Study of the evolution of particle size distributions and its effects on the oxidation of pulverized coal

Study of the evolution of particle size distributions and its effects on the oxidation of pulverized coal
of 13
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Combustion and Flame 151 (2007) 482–494www.elsevier.com/locate/combustflame Study of the evolution of particle size distributions andits effects on the oxidation of pulverized coal Santiago Jiménez a , ∗ , Javier Ballester a , b a  LITEC-CSIC (Spanish Council for Scientific Research), María de Luna, 10, 50018, Zaragoza, Spain b Fluid Mechanics Group, University of Zaragoza, María de Luna, 3, 50018, Zaragoza, Spain Received 7 August 2006; received in revised form 16 January 2007; accepted 7 August 2007Available online 12 September 2007 Abstract This paper discusses the factors influencing the evolution of particle size during the combustion of pulverizedcoal, as well as their consequences for the interpretation of burnout curves. A detailed experimental characteriza-tion of the evolution of the particle size distribution (PSD) of a pulverized coal (anthracite) burned under realisticconditions in an entrained flow reactor is presented and used as the reference data for the subsequent analysis. Thedata show evidence for particle fragmentation at relatively short times (or, equivalently, high unburnt fractions).The formation of fragments comparable in size to the parent coal/char particles is modeled with a simple fragmen-tation scheme, which results in an improved reproduction of the PSD’s evolution. The effects of fragmentation onthe burnout curves are then studied in detail. An enhancement of their curvature is observed, which results in abetter fit of the experimental data; in particular, the high conversion range, where the largest discrepancies betweenpredictions and measurements are usually found, is well reproduced with this “extended” model. Simultaneously,the increase of specific surface caused by particle fragmentation causes an increase in the conversion rate, and asmaller total conversion time. To fit the experimental data, new optimal kinetic parameters are calculated. Finally,the potential relevance of fragmentation in the simulation of industrial pf plants is discussed. © 2007 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords:  Coal combustion; Fragmentation; Burnout curves; Diameter evolution 1. Introduction Fragmentation of coal particles during combustionhasreceivedconsiderableattention, especially regard-ing its influence on fly ash formation (in the case of pulverized coal combustion) and coal conversion (influidized bed boilers). It is generally accepted that thisphenomenon does not affect the submicrometer frac-tion of the particulate emissions, for which models for * Corresponding author. Fax: +34 976761882.  E-mail address:  yago@litec.csic.es (S. Jiménez). mineral evaporation/condensation reached a separateexplanation [1,2]. In the range  > 1 µm, comparisonsbetween the particle size distribution (PSD) of thesrcinal coal with that of the resulting fly ash parti-cles showed the need for a morphological mechanismaffecting the burning particles in order to explain therelatively low mean particle size of the ash (comparedto the one expected if all the mineral matter in a coalparticle coalesced in a single ash corpuscle) [3,4].The reasonable agreements obtained between simplebreakup models (that arbitrarily assigned a number of fragments to each srcinal coal particle) and measure- 0010-2180/$ – see front matter  © 2007 The Combustion Institute. Published by Elsevier Inc. All rights reserved.doi:10.1016/j.combustflame.2007.08.001  S. Jiménez, J. Ballester / Combustion and Flame 151 (2007) 482–494  483 ments in laboratory and industrial plants [3,4] sug-gested that fragmentation is a relevant phenomenonin coal combustion.More complex models were proposed in subse-quent studies, including particle attrition [5], particle shedding by rotation caused by tangential liberationof volatiles [6], and percolative fragmentation [5,7– 9]. The latter was first introduced to study the effectof char fragmentation on burnout rate in fluidized bedcombustion (FBC); it was soon adapted to pulver-ized coal combustion (PFC) [10,11], whereas in thiscase the efforts were focused mainly on the process of fly ash