Sei sulla pagina 1di 26

Co-firing Final Period Project Energy Leeds Tel: Co-investigators: BCURA From Fax: From: Fax; 01133432507, 01/07/2007

01418853860, REPORT Report Agreement Report, University LS2 Industrial 01/01/2006 of and Manager Coal/Biomass Project: 9JT Resources December Number Professor to Supervisor: to fax; Dr email:aduncan@doosanbabcock.com. email:smithchell@doosanbabcock.com. 01/01/06 31/12/2007, 30/06/2007: Research J. BM. and 2007 01132467310, 79 Alan the Jones, toinstitute Estimation 31/12/07 Williams Mr DrAngus Professor Stuart email: Duncan, of Mitchell, fueaw@leeds.ac.uk M. Burnout Pourkashanian Doosan Doosan and Babcock NOx Babcock Formation Energy Energy Ltd,Ltd,

Co-firing of Executive Summary coal and biomass is widely used in the UK, Europe and the US and is now largely routine at least up to 20% biomass on a thermal basis. There are sti problems associated with the use of biomass, for example, often after grinding t ll biomass some particles may be larger than pulverised coal and this can result in he enhanced carbon in ash. Whilst this problem can be minimised an understanding of the combustion of the larger particles is useful. An additional issue is that bi contains a considerable amount of potassium which is easily released and can cau omass deposits and corrosion on boiler tubes. Finally although biomass usually contain se s lessand fuel-N known than a better coal the understanding detailed mechanism might make of NO possible releasethe from reduction biomass of is these not well leve The aim of this ls. underpinning science project foris the thus co-firing to improve of coal the and level biomass of understanding in utility boilers. of the T project is concerned with a modelling study of the co-firing of coal and biomass his , upthermal 20% to of the total input. The project is particularly directed towards tho biomass materials that form large, irregular particles that do not burnout readi se ly,predicting in After would and some be undertaken consideration the formation of cofiring it was of NOx. decided coal andthat biomass a computer in an industrial modelling combustion investigation tes t facility, the CTF at RWE npower at Didcot. This had rating of 0.5 MW and was fit with a considerable ted that combustion Doosan test Babcock facility number (Mitsui for ofaexperimental Babcock) wide range Mark of results III coals burner. had coal/biomass and been Thisobtained had theblends. advantage using this Thre e project computed single coals and fordata and a torrefied fivethem for coal/biomass wood. was made Thisavailable. mixtures was not part were In of selected addition, the original by results RWE proposal npower have been for but the in of view the interest in this material this case was examined although no experimental results are available at this time. A second decision was to use a commercial co Fluent, de In typical version co-firing 6. of coal and biomass the combustion of the coal itself is th e major component of the flame and an accurate mechanism for this part of the over combustionflow all including is necessary. and heat transfer Fluent contains steps and a aspects default of coal that combustion code canmechanism be modified by theand this was done during the project. Therefore in the first stage of the user, particular attention has been paid to the combustion of a single coal, Gascoigne project For Wood, the where first a step substantial of coalamount combustion, of data namely is available. devolatilisation we have previousl y used the network devolatilisation code FG-DVC and we decided to use that in this project. By comparison with other similar coals a rate expression was derived. T has a high experiments his laboratory value compared andwith thismany has a such major expressions significance obtained on the byrate small ofscale volatile which release influences flame holding and near-burner flame shape. Since the rate is fa st it that the rate heating-up of the particle plays an important role and so implies does thechoice of the specific heat. Likewise the volatiles yield was determined correct compared and with experimental data and values obtained by using rapidly heated wire previously The the meshresults case experiments. computed with obtained some Thefor was success. char for flame the burnout combustion shape, model flame andtemperature, with NOx 4% oxygen model were in radiation the ones flue used to gas the and furnac e walls, exit gas temperature, unburned carbon in ash and NOx. Different turbulence model s a 2

were tested as was the effect of different soot models on the incident radiation Comparisons measurements. determined flux. flame were In temperatures made of the addition theflame computed for similar shape data against flames was compared experimental available against in the video experimentally literature. Go agreement was obtained except for the radiation measurements for reasons which a od still uncertain at the present time. But it is probably related to variation in re as the black coal particle is converted into an ash which has a low emissivity. emissivity the question of the emissivity of the coal char/ash seemed to have been resolved Whilst ago years this does not seem to be the case and recent studies have reopened the debat A number e. 1. 2. 3. 4. The Russian South reaction of/Palm African/Miscanthus African coal data/Olive biomass coal Kernel and volatile /milled expeller waste blends Giganteus wood (15% yields were (15% (20% thermal). examined, thermal), (20% for thermal), the thermal), additional namely coals, South African an d The and Russian, biomass the miscanthus were particles obtained contained were in the examined large same particles way by SEM. as for It whilst Gascoigne was clear the PKE Wood that and coal. the the milled Olivewood waste not (although with particles which are slightly larger than the coal). This info did was effect rmation The biomass particularly. Computed usedmodels, inof heating-up the the The combustion the large effect heating-up equations biomass ofmodels. moisture were equations, particles obtained on the has devolatilisation biomass for antwo effect particles particle of and the sizes was choice burning examined. andof rates different the The heating devolatilisation conditions uprate to 10% andmoisture, yields were andestimated less accurately via theup bio-FG to 50% model, moisture. and in the case of the former the value is again a higher values. If this conclusion is this correct implies the importance of the factors controlling the heating-up rate, the of the specific heat data and the effect of moisture. The effect of the Biot num accuracy ber is important. aerodynamic Large and heating articles upwere equations. approximated The reaction by an equivalent rates of large sphere particles for bothcoul d also be approximated in this way but it is not a good approximation for burn-out. Stu of suspended single particles were made using still and video photography to che dies the compared ck The and models used. calculation against of furnace experimental exit temperatures, results for all unburned the blends carbonand in single ash were coals. made V were also calculated of NOx. For the calculations involving a biomass an allowan alues was level ce The made for of agreement the presence forof all proteins these results and amino is discussed acids in the andbiomass. the way in which t hisgeneral be In incorporated can the level into of industrial agreement models is good identified. although the model could be improved b y using a better heat transfer model incorporating a variable emissivity as the re progresses. Better data on the emissivity of the slag at high temperatures would action advantageous. powerful be computer Computation so that the of these coal combustion flames could model be improved could utilise by using a more a more detaile d coal combustion model which is available. Such models exist but can only be used conjunction in with a zone model because of the complexity of using it with a full model, but industry models could use the more detailed model at the present time CFD . 3

