONl一Time ,s ;ws,i- -Concentration of reactants in solid , kg/The quite good agreement between the results ofm3 ;model calculations and the experimental data showsW.0 ;一Initial concentration of reactants in sol-that the model presented herein is a viable tool for theid ,kg/m3 ;quantitative description of the overall combustion pro-Wsc0一Initial concentration of reactants in char ,cess of RDF pellet.kg/m中国煤化工The results of model calculations for the distribu-tion of temperature and reaction ratio in the RDF in-YHCNMHG ratio;I nermalduttuslvity ,mt/s ;dicate that the heating rate is an important parameterλ ,λ0 ,入一Apparent heat conductivity , W/( mduring the devolatilzation of RDF , where the processK);can be changed from reaction- rate- controlling to ther-ρo一Apparent density of RDF ,kg/m3 ;mal-conductiyity-ontrolling as the heating rate increases.ρs一Apparent density of solid ,kg/m3 ;Vol.12 No.3Model analysis for combustion characteristics of RDF pelletSubscripts :fluidized bed[J] Kagaku Kogaku Ronbunshu , 199.0-Initial25 :921 - 927.1- -Devolatilization process[4] Narukawa K , Chen Y , Yamazaki R ,et al. Combustion2- -Char combustion processcharacteristics of RDF[J] Kagaku Kogaku Ronbunshu ,c- -Center at cross section of RDF1996 ,22 :1408 - 1414.s- -Surface on the wall of RDF[5] Mori S. Gomi kokeika nenryo hatsuden systemn[J]Nenryo & Nensho, 1996 ,63 :563 - 570.calc.- -Calculation value[6] LiuG, Yamoeaki R ,HatanoS ,et al. Combustion char-expt. - -Experimental valueacteristics of single cylindrical RDF[J] Kagaku KogakuRonbunshu, 1999 ,25 :79- 84.Acknowledgement[7] Liu G ,Itaya Y ,Yamaraki R ,et al. Behavior of chlorineThis study was partially supported by the JSPSduring the combustion of single RDF plle[J] Kagakugrant for Research for the Future Program ( No.Kogaku Ronbunshu ,2001 ,27 :100 - 105.97100604 ) and this presentation was partially sup-[8] Koufopanos C A , Maschio G , Lucchesi A. Kinetic mod-eling of the pyrolysis of biomass and biomass componentsported by the Grantin-Aid for International Ex-[J] The Canadian Jourmal of Chernical Engineering,change from Research Foundation for the Electrotech-1989 ,67 :75 - 84.nology of Chubu.[9] Di Blasi C. Analysis of convection and secondary rection .effects within porous solid fuels undergoing pyrolys[ J][ REFERENCES ]Combust Sci Tech , 1993 ,90 :315.[10] Perry R H ,Green D W. Perry' s Chemical Enineers'[1] Kagiya T. Practical use and future topics concening RDFHandbook[ M] New York : McGraw-Hill , 1984.10technology for municipal solid waste management[ J ]- 15.Haikibutsu Gakkaishi , 1996 ,7 :352 - 362.[11] Hakamada K ,LiuG , Itaya Y ,et al. Model analysis for[2] Sugiyama H , Kagawa S , Kamiya H ,et al. Chlorine be-pyrolysis and char combustion of RDF[ A] Proeedingshavior in fluidized bed incineration of refuse derived fuelsof Autumn Conference of SCEJ[ C] Kanazawa , Japan ,[J ] Environmental Engineering Science, 1998 , 15 :971999.一105.[12] Sato I. Bussei Josu Suisnho[ M] Maruzen ,Tokyo, .[3] Kondoh M , Hamnai M , Yamaguchi M,et al. Dioxin e-Japan ,1965.129- 138.mission with the combustion of RDF in an inner cycling( Edited by YUAN Sai-qian )中国煤化工MYHCNMHG