formation. The first models consisted mostlyof pure Monte Carlo calculations for individual par-ticles, and subsequent modifications were introducedto account for statistical phenomena affecting the coalPSD in the form of probability functions for parti-cle fragmentation rates, or number and properties of the fragments produced (e.g., [5] for PFC, [12] for FBC). Also, specific fragmentation models have beenproposed for the different coal combustion modes,ranging from the initial “shrinking core” model(e.g., [9,10]) to calculations specifically designed for hollow char particles (cenospheres) [13–15].Thesemodels,withparametersconvenientlyadap-ted to each coal, are in reasonable agreement withsome experimental measurements available for pul-verized coal combustion, essentially regarding fly ashproperties. The experimental results reveal notabledifferences in the behavior of different coals and ini-tial coal particle sizes, the number of fragments vary-ing in a typical range of 2–100 (e.g., [3,13]). More recently, the existence of a third fine mode ( ∼ 1 µm)in ash fromcoalcombustion has been reported and as-cribed also to fragmentation through several possiblemechanisms [16,17].In contrast, to the authors’ knowledge, little di-rect, detailed information is available on the evolu-tion of the mass-based PSD during the initial com-bustion stages of pulverized coal. Some on-line tech-niques give number-based PSDs [5], more appropri- ate to characterize the relatively fine ash particlesthan to study the effect of char “breakage” into afew fragments comparable in size to the parent charparticle. Liu et al. [14] reported results for a bitumi-nous coal char (initial particle size  ∼ 100 µm), ob-tained in a manner similar to that used in the presentwork, supporting significant particle fragmentationat relatively low burnouts. As an exception, Hurtand Davis [18] did not observe significant particlebreakup events in experiments with individual parti-cles, belonging to several coal types and having sizes50–300 µm, embedded in an inert fiber mat, and theyfound some evidence of spontaneous agglomerationof some fragments produced in late stages of com-bustion.Although it is generally accepted that char frag-mentation can significantly affect coal conversionrates (see comment 3 and reply in [5] for PFC;[9,12] for FBC), this effect has not been widely incor-porated into the simulations performed for real com-bustion chambers or even for laboratory tests fromwhich the kinetic parameters of coal are derived. Thegeneral aim of the present paper is to gain more in-sight into the occurrence of fragmentation and its roleon the evolution of burning particles and, more gener-ally, the implications for the simulation of pulverizedcoal combustion.Ballester and Jiménez [19] recently showed theneed to consider the whole PSD (even in the caseof sieved, narrow distributions) in the derivation of the kinetic parameters from EFR tests. Following thecommon approach in the field, in that work each par-ticle was assumed to evolve independently (except fortheir contribution to oxygen consumption), its diame-ter ( D ) varying with the unburnt fraction ( U  ) accord-ing to the classical expression due to Smith [20,21] D = D 0 U  α . The parameter α  was obtained by adjust-ing the evolution of the PSD mean diameter. The evo-lution of the whole distribution is studied with moredetail in this paper. In the first place, some simpli-fications inherent to the previous approach and theirconsequences on the calculated PSDs are discussed.In the second place, the effects of fragmentation phe-nomena on the burnout curves and on the evolutionof the PSD are analyzed using a modified combustionmodel. A complete set of experimental PSD measure-ments performed with a Spanish anthracite is pre-sented and used as the reference data. 2. Experimental 2.1. Facility and instrumentation The tests were performed in an entrained flow re-actor (EFR, thereafter). It has been described in moredetail previously [19] and its characteristics will onlybe briefly outlined here. It consists of an externallyheated 1.6-m-long SiC tube, into which fuel particlesare injected pneumatically through an injection gun,whose tip can be located at any point along the up-per half of the tube. Co-flow gases are combustionproducts supplied from a natural gas burner, which