1. Co-firing Introduction of coal and biomass is widely used in the UK, Europe and the US and is now largely routine at least up to 20% biomass on a thermal basis. There are sti problems associated with the use of biomass, for example, often after grinding t ll biomass some particles may be larger than pulverised coal and this can result in he enhanced carbon in ash. Whilst this problem can be minimised an understanding of the combustion of the larger particles is useful. An additional issue is that bi contains a considerable amount of potassium which is easily released and can cau omass deposits and corrosion on boiler tubes. Finally although biomass usually contain se s lessand fuel-N known than a better coal the understanding detailed mechanism might make of NO possible releasethe from reduction biomass of is these not well leve The aim of this ls. underpinning science project foris the thus co-firing to improve of coal the and level biomass of understanding in utility boilers. of the T project is concerned with a modelling study of the co-firing of coal and biomass his , upthermal 25% to of the total input. The project is particularly directed towards tho biomass materials that form large, irregular particles that do not burnout readi se ly,predicting in and the formation of NOx. The specific objectives of this project were : To ii) devolatilisation iii) iv) biomass from 2. In i) the design develop Experimental To Topreliminary an use test and industrial sub-models understanding the coal, the approaches Methodology and robustness and Combustion discussions to char toaccurately of test combustion developed the of these Test these with heat describe Facility. above models the andIndustrial in models irregular mass in against the using CFD transfer release models shaped laboratory experimental Supervisor associated of for biomass water the experimental and co-firing data and particles, with volatiles, available RWEof data, npowe r itconcluded that the best approach to the project was to use experimental data was effectively available from determined RWE npower thefor systems theirstudied Combustion which Test areFacility. given below. ThisIt decision was decided that the Gascoigne Wood data would be used to test the model and then the binary Gascoigne 1. studied 2. 3. 4. Russian 5. South 6. 7. In (20% data coal-biomass South addition thermal) is African African/Miscanthus on coal, /Palm available African Wood the the fuels which Coal, /milled Kernel has coal-this CTF combustion /Olive been for would and is which expeller wood this included used investigated waste be is Giganteus is (20% of case. investigated. used as used (15% a one measurements thermal), blend as In as of thermal), (20% thermal). aone using all standard the of thermal), the of South blending a the CFD experiments ofAfrican because flame blending model, coals, radiation. it Coal although considered coals, has with been no torrefied widely experimental here thewood oxyg content of the flue gases was 4% (dry). en 4

on the construction Information was provided of the byfurnace RWE npower and the (Dr burner P Stephenson and these and details Dr M Whitehouse) are given in mounted Appendixin 1.aThe rectangular burner isfurnace a 0.5 MW and rated replicates Doosan a Babcock burnerMark in aIII third burner row of a wal power l-fired The combustion capability. CFD station. model of NOx used coal wasor is computed the Fluent coal/biomass using Version blends the associated 6with coalFluents combustion FluentUser post-processor model Defined to compute Function package. the combustion Most details ofof pulverized the use of coal that [1-4], modelfor have co-firing been previously [5] and with published biomass foralone the [6 , 7].of theReports Some associated important [8,computational 9] and are briefly details summarized are givenan inAppendix these references 2. Some aspects and the issue Coal volatiles CO+0.5O2= C CO All who In University. of (char) +is order User the =is O2=CO2 char a computational Defined that + member to +O2 CO2 + = O2 (R5) meet most gases (R3) =of CO Functions CO the calculations (R the +stated tar H2O 4) work Centre Rwas (R developed (1) objectives 2) for undertaken weComputational used inof the bytwo the project Dr Project step L Fluid Ma, are mechanism a Senior Dynamics the considered following Research at Leeds later. mainFellow studies A key made: decomposition were (1) The and nature products, and properties (2) Water ofevaporation, the coal and(3), biomass Turbulence and their andrates radiation of Combustion 2.1 (a). The and models, The compositions are Composition (5)Test Nature listed NO in prediction and Facility. of Appendix Properties the coal Coals program, 3.and of the and Data Biomass. and for Coal biomass (6)and other Application materials coals Biomass wasMaterials of the were that given used CFDby model by in RWE recent to npower thepr [1-8]. The values for the torrefied wood were based on data taken from reference ojects The 9. data in Appendix 3 are presented in an as received form and were used to deriv e In theorder quantities to demonstrate required the by FG-DVC properties and the of the Fluent fuels program. in relation to other commonl y usedthey are plotted on a van Krevelen plot below with the exception of the to fuels wood. The properties of this lie in an intermediate poison between coal and biom rrefied 5 ass.

The particle Figure (b) Particle 1. Thesizes Sizes van Krevelen of the for theCoals Diagram coals and were for Biomass. provided Biomass by andRWE Coal npower. The milled biomass experiments undertaken samples vary to the considerably replicate moisturetypical content depending sizes wason lowered found the type in byaair of power biomass. drying plant. and InThe thesamples milling present wasof mil miscanthus, led microscopy Figure 6 2 (b) (a) at olive Miscanthus Milled Leeds waste, University, Wood wood and and PKE these provided are shown were below. examined by electron

The sizes Figure 2 (d) (and c)PKE Olive Waste shapes of the milled Miscanthus particles are very similar to the wood (Pine) because they are both fibrous materials. The Olive waste is very sim milled to the Palm kernel because they are both nut-like materials, although in both ca ilar there are slight variations in the particle sizes. The mixtures of biomass and t ses he coalscomplex issues in the definition of the size distribution inputted in to present model. model the CFD but These there size isdistributions an issue about can the however largerbe sizes handled of the adequately wood andby Miscanthus the CFD wh willfigure ich The be discussed below shows later. the distribution in a mixture of South African coal and 1 Olive waste. It consists of a large number of smaller particles with about 5% ab 5% 150 m. The presence of larger irregular sized biomass particles has implication f ove both their or Figure 7 3. Blend reactivity of South andAfrican their aerodynamic coal and 15% behaviour, Olive waste. and are discussed later.

In the 2.2 Thecase Rates ofand theYields coals,of the the values Coal for and devolatilisation Biomass Decomposition rate constants Products Ad and Ed, for first overall the order step obtained from previous studies on devolatilisation netw codes were used. The devolatilisation rate constant parameters for the release o ork f tar, gas and total volatiles for the four coals calculated by FG-DVC [11] and values from taken On average our previous for these studies coals, [2, for 4,the 8] process are given ofin overall the Table devolatilisation, 1 below. this le ads14 -1 activation The to ancoals test energy, usedE, inof the 230 present kJ moland studies a pre-exponential are similar infactor, nature to A, those of 4.2. in10. Tab le 1 as by their close proximity in Figure 1, and consequently the above value indicated used Table Devolatilisation Coal Activation Tar Thoresby Asfordby Betts Pittsburgh#8 Ensham 8 Prodeco Experimental was 8 8 Gas for (Origin) 1. Lane 141214 141414 141314 2.4 Total 2.3 all Reaction 3.3 2.0 Energy, 2.5 x xdevolatilisation the x 2.3 102.7 Tar 102.4 102.4 102.1 x coals Rates 102.3 x Rate Gas Ed 102.2 x x102.4 x102.4 (J Total inx Parameters 102.4 102.3 x FG-DVC kmol 102.4x102.1 this 102.3x103.7 x x104.1 xrate -1) 103.0 study. 103.8 103.7 computed for Pre-exponential xx103.2 data x104.6 This Tar, 104.6 104.2 103.2 x can 104.8 curve Gas x x104.3 be x105.2 and 105.2 104.8 103.3 x analysed is 103.8 factor, Total x plotted x x10 x 10 10 10 x in Volatiles, 10 Ad ain (s-1) number Appendix [2,ways of 4. 8]. 4, so t there can be compensation between values of A and E. In the literature there are hat effectively choices between pairs of high activation energy and high pre-exponen factors or pairs of low values for E and A. The former are obtained by the netwo tial codes (FG-DVC, rk heating rates, the CPD latter or FLASHCHAIN) constants oroften by high from temperature TGA where inadequate studies with data high analysi s is The different sets of kinetics are demonstrated in Appendix 4 where the lo made. open flame and drop tube data are shown and the wire mesh and TGA results tend t wer o have lower activation energies and would lie across the plot at an angle of 45o. have We found that the choice of one set or the other does not lead to large differ ences an overall in computation but it does slightly change the location of the flame fro nt in CFD conclusions computation. a [3], and The there choiceis ofnow fast a substantial kinetics is body in of accordance evidencewith to support our previous this approach, eg. [12]. This is because in flames the reaction rate is effectively c by the heating-up rate that in turn brings in other difficult to define paramete ontrolled rs such water evaporation as (small in these experiments) and specific heats where there is debate Knowledge factor carbon still in about because of the the ash. itamount values plays The reaction of important an molten char produced conditions coal role particles. from in atdetermining devolatilisation which this devolatilisation the amount is anof important unburned takes place is an important issue. On the basis of our earlier studies [2] these are therefore taken be 105to K/s to a temperature of 1773 K in 150ms, which are typical flame conditio 8 ns.

The yields of volatiles (char is the remainder) for the coals were calculated by using in FGDVC Table the2. coal This composition approach is and significantly the above conditions differentand to the these experimental are given below method by Gibbins and co-workers who have used higher temperatures and longer residence used times with an electrically heated wire mesh [13, 14]. Using this method the rati o ofhigh temperature VM to the laboratory value is typically 1.3 to 1.6, and val the calculated on this basis are given in bold in Table 2. But these higher values a ues re not appropriate because the coal particles ignite after 150 ms and burn in th really e the heterogeneous two methods char (columns combustion 2 andmode. 3 is quite Notwithstanding good) withthat, the exception the agreement of the between Russia n Table Coal Value FG-DVC RWE 35.11 Wood Gascoigne values Experimental Russian South (a) The coalreaction npower Provided VM with 2.43.43 given 55.1 African (daf) Volatile the 55.47 by by rates high Calculated, 58.1 37.64 Dr (a) Yields volatile 57.5-69.5 P of 54.75 48.7 Stephenson, biomass (%wt, (b) 48.9-60.2 content). 58.3 these daf) (b)have for 49.92 CREthe data been Coals at 1350o widely Studied studied C [15] in using the Project TGA or drop t reactors, but there is little direct experimental information on their behaviour ube furnace in flames where the heating-up rate is important. A first order kinetic rat e model has been widely used for all biomass materials. Because of the uncertainty about high their temperaure rate constants we have used values of A= 610133 1/s and E = 2.5108 J/kgmol. The values are based on values previously used for the pf wood combusti and are calculated based on the bio-FG network program [5]. However in this stud on y we have used a single value for all biomass materials including torrefied wood. values The In the devolatisation used for biomass model and particles for coal are areshown assumed in to thebe figure spherical in Appendix and of such 4. a that the temperature is uniform across the particle, ie thermally thin. The kine size expression rate for the biomass (and the coals) is only applicable to particles tic that are may thermally particles thin, fall andoutside it is clear that category. from the electron Whether this micrographs happensthat is determined some biomass by Biot Number [16], which can be based on the surface area to volume ratio. It can the concluded be that the thermally thin regime can be defined as Bi < 0.1, and effecti that means particles of about 600 m. vely 9

The biomass volatile yields used were based on the measured laboratory value of and moisture VM Table Value Biomass RWE Milled PKE Olive Miscanthus Torrefied 2.3 The a 75.71 npower 93.13 Water correction 3. given wood VM Volatile 97.79 wood Evaporation (daf) 80.8 by 86.58.7 79.50 level 75 85.54 84.8 obtained Calculated Yields 80 in 90.91 Model solid for from 76.34 by fuel the using FGBiomass can the Fuels play FG-Biomass a significant program. role in the combustion p In coal it is usually about 3% but in industrial coals stored in the open it may rocess. 10%. reach Freshly harvested biomass contains a significant amount of water, often up 50% althoughof to evaporation non-homogeneous once water nature air-dried, from of the biomass as samples. inpresents thisMoisture project, a number release it of is problems still is therefore about because 5-10 difficult %. the of The t o predict and consists of two parts-evaporation of liquid water and the evaporatio n of drying water Many expression in capillaries. to models allowhave for this. been developed It is assumed and all that computer the progress modelsof contain dryingan is limit ed by the of heat inside the particle and the moisture evaporation rate is effec transport that of an evaporating droplet. The major assumption is that the particle is ass tively umed materials have Many touniform are a temperature anisotropic profile and the across evaporated it (as determined water has an by effect the Biot onNumber) particl e In motion thisand waycauses the heating-up rotation curves and jetfor effects the influence but this of wasmoisture not taken oninto ignition account. dela y can be At calculated the burner exit as shown and prior in Figures to combustion 4a-c. the velocity of the particles is abou t 5-9so a particle travels say 5-9 mm/ms. If the curves calculated in Figure 4a m/s, then the ignition of a 0.5mm diameter particle would be delayed by 10 ms, that i hold s 0.5Of of cm., amounts course water larger wouldparticles, penetrate further, say 1mm in possibly diameter several and containing cm as indicated significant by Fi 4c or larger ultimately resulting in incomplete burnout. gure 10

0.0300.0600.0900.01200.01500.01800.000.20.40.60.81Time 5%water function Figure 4content) (a)moisture (b) of (c) Heating-up Time Plot to Ignition of content Curves for (taken Times for different Biomass for as high thesized Particles end moisture of particles water (0.5 levels, (s)Particel evaporation) and and0.5 1mm temperature diameter mm temperature and as 2200K a ( K) 2200K 00.020.040.060.080.10.124681012Moisture 1500K 0.0300.0600.0900.01200.01500.01800.000.20.40.60.81Time 0.5 mm1 0.5mm00.010.020.030.040.050.060.0741424344454Moisture 0.5mm1500K mm 1mm2200K 0.5mm2200K 1mm(%)Ignition time (s)Particel (s)(%)Ignition temperature time (s) ( K) 2200K 00.020.040.060.080.10.124681012Moisture 1500K 11 0.5 mm1 0.5mm00.010.020.030.040.050.060.0741424344454Moisture 0.5mm1500K mm 1mm2200K 0.5mm2200K 1mm(%)Ignition time (s)(%)Ignition time (s)

These 2.4 NOx have andbeen Sootdescribed Models in previous publications [4] and are effectively the F default models with some adjustment to the approach used. In the case of coal th luent e fuel-N is converted to HCN with the subsequent calculation estimating the amount NO ofand N2 produced depending on the stoichiometry. With biomass the situation is acids more complex which contain becausethe both nitrogen HCN and[7]. NH3 Within are formed Fluent from there the are protein reaction and amino routes vi a Section either NO 3.or For NH3 the and soot notmodel, both and we used the way thethis mechanism is handled outlined hereby isus discussed in reference in [4]. But the issue is the amount of soot formed-there is an adjustable constantwas Fluent this The done inoverall the analysis single-step below.soot prediction model was used based on the total volatile concentration. The rate of soot formation is given by an empirical equa below tion The equation. choice and the of the combustion soot formation of the soot constant, particles Cs, and was the governed activation by theenergy, Magnussen Es, fo r a number of fuels has been discussed [4, 17]. Previously we chose constants for th e volatiles in these bituminous coals on the basis of their C/H ratio (approximate ly 1.5) as: equivalence ratio exponent, r, is equal to 3; the activation temperature Es/ R is K; combustion soot 2000 formation was taken constant, to be 2.7 Cs,and is the 1.5.Magnussen The stoichiometric constant for coefficient soot combustion for soot as The 4. model was computed with soot-radiation interaction. In this study the value Cs was of 2.5. Combustion There Validation is varied a considerable of a assingle of discussed the Coal, CFD amount later model Gascoigne ofin experimental Section the Wood. RWE npower 3b.dataCombustion available for Testthe Furnace: combustion Gascoigne of Wood and consequently this data was used to test some aspects of the C model. FD (a) models temperature Turbulence There for the is are Models. combustion 1597 a number K and Some ofstudies the issues: Gascoigne results were for Wood undertaken the coal. different The with experimental turbulence different models turbulence furnace are exit gi ED-RNG below ven eddy ED-SKE PDF-SKE pdf It seems model dissipation that --eddy withthe dissipation standard model best agreement with k-epsilon model RNG with turbulence in turbulence terms standard of model-1700 model-1420 the k-epsilon exit temperature K K turbulenceand mode-1650 the positi K on of the maximum of the radiation flux is obtained with the ED SKE model. In order to test the sensitivity of the model to the heat input parameter the heat of combus was increased tion about 12 1% errorby in15%. the heat The computed of combustion temperature leads to in a this 10oC case change was in 1580 temperature. K. Thus

Since there can be variations in the composition and the firing rate of approxim 5%, fluctuations in the order of 50 oC can readily occur. The choice of the turb ately model andand ulence holding) thethe devolatilisation flame shape. The model general also view determines of thethe flame flame in the stability furnace (flame is sh inphotograph own Wood Figure A 13 Figure Coal 5 (b). (a). 5a of Combustion) and General Flame this theAttachment detailed flame View provided of the near-burner Detail Contours by(for RWEstructure of Static Gascoigne npower isin Temperature Wood shown Figure Coal below. 5Combustion) b. (Gascoigne

There is Figure 5 reasonable (c), Photograph agreement of the onFlame the position of flame attachment, ignition, and measurement flame shape.of This thewas flame determined shape and using intensity a densitometer as a function to obtain of the a distance quantitative from flame. the (b) There incident Investigation was The flux experimental expansion to the of furnace angles the data Radiation are available wall, theand Properties same from there asRWE is was of the npower the information general Flame. for this flame available flame shape. onfrom the rela investigations [17, 18] on flame temperatures measurements. This information was ted The compared firstwith issue our iscomputed the choice data. of radiation model and for this investigation we u sedmodel DO the whereas previously we used the P1 model. This was done because this is a However As temperatures Figure 1650oC; 14 more far detailed as 6. more yellow,1580oC; the(for Measured powerful and a temperatures accurate Temperature similar computers blue/green, model there coal, Field are but is not 1450oC. available reasonably more [18, Gascoigne demanding reproduced and good Wood)) this on agreement with computational is as now with shown permission]: our below. first measured time. red, choice.

It was not Information obtained accuracy computed using +/-3%). values. possible wasaprovided Medtherm These to obtain results by heat RWE aare flux npower very shown transducer close onin the match Figure radiant (Has 201, 7shown below fluxes optical in together inrange Figure the 7 furnace with 0.15-5 below theTher m, e number a (a) are the accuracy of factors of the thatemissivity can come into data. play We note and these that in arerecent published work [ different 20] (b) The accuracy values of were the used model to particularly the ones commonly in circumstances accepted andwhere used there by us.are some (c) of Figure (Gascoigne The parameter. from simplifications the The radiative the measured 7. accuracy default Surface Wood): About contribution to case of the value Incident Heat the the and soot furnace flux 1.5 measurements-and the calculation, Radiation of (kg/N-m-s) as computed the geometry a Function soot Measurements one. to was on particularly 7.5, of increasing tested Distance the highest andthe by varying theSoot Computation (m) degree incident the Formation ofradiation soot coincidence formation Constant inc by about reased shape and10%, the maximum and the value location of the moved radiation slightlyflux towards are not the absolutely burner. However correct thewhi could rise from (a) incorrect value being used for the devolatilisation, (b) inc ch chemical reaction mechanism which does not include the option of free radicals i orrect n the recirculated gas stream igniting the fuel, and (c) incorrect measurement of the flux near the burner; there is some evidence that similar effects have been obse radiant rved in measurements in other research institutions possibly due to the complex furnace 15

considered geometry where further the because burner enters recentthe measurements furnace. These of coal calculations particle emissivity are being indic that there is a significant change as the particle moves from the flame front to ate main 3. On the Co-Combustion the part basis of of thethe combustor of Coal and studies [30]. with Biomass the Gascoigne in the Combustion Wood coal Test theFacility. model was used for co-firing studies, but there are additional factors in the case of biomass relat the ing The nature (a) to of the Heating-up the biomass Equations. particles. The equations used are those described in the Flue manual where there are problems due to the choice of Cp and the fact that the bi nt has equations omass (b) The a different The Aerodynamic used emissivity are Equations the standard to coal. Fluent equations but the large biomass parti sizes present a problem. The drag coefficients, CD, are as in equation (1) and s cle factors hape CD b1 b2 b3 b4 b1,b2,b3 =2.3288-6.4581f+2.4486f2 = =1.4681+12.258f-20.7322f2+15.8855f3 (24/Re) 0.0964+0.5565f 4.9050-13.8944f+18.4222f2-10.2599f3 were and b4 used (1 + b1 are toexpressed (3) account ) + [b3(2) for Re/(b4 in terms the+irregular Re)] (5) of (4) the (1)shape biomass factor, particles f, which shapes. is defined as far ratio particle As theof asand the thethe surface combustion actual area surface of ofthe an large equivalent area of particles thesphere particle. they having seemthe to generally same volume follow as the th e same laws particles as coal hadparticles. the expected Examination ratio of the using radius videoof images the flame showed zone/radius that suspended of the p Examination of this radius ratio provides confirmation of the amount of volatile article. s associated with the devolatilisation of the biomass particles. Likewise examinat ion video the of gives confirmation of the burning times. Some jetting of material took place at the ends of the particles which could cause tumbling in a flame but the was 16 effect considered not to be a major factor and not included in the calculations.

In these Figure 7.flames Examination it is possible of the Combustion to quantify of the a Single ratioParticle of the flame of Wood zone radius to particle the radius (rc/rp) and make conclusions about the rate of evolution of the and indeed the NO and smoke release. This will be followed up in a future projec volatiles t (c) Comparisons under Flame SuperGen-Bioenergy. Shapes. could be made of the computed flame shapes with experimental studies . A typical flame is shown below and it has a well-defined flame shape with the fl . ame edge diverging from the centerline at an angle of 10o In principle it is possibl e to quantify these flame parameters but there has not been an opportunity to do this any Figure Generally to extent 8. biomass Experimental in thisflames project. Furnace are attached Flame Shape more firmly to the burner and have a large r flame divergence angle, both consistent with the higher reactivity of the biomas s and the (d Generally reasonable torrefied 17 )larger NOx Calculations woods wood quantity assumption the contain same offor volatiles about assumption Biomass. miscanthus 70-90% produced. istoo. protein made. InIn anthat the amino case case acids of PKE, we and have Olive this followed waste is a the and

HCN andpreviously method NH3 routesused is 1:3. by us Effectively [7] and assumed in thethat casethe of a ratio 20% biofuel of NO produced the actual via N is 10% higherwere Ox Computations thancarried if thison route the basis is notdescribed taken into and account. the results are given in th e Generally Table below. the computed results are in reasonable agreement with the experimental results. There are a number of factors that influence the accuracy of the experi results. These mental (typically 5% although include the C-in-ash experimental values errors are more for the like individual 20%), butmeasurements a major facto r is the experiments are made under conditions where the furnace is being used t that o operating undertake under a large steady number state of commercial temperature data conditions. collection Therefore operations theand input is data not fo r the CFD calculation does not exactly match the conditions under which the experiment Table Fuel Measured ppm NO Calc (dry) % Ash, C Mass exit Calculated Exit Experimental Wood Gascoigne 313 Russian Russian/PKE 288 wood 236 (20%) Miscanthus waste 276 (20% Wood* African/Torrefied S. 239 0.3 1724 *Assuming available 4. The validation experimental data in ppm African African/ Discussion 325 319 242 230 Not Combustion were Calc Ash, % T, coal thermal) Not 4.0.62 (15%) KComparison 1.32 coal 5.6 8.9 obtained. the coal coal/ studies 316 5.9 1.94 Olive 3.0 296 0.98 results milled and Test 321 330 1583 1605 325 1664 1646 Conclusions with of 3.5 312 Facility 1.24 1527 wood were 1597 1470 Experimental 1539 4.8 5.87 Gascoigne 3.4 has obtained 1487 6.7 has 1617 a particle 1488 1668 been Wood 1407 although and 1424 modelled coal Computed size reasonable there which using Data. isis a agreement still the some single same coal. as uncertainty with coal. Inthe the abou t the radiation The properties. coal and data. In biomass particular models wegive haveathe satisfactory major charprediction combustionof and the a devolatilisati combustion 18 on

particle model measurements that sizes includes are and necessary. shapes prediction inIt a reasonable of the has been rates possible manner. and yields to Attention deal and with no has experimental complex been drawn particle to th e importance of accurate data on the specific heats of the coals and in particular biomass The predictions of computer the particles. and model it has been run modified usingto three improve coals the and accuracy four coal/biomass of the temperature mixtures , hypothetical a and case of a blend containing torrefied wood. The latter case was st because of the current interest in the combustion of this pre-processed fuel. Th udied e agreement between computed and experimental radiation results is not precise and kinetic this could rate arise through fromto a the number turbulent of issues model ranging used.from It is the our conclusions intention to about revisit the of Computations satisfactory the radiation issue as modelling undertaken were the in using calculations a future the N-model project. of the to unburned compute the carbon-in-ash, NOx were although gr sophistication is required and this is being addressed by current research under eater SuperGen-Bioenergy The themajor issues that project. caused problems are the fact that the research facility wa s designed for industrial testing rather than for an academic study, and therefore extensive academic lacked study. combustion It would diagnostic be an extremely equipmentadvantageous of the type to required run such foraarig detailed using th e Equipment biomass The coal instrumentation Project andfuels. coal-biomass Loan hasBank thatusing provided is blends available information a systematic for furnace to of UKrange research value prediction of tohighly institutions groups forcharacterised using CFD industrial via modelling the power coals EPSRC station and of 5. We Supervisors information design Acknowledgements are purposes. grateful for considerable obtained to BCURA on their for assistance. the Combustion award ofTest We this are facility grant also grateful and atto Didcot. the RWE Industrial npower We are extreme for grateful to Dr Peter Stephenson, Dr Michael Whitehouse and Mr Jason Powis at RWE ly 19 npower for information and helpful discussions.

1. Ma, References Some A. 2. 2000. 3. Arenillas, Tangentially 2196-2203, 4. Modelling Model. R.I. 5. Williams. Combust 6. Predicting Barcelona, L. 7. Fuel, J.M. 8. Modelling-Project through The 9. Model T.G. 10. Canary Combustion Advanced 11. T.H. 12. Expressions. 13. Conference, 17-19 C.K. 14. NOx 20 Williams, Development Prediction Other Backreedy, Jones, Bridgeman, Fletcher, Man, 86, for August Combustion M. J.M. Grass, Inst, an Fuel Coal Pulverised Co-firing Pourkashanian, 1959-1965, CFD J.R.Gibbin, J.R. Solid J.J. 2005 Biomass 2006. of interface Properties, Jones, M. The M. R.I. Fired Sandia 1994. 30, Pulverized Codes. Wheat Research Combustion: Pourkashanian Sensitivity Gibbins, in Pourkashanian Pis L.M. L. and J.M. Effects Fuels, Backreedy, Science 2955-2964, No. Air-Staged Power Particle M. and Pulverised Implementation Straw National Fletcher, Coal 2007. with EPSRC Jones, Fletcher, Pourkashanian 127-ACCCT F.C. Fuel Inc, Proceedings F of J.G Station J.M. Biomass Rubiera, Combustion Alstom and the GR/R02030/01, of Lockwood, Coal L. Depositions I. 2008, New .Witkamp, R. 2005. and Laboratories Coals and Technology, Jones, Willow Combustion Current Ma, L. Coal Shield, Habib, J. Paper Haven, Quality using in A. Power), J.M. Ma, to Prediction M. of Fuel of and an Williams, Using and to be J Pourkashanian, M. Jones, No. an J. Jones, the R., Position. Single Industrial P.T. CT, Williamson, Biomass: in A. Enhance published. on 84, Pourkashanian A. FINAL Advanced (2003). Calculations Zhang, 178, 20 Report Combustion J.M. a Williams, USA. Power Williams. Williams, 2190-2195, Detailed L. PF (linked The of Coals Advanced 763-787, REPORT, Jones, Ma, Furnace, Solid Coal Fuel, Unburned A No. Plants Combustion Coal modelling Proc. and M. A.Williams, Modelling to Sand85-8854, Characterisation Institute, Coal A Torrefaction Fuel to 81, and Pourkashanian Coal October, Combustion 2005. Blends. 2006. Numerical Project Charleston ASME Devolatilisation Fourth Carbon M. 605-618, A. Combustion Qualities Combustion of approach. Test Pourkashanian, Williams, Conference, the Coal Fuel, No.139 2002. Int 28, B. and Model Furnace Sub1985. of SC, 2002. Arias, EPRI 2181, NOx and 84, Reed for Proc USA, for in A. Rate A. a

P.Yan 15. NOx 16. Energy B. 17. Soot Combustion Y 18. Pulverised G. 19. C. 20. published. Emissivity Fired Inst., List 1. Coal Research September 7th 2. Moisture Journal L. 3. Williams, submitted applications, Y.B. 4. M. Biomass, 21 Cahill Simmons, Lu, Lou, Ma, Pourkashanian Burners. Formation and of with Yang, Concurrent Boiler and Y. 31, M. H-C Publications of Biomass and Content Energy and Yan, Biomass 2006 Pourkashanian, Gharebaghi, Modelling to Fuels, Coal and 2771, the V.N. B. Zhou Results furnaces its the S.J.Vollander, Cardiff and for Flame, S. Radiative Energy Flames. Sharifi and Measurements P-F 2007 Applications, on 7th and 22, Cornwell, in A. Benzene, Flame the Methods Fuels, of A Yu Industrial European A. 306-316, Williams, 71, by Institute, R. University, Williams, Devolatilisation DTI Ignition and Williams, Properties Image J.M. Porter, 219-232, Characteristics 22, Toluene Project M. Z-W For J. Studies of Jones Conference 2008. Whitehouse processing University Swithenbank, 306-316, AM. Jiang, Furnaces, Co-Fired Combustion temperature of 2008 Shock 2008 Combustion 1987. No of and Biomass Pourkashanian, onParticulate 118, Measurements Power Tube Toluene/n-Heptane A. 2008. Tests, Pulverised on and 6th of Williams of Final Particles, Coal L. Investigation and Visible Kent, of Station G. Stability, European aa Ma, Report Single Soot Riley. Research Report. Single Media Canterbury, J L.I. of A. A Fuel Radiation. M Grade Concentration The No the Co-firing Conference To Jones Particle Impact Darvell, Fuel in Furnaces, Particle PGB Mixtures, Effect be and of Flame Coals Pulverised-Coalpublished, and 48. the 2007, its of Kent Proc. J. Pulverised of A. J.M. for CRE Rate Co-firing on of Paper M. Biomass, of to 5th Coal Low 1989. Combust. Jones, Jones, of be

The Appendix Details A 1.1. inside Burner of 1 of Details the Face the 0.5furnace of Doosan MW Details the is Burner Babcock shown and below Furnace Mark indicating III burnerthe arepositions shown below. of the ports u sed the A 22 1.2. 1.3. forDiagrammatic radiation Furnacemeasurements. Geometry Representation of the Doosan Babcock Mark III Burner

devolatilisation Appendix The models 2.have Details been model of described the Char based onin Model FG-DVC earlier [2]and publications the char model particularly [4], and the these also give details of the nomenclature used. The char model used here includes a numbe r The (dmp/dt) dp2 Yox)/Mw, t where dp char p environment T local Yox of extra time conventional [(pR particle mass = ox] features T gas fraction [D0 diameter density temperature char /(D0+)] and ofcombustion these oxidant (6) arespecies indicated model in is:gas below. surrounding combusting char partic Mw, le molecular diffusion D0 kinetic ox rate weight rate incorporating coefficient of oxidanteffects speciesof inchemical gas surrounding reactioncombusting on internal particle surface of c har particle (intrinsic Non-default 1. diffusion limited reaction) valuesrate were and constant, selected pore diffusion foe intrinsic kinetics limited char rate model pre-exponential parameters [4]: factor massan d combustion Inclusion 2. mean pore of radius, rate a variable to improve/optimise char particle surface the accuracy area scaling of the calculated factor, fscal char(7) (8) the represents within intrinsic thechar specific burning internal rate expression surface area (6), ofvia theuser-subroutines, char particle andwhere ki repres Ap, ents intrinsic (dp/6) 432 (7) 85.33 Inclusion 3. = fscal Ap,the BC ip +of (-28.44 char ki 28.44 a coal particle BC BC+maceral +1) 85.33 reactivity. (8) BC correction = factor, fmac, as described previously [4] , within the char burning rate expression (5), via user-subroutines. Thus the char Inclusion 4. burning isof determined a simplified by the thermal maceral annealing composition. correction factor, fann (9), within char[(pR the carbon fann Thus 5. represented is (dmp/dt) dp2 Yox)/Mw, 23 burning = we content 0.9 have =T ox] if fann by rate simulation [D0 {70 of (7) fscal /(D0+)] expression the %wt and char < (8). Cp (9) of particle. < of (5), 90 char %wt, via combustion 20 user-subroutines, m < dp for < different 100 m, where Tpcoals >Cp 1400oC} represents where(9) the

Appendix Coal Inherent Ash Table Biomass moisture ash VM FC C H N S K Cl GCV NCV *Torrefaction 24 63.99 4.21 1.36 1.112 44.91 5.70 2.43 0.22 0.4 27.48 50.78 0.05 80.26 6.62 17.74 3.9 MJ/kg Gascoigne A0.13 1.54 5.92 1.08 3.1. 65.42 0.44 3.2. PKE 7.96 20.8 46.42 318.72 moisture 36.19 9.22 64.83 12.15 26.47 25.44 Coal 1.56 Olive 5.34 0.32 0.005 Coal 0.48 Biomass 2.4 12.05 67.01 6.41 47.51 temperature-563 Wood 30.01 77.76 and 12.15 26.83 18.60 5.45 0.50 Properties 4.00 waste 0.08 7.79 16.56 Biomass Russian 42.81 Properties 69.57 26.75 8 0.7 18.61 5.56 Milled 8.42 2071 Properties South 5.47 24.93 (%as 1.0 Kwood (%as received) African 21.8 Miscanthus received) Torrefied wood*

The plot below Appendix 4. Summary showsof the Kinetic variety Analysis of data available for kinetic studies. The cent black area shows data from wire mesh (mainly for coal) and from TGA studies (bot ral h laboratory for coal and flames for biomass). but without Higher surrounding rates are radiating obtainedwalls. from drop The highest tubes and rates openare applicable to high intensity flames in radiant furnaces and these are used here on theBiomass. based Figure Coal. 25 assumption A 4. Plot of that Rate these Constants simulate as a real Function furnace offlames.. temperature

Potrebbero piacerti anche