si-multaneously preheat gases and serve to control theoxygen concentration in the reactor.Solid samples are collected by means of an isoki-netic particle-sampling probe, which can be insertedthrough the outlet section to any location along thelower half of the tube. The particles are finally re-tained in a sintered-bronze filter. The distance be-tween injection gun and probe (or, equivalently, the  484  S. Jiménez, J. Ballester / Combustion and Flame 151 (2007) 482–494 residence time of the particles in the combustion at-mosphere) can thus be varied within a wide range(typically 0.2–4 s for oxidation tests, and 0.05–1 s fordevolatilization studies).The srcinal fuel and the char particles collectedhave been analyzed to obtain their physical-chemicalproperties:– Ultimate and proximate (ASTM) analyses of thefuel.– Thecombustiblefractionineverysamplewasde-termined according to the ash-as-tracer method.The thermogravimetric method developed byMayoral el al. [22] was followed due to the smallamount of sample collected.– Selected samples were observed in a scanningelectron microscope (SEM) to determine the evo-lution of the particle morphology.– Particle size distributions have been measured bylaser diffractometry (Malvern Mastersizer S). Inorder to assure a correct size characterization,suspensions were dispersed by ultrasonication ina bath sonicator (40 kHz, 50 W) immediatelybefore wet measurement to break up particle ag-gregates. In prior measurements with pulverizedcoal samples, increasing sonication times weretested; the effect of the ultrasounds was evidentfrom the firsts seconds of application, and rapidlysaturated, so that over 30 s no differences wereobserved and the measurements were repeatable.As a consequence, sonication periods of approx-imately 30 s were used in this study. 2.2. Combustion experiments The main set of tests refers to experiments per-formed with a Spanish anthracite, whose results con-cerning burnout curves served as a reference for dis-cussion in a previous work  [19]. The pulverized coal was thoroughly sieved in the range 53–63 µm. Fivecombustion conditions were used: four different re-actor ( =  gas and wall) temperatures (1040, 1175,1300, and 1450 ◦ C) with a constant oxygen concen-tration at the reactor’s exit (4% in a molar, dry ba-sis), and an additional series at 1300 ◦ C and 8% O 2 .In all cases, the oxygen concentration at the fuel in- jection point was  ∼ 3% higher, the difference beingconsumed in the combustion of the particles alongthe tube. A highly diluted flow was obtained (par-ticle loading ratios  ∼ 0.02 and  ∼ 3 × 10 − 6 in massand volume bases, respectively), so that particle-to-particle interactions were negligible. On the otherhand, the calculated critical velocity is a few cen-timeters per second, one order of magnitude smallerthan the gas velocity, thus ensuring entrained flowconditions. These conditions are thought to ensure a Table 1Proximate and ultimate ASTM analysis of the fuels (as fired,by weight)Anthracite CokeMoisture (%) 1 . 46 0 . 65Ash (%) 19 . 17 1 . 26Volatiles (%) 10 . 28 11 . 88Fixed carbon (%) 69 . 09 86 . 20C (%) 70 . 3 87 . 30H (%) 3 . 03 3 . 82N (%) 1 . 63 2 . 52S (%) 2 . 28 4 . 36HHV (MJ / kg) 27 . 59 35 . 6 uniform combustion history of all particles along thereactor.In addition to these experiments, results from an-other series performed with sieved (53–63 µm) pe-troleum coke at 1300 ◦ C and 4% O 2  at the reactor’sexit will also be shown. Table 1 presents the proxi-mate and ultimate ASTM analysis of both fuels. In allthe series, seven char samples were collected for eachrun, with distances between injection gun and sam-pling probe from 0.2 to 1.6 m.Thecalculatedheatingrateoftheparticlesinjectedinto the EFR is ∼ 10 5 K / s, which is representative of pfsystems[23].Theoxygenconcentrationswerecho- sen in [19] as representative of the longest stage alongthe combustion history of the particles, i.e., char ox-idation after the flame, in order to derive the kineticparameters for char oxidation [19,24]. Important vari-ations in this respect are found among other fragmen-tation studies: the use of similar [13] and somewhathigher concentrations [5] has been reported; in othercases high oxygen concentrations were used (e.g.,from 10% up to 80% O 2  in [6]) in order to achievehigher particle temperatures. To the authors’ knowl-edge, there is a lack of detailed data on the evolutionof the PSDs in industrial or laboratory facilities forconcluding likely effects of stoichiometry or particleconcentration on fragmentation. 2.3. Experimental results Fig. 1 shows the burnout curves obtained for theanthracite in the EFR, in terms of unburnt fraction( U  , fraction of the srcinal combustible matter re-maining in the particle) as a function of the distancebetween injection gun and sampling probe ( L ). Theslightly different slip velocities for the various parti-cle sizes prevent a direct translation of these curves todependences on residence time of the particles in thereactor.The evolution of the PSD was measured for threeoftheseseries.Figs.2–4displaytheresults,comparedin all cases to the PSD of the srcinal sieved coal.  S. Jiménez, J. Ballester / Combustion and Flame 151 (2007) 482–494  485Fig. 1. Unburnt fraction of the coal along the EFR length for the five combustion conditions tested.Fig. 2. Measured evolution of the coal/char PSD with the unburnt fraction  U  . Combustion conditions 1300 ◦ C, 4% O 2  in fluegases. In these graphs, the  Y  -axis represents the percentage of volume contained in a narrow interval around diameter  D i , with D i + 1 /D i  ≈ 1 . 16. The general aspect of these graphs is very similar: be-sides a progressive shift of the peak toward smallerdiameters, the width of the distribution increases withthe degree of burnout (since the distributions are nor-malized, this is more clearly represented by a lower-ing of the mode’s height). With respect to the src-inal coal distribution, the fraction of particles in therange 20–40 µm displays a fast rise, even for rela-tively high  U  s. Except for the curves correspondingto very low  U   levels ( < 0.07), a clear gradation of thedistributions with the degree of burnout is apparentin Figs. 2–4.The presence of particles in the range 20–40 µmat high  U   levels and their absence within the srcinalsample are apparent also in the SEM pictures shownin Fig. 5. Incidentally, Fig. 5a also shows the pres- ence of small particles apparently adhering to the sur-face of the large ( ∼ 53–63 µm) particles, despite thesample having been sieved repetitively (four times).These small particles were, however, detected in themeasurements performed by diffractometry as a sec-ondary mode  < 10 µm in Figs. 2–4 and are not relatedto the changes in the intervals ∼ 20–40 µm.The PSD evolution is markedly different for thecombustion of petroleum coke sieved in the same sizerange (Fig. 6). In this case, the mode “shifts” toward smaller diameters without significant modification of its shape for  U >  0 . 3. A progressive increase of themode’s width, together with the appearance of a no-ticeable mode below 10 µm, is observed afterward.  486  S. Jiménez, J. Ballester / Combustion and Flame 151 (2007) 482–494 Fig. 3. Measured evolution of the coal/char PSD with the unburnt fraction  U  . Combustion conditions 1450 ◦ C, 4% O 2  in fluegases.Fig. 4. Measured evolution of the coal/char PSD with the unburnt fraction  U  . Combustion conditions 1300 ◦ C, 8% O 2  in fluegases. 3. Evolution of particle size distributions 3.1. An apparent paradox  The experimental data shown in Figs. 1–4 wereused to characterize the reactivity of the anthracite, byderiving the parameters that best describe its oxida-tion behavior. The combustion model is derived fromthe early work of Field et al. [25] and will only besummarized here; a more detailed description can befound in that reference or in [19]. The model consid- ers independently coal devolatilization and char ox-idation; a single-step model is adopted for the firstprocess, whereas char oxidation is modeled by an ap-parent kinetics based on the outer surface of the charparticles. Two parameters,  A c  and  E c , respectivelythe pre-exponential factor (kg / (m 2 sPa) for reactionorder 1) and the activation energy (J / mol), combinedin an Arrhenius-type expression, are used to describethe oxidation rate.Theevolution ofcoalparticle diameter anddensityduring combustion was modeled in [19] according tothe expressions, proposed by Smith [20,21],(1) D = D 0 U  α  , (2) ρ = ρ 0 U  β  , where 3 α  + β  = 1 for a spherical particle. AlthoughEqs. (1) and (2) can adequately fit the experimen-
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks