文档库

最新最全的文档下载
当前位置:文档库 > Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Journal of Analytical and Applied Pyrolysis 114(2015)22–31

Contents lists available at ScienceDirect

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Journal of Analytical and Applied

Pyrolysis

j o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /j a a

p

Study of the high heating rate devolatilization of a pulverized bituminous coal under oxygen-containing atmospheres

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

R.Lemaire a ,?,C.Bruhier a ,D.Menage a ,b ,E.Therssen c ,P.Seers b

a

Mines Douai,EI,F-59508Douai,France

b

Department of Mechanical Engineering,école de technologie supérieure,Montréal,Québec H3C 1K33,Canada c

Laboratoire PC2A,UMR CNRS 8522,F-59655Villeneuve d’Ascq,France

a r t i c l e i n f o Article history:

Received 24December 2014

Received in revised form 3April 2015Accepted 4April 2015

Available online 18April 2015Keywords:Coal

Oxygen-enriched combustion Flat-?ame reactor Devolatilization Kinetics

Gaseous emissions

a b s t r a c t

This study deals with the analysis of pulverized coal devolatilization in oxygen-containing atmospheres.To do so,a newly developed ?at ?ame reactor allowing coal jet ?ames to be stabilized with fuel heating rates of the order of 106K s ?1has been used to mimic operating conditions close to those met in practical combustors.A high volatile bituminous coal milled in an industrial grinder has been ?uidized by carrier gases containing various amounts of oxygen (air and pure oxygen)and then injected in hot gases (temper-atures ranging up to ~1240K)generated by propane/air ?at ?ames.The thermal history of coal particles has been monitored by coupling particle image velocimetry (PIV)and pyrometric measurements to be integrated in constant-rate,single-kinetic-rate,two-competing-rate and DAEM devolatilization models.Devolatilization pro?les derived from the analysis of the char collected at different residence times have then been compared with simulated data and a simpli?ed ?tting procedure has been implemented to ?nd kinetic parameters leading to reproduce experimental data so as to evaluate the relative in?uence of the surrounding atmosphere on the devolatilization kinetics.Obtained results con?rm that devolatilization is faster and more complete under oxygen enriched environments which has been related to an enhanced combustion of volatiles inducing a rise of the temperature of fuel particles.On the other hand,apparent devolatilization rates appeared to be only indirectly affected by the surrounding atmosphere through the way this one in?uences the heating of the coal.Furthermore,results obtained in this work tend to indicate the absence of overlapping between devolatilization and char oxidation stages.Measurements of gaseous species released during or formed after the devolatilization process ?nally con?rm that the combustion of volatiles is more complete under oxygen enriched combustion (OEC).More CO 2is thus produced while CO concentrations signi?cantly decrease.An enhanced fuel-N conversion has moreover been related to the higher temperatures and oxygen partial pressures met under OEC while SO 2concen-trations ~50%higher have been measured due the higher quantities of fuel-sulfur and oxygen available in the medium.

?2015Elsevier B.V.All rights reserved.

1.Introduction

Coal is currently used to generate around 41%of the electricity in the world with 65%of the global coal demand consumed in the sole power sector in 2010[1].Its combustion in coal-?red power plants however results in the generation of carbon dioxide whose emis-sions account for around 40%of the total quantity of greenhouse gases emitted into the atmosphere [2].To reduce such emissions,

?Corresponding author at:Ecole des Mines de Douai,Département Energé-tique Industrielle 941,rue Charles Bourseul,CS 10838,59508Douai Cedex,France.Tel.:+33327712129;fax:+33327712915.

E-mail address:romain.lemaire@mines-douai.fr (R.Lemaire).identi?ed strategies mainly concern the increase of the ef?ciency of energy production systems and the use of carbon capture and storage (CCS)processes [1].

Among existing CCS techniques,oxycombustion refers to a combustion conducted using a mixture of oxygen and recircu-lated ?ue gases (mainly composed of CO 2)[3,4].An intermediate step between air combustion and full oxycombustion consists in oxygen-enriched combustion (OEC)which is based on the use of air–oxygen mixtures containing elevated O 2concentrations (>21%).OEC is actually commonly used in a wide range of industrial applica-tions including mineral production and metal heating or melting for instance [5].OEC offers a number of advantages in such applications including higher ?ame temperatures,increased heat releases and boiler thermal ef?ciency,fuel saving and increased processing rates

http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.html/10.1016/j.jaap.2015.04.008

0165-2370/?2015Elsevier B.V.All rights reserved.

R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–3123

Nomenclature

A char Dry mass fraction of ash in the char[?]

A coal Dry mass fraction of ash in the raw coal[?]

dt Time step for the devolatilization modeling[s]

E Activation energy[J mol?1]

E i Activation energy of the i th reaction[J mol?1]

E0Central value of the Gaussian distribution of the acti-vation energy[J mol?1]

F d Devolatilized mass fraction(expressed as a percent-

age of the original coal)[wt%(DAF)]

f(E)Gaussian distribution function of the activation energy[?]

f0Volatile fraction contained in the coal(on dry ash free basis)[?]

k Rate constant[s?1]

k0Pre-exponential factor of the Arrhenius equation [s?1]

M Char residual mass[kg]

M dev Devolatilized mass[kg]

M0Initial mass of the coal particles[kg]

M inf Mass of volatiles released for an in?nite time[kg]

R Ideal gas constant[J mol?1K?1]

T p Coal particle temperature[K]

T g Temperature of hot gases[K]

V Mass fraction of volatiles released for a given time t (expressed as a percentage of the original coal)[wt%

(DAF)]

V inf Mass fraction of volatiles released for an in?nite time(expressed as a percentage of the original coal)

[wt%(DAF)]

VM0Volatile matter content of the coal obtained from the proximate analysis[wt%(DAF)]

X Volatile fraction[?]

?i Mass stoichiometric coef?cient of the i th reaction [?]

Standard deviation of the activation energy Gauss-ian distribution[J mol?1]

Equivalence ratio[?]

Acronyms

C Coal

CCS Carbon Capture and Storage

CG Carrier Gas

DAF Dry Ash Free

EFR Entrained Flow Reactor

FF1Flat Flame1(propane/air =1.06)

FF2Flat Flame2(propane/air =1.00)

FFR Flat Flame Reactor

HAB Height Above the Burner

OEC Oxygen Enriched Combustion

PC Pulverized Coal

PIV Particle Image Velocimetry

R c Residual char

VM Volatile Matters

VM i Volatile Matters issued from the i th reaction

[5].Since OEC basically involves removing nitrogen from the oxi-dizer,the?ue gas volumes produced are therefore reduced while their CO2fraction increases thus favoring its post-combustion low-cost capture[5,6].A recirculation of burnt gases(mainly composed of CO2and N2)can moreover be implemented to increase pollut-ant concentrations(and ease their treatment)and/or to manage furnace temperatures which in turn leads OEC to be considered as a retro?t option for traditional pulverized coal(PC)power plants as well as an interesting compromise between conventional air-?ring combustion and emerging oxy-fuel combustion systems[6].

This being said,the use of oxygen-enriched atmospheres inevitably affects the ignition and combustion characteristics of pulverized coal particles in addition to pollutant emissions result-ing from volatile oxidation and char burnout(see[4,5]for instance and references therein).A thorough understanding of the impact of OEC on the physical–chemical mechanisms involved in the main steps of the thermochemical conversion of pulverized coal is thus needed to analyze the main trends observed during pilot-scale tests as well as to predict scale-up performance through CFD modeling. Devolatilization is the?rst step in the coal conversion in most com-mon PC boilers.Such a process is consequently of prior importance as it directly in?uences the subsequent stages.It therefore needs to be well predicted to be able to achieve consistent simulation of PC ?ames.Different works have thus been undertaken to better under-stand the in?uence of the oxygen content of the oxidizer stream on coal devolatilization.

Timothy et al.especially measured average coal particle tem-peratures between700and800?C higher when burning lignite and bituminous coals(classi?ed in38–45and90–105micron size ranges)at1700K in a laminar?ow furnace using atmospheres con-taining100%of oxygen instead of15%[7].They also estimated devolatilization times and particle burnout durations between 4and5times lower in100%O2than in air.Finally,predicted devolatilization yields passing from60to80%in15and100% O2,respectively,were reported[7].Similar increases of particle temperatures along with decreases of char and volatile burnout times(depending on the oxygen concentration into the medium) have been reported since then during different drop tube furnace studies[8–10].As far as devolatilization is concerned more specif-ically,Murphy and Shaddix reported an apparent enhancement of devolatilization rates of subbituminous and bituminous coals with increasing oxygen concentrations(from6mol%to36mol%)in a combustion-driven entrained?ow reactor(EFR)fueled with parti-cles ground and sieved into a106–125-?m size fraction[11].These authors especially related this observation to the combination of two factors including a closer proximity of the volatile?ame to the coal particles and the higher temperature of such a?ame due to the oxygen-enhanced oxidation of devolatilized matters.Molina and Shaddix[12]then conducted additional experiments in the San-dia’s char kinetic EFR with gas temperatures of around1250K and concluded that increasing the oxygen concentration(from21%to 30%)in N2accelerates the particle ignition.No clear-cut conclusions were drawn,however,concerning a reduction of the devolatiliza-tion time which contrasts somewhat with the results obtained by the same authors when analyzing the ignition and devolatilization characteristics of bituminous and subbituminous coals at a gas tem-perature of1700K and oxygen fractions varying from12to36vol% [13].A reduction of around59%of the devolatilization time was then observed when increasing the O2concentration in the pro-portions indicated above.This trend has been related,as in[11],to an increase of the mass?ux of oxygen to the volatile?ame which thereby induces enhanced devolatilization rates.Such explanations are consistent with the conclusions issued from a modeling work from Cho et al.who investigated the effects of gas temperature and oxygen concentration,among others,on the volatile matter releases and char burning rates[14].They demonstrated that the greater the O2concentration,the shorter the devolatilization time. By analyzing in detail the results issued from their simulations,Cho et al.indicated that the increase of the devolatilization rate leads to a decrease of the oxygen mass fraction around the fuel particles. These authors argued that the volatiles ejected from the coal par-ticles push away the oxygen and thus generate a depletion zone around the fuel.According to them,this depletion zone is likely to

24R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–31

be accentuated by the consumption of the supplied oxygen from the bulk?ow in the high temperature?ame surrounding the par-ticles(such a?ame resulting from the combustion of the ejected volatile matters in the oxygen enriched medium).Finally,Cho et al. explained that the char oxidation begins after the devolatilization process.This analysis may however be somewhat contrasted by the conclusions drawn in a previous modeling work from Gurgel Veras et al.who noted that a possible overlapping of the devolatilization and char oxidation stages could occur depending on parameters such as the particle size and the oxygen concentration of the reac-tion medium[15].These authors especially noted that the oxygen content at the surface of particles having diameters of a few tens of microns(<80?m)can signi?cantly exceed zero during a low temperature devolatilization process(<1400K)despite the?ux of volatiles released from the surface of the fuel particles.The increase of the oxygen diffusion would then enhance the coal oxidation and therefore contribute to accelerate the loss of volatiles.

Though the main trends reported in the literature indicate that an increase of the oxygen concentration induces a rise of the devolatilization rates,some aspects of the thermochemical mechanisms involved in such a conversion process appear to be not fully elucidated yet.In addition,fundamental works coupling experimental analyses and kinetic modeling of the devolatilization process are still fundamental in order to support and validate simu-lations of full-scale PC boilers.That said,analysis and confrontation of data issued from the literature can be made dif?cult due to the variability of the experimental conditions used during lab-scale and/or pilot-scale experiments.Such parameters inevitably affect the obtained results depending on factors such as the fuel heating rate or the particle size and residence time as recalled in a recent work from Brix et al.dealing with coal OEC in N2-and CO2-based atmospheres[16].It is therefore of importance to use or develop experimental facilities that mimic some speci?c characteristics of the coal combustion under industrial combustors to obtain broad consistent datasets that are likely to promote the understanding of coal OEC under practical conditions.To be able to investigate coal devolatilization with high heating rates typical of those met in blast furnace and pulverized coal power plant applications this paper focuses on(i.e.,>105K s?1[17]),we recently designed a?at?ame reactor(FFR)based on the use of a hybrid McKenna burner similar to the one previously developed in[18]to investigate the combustion of liquid fuels.Such a type of facility indeed appears to be particu-larly relevant to reproduce the thermal conditions prevailing in the near-burner zone of industrial boilers[19].This newly developed device(described and characterized in detail in next section)allows studying the devolatilization of coal particles(having an industrial size-range)with heating rates of the order of106K s?1and maximal gas temperatures?xed to~1240K in the present work.Such exper-imental conditions have been only seldom studied in the literature despite the extensive need of experimental data obtained over a wide range of conditions(including the high-heating rate low-temperature devolatilization regime)to validate kinetic models thus justifying the interest of the experiments conducted here.The thermal history of coal particles has been monitored by coupling particle image velocimetry(PIV)and pyrometric measurements. Devolatilization pro?les derived from the analysis of the char col-lected at different residence times have then been analyzed and confronted with data issued from the use of4different types of empirical models.To the best of the authors’knowledge,experi-mental devolatilization pro?les thoroughly determined under such speci?c conditions and then simulated with various kinetic models are relatively scarce.Finally,even if reductions of CO concentrations [5]and increases of CO2emissions[5,20]are generally reported under OEC due to a more complete combustion,trends regarding species like NO x are more dif?cult to predict as the formation of such a pollutant depends on numerous inter-connected factors including devolatilization yields as well as local temperatures and equivalence ratios among others.Consequently,to complete the analysis of devolatilization kinetics proposed herein,the impact of the oxygen enrichment on CO,CO2,NO and SO2emissions has also been studied.

2.Experimental setup and methodology

2.1.Description of the experimental setup

The experimental setup is made up of three distinct parts:a fuel feeding system based on the use of an acoustic sower,a hybrid?at ?ame burner and a sampling line connected to cyclones and gas analyzers.The whole system is represented in Fig.1.

The coal feeding system is composed of a tank connected to a motorized sprocket wheel.The rotation speed of the sprocket

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

wheel Fig.1.Diagram of the experimental setup(a)and picture of a coal jet?ame(b).

R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–3125

Table1

Coal properties.

Moisture* (wt%)Volatile matter**

(wt%)

Ash***

(wt%)

C***

(wt%)

H***

(wt%)

O***

(wt%)

N***

(wt%)

S***

(wt%)

HHV***

(kJ kg?1)

1.635.9 4.88

2.1 4.9 6.560.810.8333145

*As received.

**Dry-ash-free basis.

***Dry basis.

can be adjusted and monitored to allow the injection of a controlled and constant quantity of coal particles into the acoustic sower.A mass?ow rate of12g h?1has been?xed for all the experiments reported in this paper.The acoustic sower that has been designed to feed the burner with pulverized coal particles consists of a glass-made enclosure positioned on a thin vibrating membrane driven by a loudspeaker whose frequency is adjusted by a frequency gen-erator.A cloud of particles is thus formed on the membrane as coal falls from the sprocket wheel.This particle cloud is then?uidized and pneumatically transported to the?at-?ame burner by carrier gases containing various amounts of oxygen.Air and pure oxygen have been used in the present work with a constant?ow rate of 1l min?1.The carrier gas is injected in the glass enclosure through four lateral inlets whose positions have been optimized by numer-ical simulations in order to achieve the best quality mixing of coal particles with the gas,to avoid fuel accumulation on the vibrating membrane and to maintain a constant mass?ow-rate of coal at the sower outlet[21].

The hybrid burner used to ignite the coal particles is based on an atmospheric McKenna?at?ame burner(manufactured by Holthuis &Associates).It is composed of a60-mm diameter bronze porous plug with a central6.35-mm diameter tube allowing the introduc-tion of a fuel injector(a1.75-mm internal diameter stainless-steel tube).Propane–air mixtures have been used to stabilize premixed ?at?ames on the porous plug.The burnt gases issued from these ?at?ames heat the coal particles transported through the central injector resulting in the generation of pulverized coal jet?ames of approximately35cm high and3cm wide.In the present work, two different propane–air mixtures have been used.The?rst one (FF1; =1.06)has been obtained using0.53l min?1of propane and 11.9l min?1of air in order to reproduce the operating conditions ?xed in a reference work from Therssen et al.[22].The second ?at?ame(FF2; =1.00)has been stabilized using0.65l min?1of propane and15.4l min?1of air.In this case,more propane is burnt to increase the temperature into the combustion chamber.The obtained?ames are isolated from the surrounding atmosphere and from ambient perturbations by a40-cm high quartz chimney that is sheltered by a stainless-steel cover to avoid thermal losses and to keep a high temperature into the reactor(adapted accesses being built into the stainless-steel cover to allow the implementation of optical diagnostics).

The sampling system positioned above the FFR allows char sampling and combustion gas analysis.It is made of a51-cm high quartz collector(based on a system similar to the one previously used in[23]for soot sampling)made of two concentric tubes.Nitro-gen is introduced into the annulus area and injected through the internal tube by the means of four holes placed a few millime-ters downstream of the collector inlet.By this way,fast samplings with an immediate quenching of collected species can be operated. Samples are then driven to either a gas analyzer rack or a char col-lecting system composed of heated cyclones.Environnement SA Topaze32M and MIR2M analyzers have been used.Sampled gases are transported to these devices through a heated probe whose temperature is regulated to190?C to avoid water condensation. Then,gas concentrations are measured by non-dispersive infrared analysis with gas correlation?lter(CO2,CO,SO2),chemi-luminescence(NO/NO x)or paramagnetic resonance(O2).The char collecting system is composed of a heated cyclone(Dekati SAC-65) having a cut size of14?m.Its temperature is regulated to180?C. Other species such as soot particles then pass through a secondary ?ltration system where they can be collected for further analysis. The mass?ow rate at the exit of the vacuum pump is?nally con-trolled by a mass?ow controller whose set point can be adjusted as a function of the cold nitrogen?ow sent to the quartz collector. The sampling line is positioned at a?xed distance from the burner while the position of the latter can be adapted thanks to a motor-ized translation stage.By this way,samplings can be achieved at different heights above the burner(HAB)that is to say at different residence times.

2.2.Fuel

A French high volatile bituminous coal from Freyming has been used in this work.The results of its analyses are reported in Table1. The particle size distribution of this coal issued from an industrial grinder is characterized by a measured mean diameter of107?m; the Rosin–Rammler parameters being X0=0.136mm and n=2.57.

2.3.Residence time and temperature?elds characterization

A complete characterization of velocity and temperature?elds has been carried out above the?at?ame burner.Such a procedure is required to be able to derive consistent kinetic data from the analyses conducted in the studied?ames.

The velocity and temperature?elds have been?rst analyzed without injecting coal particles in the central carrier-gas jet (non-reactive conditions).One can examine this way in which envi-ronment the fuel will be introduced.The speed of the particles transported by the carrier-gas jet passing through the?at?ame burnt gases has been estimated using particle image velocimetry (PIV).Alumina spheres were used as seeding particles in this case with a volume fraction of0.02%into the carrier-gas.This is equal to the volume fraction of the fuel when stabilizing coal jet?ames.The size of the particles has been adjusted to ensure that the velocity of alumina and coal particles could be considered as identical during the residence time range investigated here;the particle fall speeds being similar and signi?cantly lower than the?ow characteristic speed.A dual cavity Nd:YAG New Wave Research laser(Solo120XT) generating laser sheets at532nm has been used.It has been cou-pled with a CCD FlowSense camera(2048×2048)equipped with a532nm band-pass?lter positioned at right angle from the laser propagation axis.The correlation(derived from PIV measurements) between the residence time of coal particles at the center of the burner and HAB is depicted in Fig.2.No measurable difference has been observed between measurements carried out using air or oxy-gen as carrier-gases regardless of the?at?ame used(FF1or FF2). Temperature of hot gases above the burner has been measured axi-ally and radially using a200-?m diameter S thermocouple coated with a thin layer of beryllium and yttrium oxides to reduce catalytic effects[24].The position of the beaded wire thermocouple shel-tered into an Inconel sheath has been adjusted thanks to a3-axis translation stage.Radiation losses have?nally been corrected using the usual electric compensation technique[25].As one can see by looking at Fig.3,the temperature of the central carrier-gas jets

26R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–31

Fig.2.Correlation derived from PIV measurements between the residence time of coal particles at the center of the burner and HAB in both combustion atmospheres with FF1and FF2.

(consisting in air or pure oxygen)rapidly increases from ambient temperature to~1140and~1240K for FF1and FF2,respectively, while no signi?cant difference can be observed depending on the nature of the carrier gas used.

In a second step,coal has been introduced into the acoustic sower(reactive conditions).The fuel has been?uidized by either air or oxygen and then injected at the center of the burner.The temperature of the coal particles has been monitored using a two-color infrared pyrometer Keller MSR-CellaTemp PA40(0.95?m and 1.05?m).The measurement range of this pyrometer is comprised between923and1973K.A monochromatic infrared pyrometer Keller MSR-CellaTemp PA20(1.1–1.7?m)having a measurement range comprised between523and2273K has thus also been used to assess temperatures outside of the PA40range.Measurement continuity has been ensured by adapting the emissivity of the PA20 to obtain the same temperature as the one given by the PA40in spe-ci?c positions of the?ames where the particle temperatures were included in both ranges of measurement.The pyrometers focused on a1.3-mm diameter target along the center of the coal jets thus leading to pro?les of the area-weighted mean temperatures of all the particles passing through the measurement region as explained and discussed in[22,26].The temperature pro?les obtained using air or oxygen as carrier-gases with FF1and FF2are plotted in Fig.3. Heating rates of the order of1.5×106and1.8×106K s?1can be estimated for FF1and FF2by determining the slope of the tangent to the origin of the particle temperature pro?les as a function of the residence time following the procedure adopted in[26](the conversion between HAB and residence time being achieved on the basis of the curves presented in Fig.2).Such values are rep-resentative of industrial combustion conditions[17].Furthermore, the heating of the fuel is insured by hot gases(and not by electri-cal means)which is also of interest to mimic operating conditions close to practical ones.The temperature of particles and gases are very close for HAB up to around50–60mm(corresponding to a coal particle residence time of~10ms).Nevertheless,after this height, the coal particle temperature becomes less important than the temperature of hot gases under non-reactive conditions

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

(between

Fig.3.Thermal history of carrier gases(CG)and coal particles in the central axis of the burner under both combustion environments with FF1(a)and FF2(b).

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Residence time (ms)

Experimental data

Model from Baum and Street

Model from Badzioch and Hawksley

Model from Kobayashi et al.

Model from Anthony et al.

20

40

60

80

100

D

e

v

o

l

a

t

i

l

i

z

a

t

i

o

n

y

i

e

l

d

(

w

t

%

)

Residence time (ms)

Experimental data

Model from Baum and Street

Model from Badzioch and Hawksley

Model from Fu et al.

Model from Kobayashi et al.

Model from Anthony et al.

http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlparison of calculated devolatilization yields using literature(a)and adjusted parameters(b)with experimental results issued from the analysis of the bituminous coal from Freyming using FF1and air as an oxidizer.

80and170K lower depending on the operating conditions).Such a phenomenon has also been observed during experiments con-ducted with a similar burner con?guration[22,26]and especially attributed to thermal losses induced by particle radiation.Such lower particle temperatures are moreover consistent with the heat sink effect that coal takes up to release volatiles[27].On the other hand,the graphs of Fig.3clearly illustrate that the particle temper-ature is signi?cantly higher when oxygen is used as a carrier-gas instead of air(differences of the order of60and80K being observed for FF1and FF2,respectively)which will be discussed in Section4.

3.Measuring and modeling devolatilization rates and

kinetics

3.1.Experimental approach

Char samples have been collected at different HAB as mentioned above and standardized proximate analyses of these samples have been carried out.Ash has been used as a tracer to determine the devolatilized fraction of the collected samples as in[22,28,29].The conservation of the quantity of ash during the devolatilization pro-cess implies the following equality:

M0×A coal=M×A char(1) where M0is the initial mass of coal,A coal designates the dry mass fraction of ash in the coal,M stands for the residual mass of char and A char corresponds to the dry mass fraction of ash in the collected char.The devolatilized mass M dev corresponds to the difference between M0and http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbining this difference with Eq.(1)leads to the determination of the devolatilized fraction F d expressed in mass percentage of the original dry-ash-free coal as explained in

[29]:

F d=100×

1?

A coal

1?A coal

×

1?A char

A char

(2)

This fraction can then be reported to the volatile matter con-tent of the original coal(on dry-ash-free basis)issued from the proximate analysis(VM0)to derive a fractional volatile conver-sion(as equated in[30]for instance)which is equivalent to a global devolatilization yield expressed as100×F d/VM0.One should note that three analyses were performed and averaged for each char sample(error bars represented on the graphs of Figs.4and5 being directly representative of the dispersion of experimental data around the mean values).

3.2.Modeling devolatilization kinetics using empirical models

3.2.1.Presentation of the models used in this work

The devolatilization rate modeling has been carried using4dif-ferent types of empirical models from the literature.These models

R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–31

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

27

http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlparison of calculated devolatilization yields with experimental results

using FF1and air as an oxidizer(a),FF1with O2as an oxidizer(b),FF2with air as

an oxidizer(c)and FF2with O2as an oxidizer(d).The insets of each?gure rep-

resent the devolatilization pro?les obtained using a logarithmic scale for the axis

corresponding to the residence time.

can be classi?ed depending on the number of?rst-order reactions

they take into account[31].For instance,the model proposed by

Baum and Street[32]simply considers a constant devolatilization

rate.Badzioch and Hawksley[29]or Fu et al.[33]developed models

based on a single-reaction scheme.Models proposed by Kobayashi

et al.[28,34]and Ubhayakar et al.[35]consider two-competitive

reactions while Anthony et al.[36]integrated an in?nite number

of reactions in their calculation procedure.Each of these models is

brie?y described thereafter.

In the work of Baum and Street[32],it is assumed that volatiles

are released at a constant rate leading to the following expression

of the coal particle weight loss:

dM

dt

=?k×f0×M0(3)

where M is the particle mass,k is the rate constant,f0refers to the

volatile fraction contained in the coal and M0is the initial parti-

cle mass.Thus,devolatilization mainly depends on the value of k.

It begins when the vaporization temperature controlling the onset

of devolatilization is reached(~600K[32])and it stops when the

total weight loss reaches a value equal to M inf which represents the

hypothetic mass of volatiles released at the end of the devolatiliza-

tion process.The incremental weight losses can then be summed

and divided by the initial mass of volatiles contained in the original

dry-ash-free coal(i.e.,f0×M0which is equal to VM0)to obtain a

devolatilization yield as de?ned in Section3.1.

In the case of single-kinetic-rate models,the devolatilization

process is summarized by a simple reaction written as[19,31]:

C k→XVM+(1?X)R c(4)

where C is the coal,VM designates the volatile matters,X is the

volatile fraction and R c stands for the residual char.In this case,the

rate constant k follows an Arrhenius equation of the type:

k=k0×exp

?E

R×T p

(5)

where k0is the pre-exponential factor,E the activation energy,R

the ideal gas constant and T p the particle temperature.The weight

loss of volatiles during a time step dt is then expressed as:

dV

dt

=k×

V inf?V

(6)

where V inf corresponds to the mass fraction of volatiles released for

an in?nite time(i.e.the total volatile content of the coal[37]?xed

to VM0in the following considering the investigated temperature

range)and V is the fraction of volatiles released by time t[37].Two

types of single-kinetic-rate models have been considered in the fol-

lowing.They are issued from the works of Badzioch and Hawksley

[29]and Fu et al.[33].In[29],the pre-exponential factor of the rate

constant depends on the nature of the coal.Values for this param-

eter have been tabulated for bituminous coals as a function of the

fuel carbon content[29].The values of E and k0in[33]depend

on the heating conditions and on the temperature reached by coal

particles.They are tabulated for temperatures comprised between

1000and2100K[33].

In the case of two-competing-rate models,the formation of

volatiles and char is described by the means of two?rst-order

reactions which compete for the remaining un-reacted fuel:

C k1→?1VM1+(1??1)R c1

k2

→?2VM2+(1??2)R c2(7)

where?1and?2are mass stoichiometric coef?cients referring to

the reactions occurring at low and high temperatures,respectively,

and k1and k2are their respective rate constants which follow

Arrhenius equations(see Eq.(5)).Following the thorough presen-

tation proposed in[38],the volatile release rate can be expressed

as follows:

dV

dt

=(?1×k1+?2×k2)×C(8)

with

dC

dt

=?(k1+k2)×C(9)

In the formulation proposed by Kobayashi et al.[28,34],?1is cho-

sen depending on the fuel proximate analysis while?2is taken

equal to unity as a complete weight loss is expected at extremely

high temperatures[28].Finally,values for the pre-exponential fac-

tors and activation energies used in the calculation of k1and k2are

proposed in[34].

In contrast with the models presented above,the distributed

activation energy models(DAEM)consider an in?nite number of

independent?rst-order reactions.In the formulation proposed

by Anthony et al.[36],the activation energy of the rate con-

stant is continuously distributed following a Gaussian function

f(E)centered on E0with a standard deviation .In this case,

the weight loss of volatiles can be estimated using the following

expression[37]:

V=V inf×

?

?1?1

×

×

+∞

exp

?

??k

t

exp

?E

R×T p

×dt?(E?E0)

2

2× 2

?

?×dE

?

?(10)

3.2.2.Implementation of the models and?tting procedure

The use of the model proposed by Baum and Street only

implies to?x a constant devolatilization rate whose value(derived

28R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–31

Table2

Summary of the devolatilization models and kinetic parameters.

Model type Kinetic expression Initial parameters

(Fig.4(a))Adjusted parameters (Fig.4(b))

Constant devolatilization rate model dM

dt

=?k×f0×M0k=12s?1k=44s?1(a)

Single kinetic rate model C k→XVM+(1?X)R c

dV

dt

=k×

V inf?V

k0=4.94×105s?1

E=7.40×104J mol?1

k0=6.90×105s?1

E=7.40×104J mol?1

Two-competing-rate

model

C k1→?1VM1+(1??1)R c1

k2

→?2VM2+(1??2)R c2

dV dt =(?1×k1+?2×k2)×C

?1=0.359

?2=1

k01=2.0×105s?1

k02=1.3×107s?1

E1=1.05×105J mol?1

E2=1.68×105J mol?1

?1=0.359

?2=1

k01=2.0×105s?1

k02=1.3×107s?1

E1=6.45×104J mol?1

E2=1.68×105J mol?1

Distributed activation energy

model V inf?V

V inf =1

×

.+∞

exp

?

??k0×

t

exp

?E

R×T p

×dt?(E?E0)

2

2× 2

?

?×dE

k0=1.67×1013s?1

E0=2.29×105J mol?1

=7.2×104J mol?1

k0=1.67×1013s?1

E0=2.29×105J mol?1

=1.5×104J mol?1

(a)The calculation being stopped for a yield of79%.

from[39])has been taken equal to12s?1in?rst approach. The single-kinetic-rate and two-competing-rate models have been implemented using a step-by-step scheme.To do so,the particle temperatures(considered as constant during each time step dt fol-lowing explanations given in[28,29])have been derived from the experimentally monitored pro?les of Fig.3to calculate the rate constants of each model(k or k1and k2)for every time interval (0.1ms).Incremental weight losses have then been estimated using Eq.(6)or(8)and summed to obtain total weights of volatiles emit-ted leading to devolatilization yields when referring to VM0.The kinetic parameters of each model(summarized in Table2)have been?rst selected based on the values proposed in[29,33,34]. Finally,Eq.(10)has been used for the implementation of the model from Anthony et al.[36].The activation energy integral has been calculated using the trapezoidal integral method applied in the [E0?2 ;E0+2 ]interval while the values selected for k0,E0and are issued from[40].

The devolatilization pro?les predicted by the different mod-els are plotted in Fig.4(a)together with the experimental data obtained using FF1and air as an oxidizer.These speci?c oper-ating conditions have been selected as they are close to those investigated in[22]where the high heating rate devolatiliza-tion of the French bituminous coal used herein was analyzed under N2.This being said,one can?rst note that the experi-mentally monitored devolatilization pro?le of Fig.4is similar to those reported in the literature[22,41]for heating rates of the order of106K s?1.On the other hand,the?nal devolatilization yield measured in the present work(around79%)is lower than the value reported in[22](>100%)for the same coal which is consistent with the lower temperature range investigated here. Fig.4(a)also illustrates the inadequacy of the tested models to simulate the experimental results.This can be related to the par-ticular experimental conditions investigated in the present study which differs signi?cantly from those associated to the datasets used to build the tested models(heating rates between104in [29,33,36]and105K s?1in[28,34]against around106K s?1herein and temperatures mainly higher than1000K in[28,33,34]).A?t-ting procedure of the parameters used for each model is therefore necessary.

For the model of Badzioch and Hawksley the pre-exponential factor of the rate constant k0can be adjusted(the activation energy E being kept constant)as explained in[29]and done in[22]for instance.We therefore implemented a simple optimization algo-rithm aiming to?nd a value for k0which allows minimizing the mean squared deviation between the experimental and the modeled data.A value of6.9×105s?1has then been obtained for k0 which is relevant considering the need to increase the devolatiliza-tion rate as noted based on the results of Fig.4(a).By the same way, a?tted value of64.5kJ mol?1has been obtained for the activation energy of the?rst reaction of the two-competing-rate model.In fact,the temperature of coal particles monitored in the present case being mostly below1000K,only the?rst reaction occurring at low temperature has been considered here.Such an approach leads to reduce the value of E1which in turn increases the number of low temperature reactions considered in the calculation(as explained in[28])consistently with the speci?c operating conditions con-sidered in this work.This way of proceeding moreover leads to obtain similar overall kinetic rates for single-and two-competing-rate models on the range of temperatures encountered by the coal particles.Finally,the activation energy distribution function of the model proposed by Anthony et al.has been truncated close to the mean activation energy E0to account for the high heating rate induced by the use of our?at?ame burner according to the expla-nations given in[22].The mean standard deviation has thus been decreased as in[22]to a value of15kJ mol?1which gave the best agreement between simulated and experimentally monitored devolatilization pro?les.The changes proposed for the above-mentioned parameters(summarized in Table2)allow obtaining the results plotted in Fig.4(b).One can note that the model of Anthony et al.slightly overestimates the measured weight losses during the ?rst40ms while it underestimates the?nal devolatilization yield. The agreement stays satisfactory,however.Moreover,the aim of the present parametric study is not to determine kinetic constants that would be universally valid but to?t the tested models to sim-ulate the devolatilization behavior of our bituminous coal under air condition.The obtained results will then serve as a reference to test the relative in?uence of the combustion atmosphere and especially of the oxygen enrichment on devolatilization kinetics

R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–3129

which is in the scope of the present paper.Concerning the model from Baum et al.,the linearity of the weight loss pro?le induced by the use of a constant kinetic rate as well as the need to de?ne a?nal devolatilization yield to stop the calculation makes it inap-propriate to?t our experimental data in addition to prevent such a model to be used as a predictive one.It will therefore not be considered anymore in the following.In the same way,the param-eters involved in the model from Fu et al.have not been adapted since values of E and k0only depend on the?nal temperature of coal particles and are supposed to be universal and indepen-dent of the coal type.Consequently,such a model appears to be unsuitable to simulate the devolatilization pro?les measured herein and it has therefore not been considered anymore here-after.

4.Results and discussion

4.1.In?uence of the oxygen enrichment on devolatilization

kinetics

Experimental and modeled devolatilization pro?les obtained using FF1and FF2with air and oxygen as carrier gases are plot-ted in Fig.5.The parameters used for each model are issued from the?tting procedure detailed in Section3.2.2.

The obtained experimental data?rst show that the devolatiliza-tion process ends for a residence time of around30ms(as in [22,41])for all the considered atmospheres.The devolatilized frac-tions also appear to be higher under oxygen enriched environments compared to the data obtained using air as an oxidizer.The?nal devolatilization yields thus pass from79to88%with FF1and from 86to93%with FF2.Such results are in agreement with the main trends depicted in the literature(see Section1)indicating that oxy-gen enrichment leads to a rise of the devolatilization yield.These results are moreover well correlated with the increase of the fuel particle temperature observed in Fig.3.Considering the in?uence of the?at?ame used to stabilize the coal jet?ames,one can note that the higher temperature in the reactor issued from the use of FF2induces an expected rise of the measured weight losses. The obtained devolatilization yields thus pass from79(with FF1) to86%(with FF2)when air is used as a carrier gas while rising from88to93%under OEC conditions.It is worthy to note that the increase in the fuel particle temperature induced by the oxygen enrichment occurs for a coal particle residence time of the order of10ms while the devolatilization process is still not completed. Such a behavior can be explained by an enhanced combustion of the ejected volatile matters due to the higher amount of oxygen available in the medium.The higher temperature of the volatile ?ame surrounding the fuel particles then induces an ampli?cation of heat transfers towards coal(clearly observable in Fig.3)enhanc-ing the devolatilization process.Experimental results obtained in this study thus tend to corroborate the conclusions drawn in the modeling work from Cho et al.[14].

The pro?les of Fig.5also illustrate that all the tested mod-els?t well the experimental data.Since kinetic parameters have been kept constant regardless of the considered atmosphere,the adequacy between modeled and experimental data implies that changes in devolatilization rates are only due to changes in the thermal history of the fuel particles.To clearly illustrate such a correlation between the fuel particle temperature and the quan-tity of volatiles released,we plotted in Fig.6the evolution of the ?nal devolatilization yields obtained by experiment and modeling as a function of the peak temperature of coal particles as done in [42].The obtained results clearly show that modeled values match the experimental data within a~+/?5%error range correspond-ing to the overall uncertainty related to the determination of

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

the

70

80

90

100

9901020105010801110

F

i

n

a

l

d

e

v

o

l

a

t

i

l

i

z

a

t

i

o

n

y

i

e

l

d

(

w

t

%

)

Peak temperature (K)

Experimental data

Single-kinetic-rate model

Two-competing-rate model

Distributed activation energy model

Fig.6.Correlation between the?nal devolatilization yield and the peak temperature of the fuel particles.

devolatilization yields(see error bars plotted in Figs.4and5).Such a conclusion is consistent with the statement from Scheffknecht et al.

[43]who noted that the devolatilization stage was a purely thermal decomposition process which was only indirectly affected by the surrounding atmosphere via the way this one in?uences heat trans-fers towards coal particles.On the other hand,another fundamental conclusion drawn by the set of results presented here concerns the apparent absence of overlapping between devolatilization and char oxidation stages as de?ned and discussed in Section1based on the work from Cho et al.[14].Indeed,if the oxygen diffusion towards coal was signi?cant during the devolatilization stage and that the particle oxidation was therefore active,increasing the oxy-gen concentration during OEC experiments should enhance the char oxidation and then accelerate the particle mass loss.How-ever,as an increase of the oxygen concentration does not affect the devolatilization kinetic parameters(which are kept constant as explained above),results presented here thus suggest the absence of any overlapping process(?nal devolatilization yields measured herein being moreover consistent compared to the values reported in[22]using the same pulverized coal under inert atmosphere). Such information is of interest to better understand the charac-teristics as well as the main steps leading to the thermochemical conversion of pulverized coal particles under OEC conditions with high heating rates close to those met in practical combustors.This conclusion is moreover in agreement with the results reported in a late work from Riaza et al.[44]who analyzed the combustion of bituminous coal particles ground and sieved to a particle size cut of75–150?m(i.e.,relatively close to the size range considered here knowing that~90%of the Freyming particles have a mea-sured diameter comprised between65and200?m).Riaza et al. especially observed a two-phase combustion process by means of three-color pyrometry and high-speed high-resolution cinematog-raphy.To these authors,volatiles?rst evolve,ignite and burn in luminous enveloping?ames.Upon extinction of these?ames,the char residues then ignite and burn[44].This general combustion behavior corroborates the analysis proposed here though strikingly disparate combustion behaviors were noted in[44]depending on the coal rank thus demonstrating the interest of leading such addi-tional experimental characterizations.

4.2.In?uence of the oxygen enrichment on gaseous emissions

To complete the analysis of the impact of OEC on the devolatilization behavior of the Freyming coal,concentrations of O2,CO,CO2,SO2and NO have been measured in the reaction medium.Pro?les of such species are plotted in Fig.7as a function of the residence time of the solid fuel in the reactor when using

30

R.Lemaire et al./Journal of Analytical and Applied Pyrolysis 114(2015)22–31

Study of the high heating rate devolatilization of a pulverizedbituminous coal under oxygen-containi

Residence time (ms)

CO2O2CO

SO2

NO

0.00

2.304.606.909.2011.50C O 2a n d O 2(%)

Residence time (ms)

CO2O2CO

SO2

NO

Fig.7.Concentration pro?les of O 2,CO,CO 2,SO 2and NO as a function of the resi-dence time of coal particles using FF2and air (a)or O 2(b)as carrier gases.

FF2and air or oxygen as carrier gases.One can note that addi-tional information regarding other species (not measured here)which are likely to be released during coal devolatilization under complex atmospheres containing oxygen can be found in [45]for instance.Finally,data obtained using FF1have not been reported here since they have already been presented and commented in [41].Furthermore,the slight modi?cation of the equivalence ratio operated between FF1and FF2does not modify the main trends and conclusions presented hereafter.

When looking at the pro?les of Fig.7,one can ?rst note that the oxygen fractions measured immediately after the coal injec-tor outlet are signi?cantly lower than the oxygen fractions of the carrier gases used to ?uidize the fuel particles.O 2concentrations are thus of around 10%(instead of 21%)and 43%(instead of 100%)when using air and oxygen as carrier gases,respectively,for a residence time of ~3.5ms.This is due to the fact that measure-ments are integrated over the section of the quartz collector whose internal diameter is of 4cm.As a consequence,burnt gases issued from the ?at ?ame are also pumped into the probe and dilute the carrier-gas/coal-particles mixture.Additional measurements have been performed using a micro-probe similar to the one used in [46]for soot sampling.In this case,no coal particles were injected at the center of the burner.The obtained results revealed that the oxygen concentration in the central axis of the burner was very close to the oxygen content of the carrier gas used while it pro-gressively decreases as the probe was laterally moved outside of the central core of the burner.As spatially resolved measurements using such a micro-probe were not possible in ?ame conditions due to perturbation and plugging problems (the diameter of the micro-probe being of the same order of magnitude than the size of the fuel particles),measurements were performed using the quartz collector and integrated over its whole section.That way,one can have a relevant indication concerning the composition of the overall environment in which the whole coal/volatiles jets evolve.

The concentration pro?les plotted in Fig.7show that the com-bustion is more complete under OEC.Indeed,the CO concentration peaks for a residence time of around 3.5ms when oxygen is used as a carrier gas whereas it peaks at around 8.5ms under air condi-tions.Furthermore,a signi?cantly higher initial peak of CO (0.43%for OEC instead of 0.12%under air)can be measured and related to the enhanced devolatilization process that occurs under OEC (illustrated in Section 4.1)leading to the formation of a fuel-rich environment around the coal particles during the ?rst milliseconds of the conversion process.Then,the fraction of CO in the exhaust gases becomes lower under OEC (~115ppm versus ~270ppm on average between 17ms and 100ms)as CO reacts to form CO 2the concentration of which is higher (around 12.64%versus 9.76%on average until 100ms).The oxidation of volatile matters,which are released in higher quantity under OEC,is thus more complete due to the higher amount of oxygen available in the medium which is consistent with general trends depicted in [6,20].

are concerned,they are multi-when using oxygen instead of air as a increase of NO emissions with the oxygen to an increase of the fuel-N trans-the production of prompt-NO during short contributes to less than 5%of the over-the formation of thermal-NO begins to temperatures (above 1800K [47])that are not reached in this study.We can thus deduce that the observed increase of the NO concentration under OEC mainly relies on an increase of the fuel-NO formation.This concurs well with the fact that devolatilization is more complete under OEC which in turn is likely to increase the quantity of nitrogen from volatiles available in the medium.Furthermore,the higher oxygen partial pressure met under OEC conditions is also a factor favoring the NO forma-tion [48].One can note that NO concentrations peak for residence times comprised between around 7(Fig.7(b))and 10ms (Fig.7(a))corresponding to the regions where the measured temperatures are maximal (see the pro?les of Fig.3).Such an observation is consistent with the in?uence of the temperature on the NO pro-duction [48]as well as with the relatively high temperature levels needed to insure the tar-nitrogen conversion [47].This also con-curs to explain why the NO production is much more marked under OEC since the temperature of the volatile ?ame surround-ing the fuel particles is higher (as commented in Section 4.1)thus contributing to promote the transformation of the fuel-N released during the enhanced devolatilization process.Note also that NO volume fractions seem to decrease after the observed peak val-ues which can be related to a less pronounced NO formation due to the lower temperatures measured for residence times higher than 15ms coupled with a probable dilution effect induced by the increase of the concentration of other gaseous species released dur-ing the devolatilization process which then react in the volatile ?ame.

Finally,the SO 2volume fractions in the exhaust gases are around 50%higher on average when O 2is used instead of air.The maximal SO 2concentration is furthermore quite immediately reached (a few milliseconds after the beginning of the devolatilization)whatever the considered atmosphere which is consistent with the relatively low temperature range needed to release such a pollutant as shown in [49].The higher production of SO 2is ?nally well correlated with the fact that SO x emissions are mainly governed by the quantity of fuel-sulfur emitted during the devolatilization process and by the O 2concentration in the reaction medium.As a consequence,the higher the devolatilization yield and the higher the O 2concentra-tion,the higher the SO 2emissions as con?rmed by the pro?les of Fig.7.

5.Conclusion

The high heating rate devolatilization of a bituminous coal has been analyzed under air and OEC conditions by coupling experi-mental and modeling approaches.To do so,a newly developed FFR has been used to generate operating conditions of interest for prac-tical applications.The following conclusions can be drawn from the present work:

?The oxygen enrichment of the reaction medium induces a rise of the fuel particle temperatures that subsequently leads to higher devolatilization yields.Such a phenomenon has been related to an ampli?cation of heat transfers towards coal particles resulting from the oxygen enhanced combustion of devolatilized matters as con?rmed by gas analyses.

?The thermal history of the fuel particles,characterized by cou-pling PIV and pyrometric measurements,has been integrated into

R.Lemaire et al./Journal of Analytical and Applied Pyrolysis114(2015)22–3131

different devolatilization models including single-kinetic-rate, two-competing-rate and DAEM models.A?tting procedure has then been operated to?nd adjusted kinetic parameters leading to reproduce the devolatilization pro?les experimentally moni-tored.

?The rate constant parameters used during the modeling pro-cedure did not need to be modi?ed to simulate well the experimental data regardless of the considered atmosphere. Apparent devolatilization rates thus appeared to be only indi-rectly affected by the surrounding atmosphere through the way this one in?uences the heating of coal particles.Furthermore, since an increase of the oxygen concentration does not affect the devolatilization kinetic parameters,the results reported in this paper also suggest the absence of overlapping between devolatilization and char oxidation stages which is of practical interest to?gure and then model the main steps involved in the high heating rate thermochemical conversion of pulverized bituminous coal particles under oxygen containing atmospheres.?Coupled with the characterization of devolatilization kinetics, measurements of gas concentrations illustrated that the combus-tion of volatiles is more complete under OEC.More CO2is thus produced while CO concentrations signi?cantly decrease.?Volume fractions of NO are around two times higher when oxygen is used as a carrier gas instead of air due to an enhanced fuel-N conversion related to the higher temperatures and oxygen partial pressures.

?SO2concentrations are~50%higher on average under OEC due to the higher quantity of fuel-sulfur emitted during the devolatiliza-tion and to the higher amount of O2available in the medium.

Acknowledgments

Mines Douai and PC2A laboratories participate in the Insti-tut de Recherche en ENvironnement Industriel(IRENI)which is ?nanced by the CommunautéUrbaine de Dunkerque,the Région Nord Pas-de-Calais,the Ministère de l’Enseignement Supérieur et de la Recherche,the CNRS and the European Regional Development Fund(ERDF).The authors thank the?Programme Interdisciplinaire Energie?of the CNRS that supports the OxyChar research program.

A.Bocheux from CERCHAR is warmly thanked for providing the coal that has been used in this work.Finally,the authors thank

B.Lecrenier from the PC2A laboratory for her contribution to the char analysis,S.Menanteau from Mines Douai–EI for fruitful dis-cussions and

C.Evrard,J.L.Faille,C.Ghyselen and

D.Jacquart from Mines Douai–EI for their technical support.

References

[1]International Energy Agency,World Energy Outlook(2012).

[2]S.Ansolabehere,J.Katzer,J.Beer,J.Deutch,A.D.Ellerman,S.J.Friedmann,H.

Herzog,H.D.Jacoby,P.L.Joskow,G.McRae,R.Lester,E.J.Moniz,E.Steinfeld, An Interdisciplinary MIT Study,Massachusetts Institute of Technology,2007, ISBN:978-0-615-14092-6.

[3]B.J.P.Buhre,L.K.Elliott,C.D.Sheng,R.P.Gupta,T.F.Wall,Prog.Energy

Combust.Sci.31(2005)283–307.

[4]L.Chen,S.Z.Yong,A.F.Ghoniem,Prog.Energy Combust.Sci.38(2012)

156–214.

[5]C.E.Baukal Jr.,Oxygen-Enhanced Combustion,CRC Press LLC,1998,ISBN:

0-8493-1695-2.

[6]S.S.Daood,W.Nimmo,P.Edge,B.M.Gibbs,Fuel101(2012)187–196.

[7]L.Timothy,A.F.Saro?m,J.M.Beér,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.19(1982)1123–1130.

[8]B.R.Stanmore,Y.C.Choi,R.Gadiou,O.Charon,P.Gilot,Combust.Sci.Technol.

159(2000)237–253.

[9]P.A.Bejarano,Y.A.Levendis,Combust.Sci.Technol.179(2007)1569–1587.

[10]P.A.Bejarano,Y.A.Levendis,Combust.Flame153(2008)270–287.

[11]J.J.Murphy,C.R.Shaddix,Combust.Flame144(2006)710–729.

[12]A.Molina,C.R.Shaddix,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.31(2007)1905–1912.

[13]C.R.Shaddix,A.Molina,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.32(2009)2091–2098.

[14]C.P.Cho,S.Jo,H.Y.Kim,S.S.Yoon,Numer.Heat Transfer,Part A:Appl.52

(2007)1101–1122.

[15]C.A.Gurgel Veras,J.Saastamoinen,J.A.Carvalho Jr.,M.Aho,Combust.Flame

116(1999)567–579.

[16]J.Brix,P.A.Jensen,A.D.Jensen,Fuel89(2010)3373–3380.

[17]Y.C.Guo,C.K.Chan,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlu,Fuel82(2003)893–907.

[18]R.Lemaire,M.Maugendre,T.Schuller,E.Therssen,J.Yon,Rev.Sci.Instrum.80

(2009)105105.1–105105.8.

[19]J.Zhang,Coal Combustion Research Advances,Nova Science Publishers,2010,

ISBN:978-1616689353.

[20]Y.Hu,S.Naito,N.Kobayashi,M.Hasatani,Fuel79(2000)1925–1932.

[21]G.Branchet,R.Lemaire,E.Therssen,33rd International Symposium on

Combustion,Tsinghua University,Beijing,China,2010.

[22]E.Therssen,L.Gourichon,L.Delfosse,Combust.Flame103(1995)115–128.

[23]Y.Bouvier,C.Mihesan,M.Ziskind,E.Therssen,C.Focsa,J.F.Pauwels,P.

Desgroux,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.31(2007)841–849.

[24]J.H.Kint,Combust.Flame14(1970)279–281.

[25]U.Bonne,T.Grewer,H.G.Wagner,Z.Phys.Chem.26(1960)93–110.

[26]E.Therssen,L.Delfosse,Rev.Sci.Instrum.66(1995)4041–4044.

[27]Y.Xie,V.Raghavan,A.S.Rangwala,Combust.Flame159(2012)2449–2456.

[28]H.Kobayashi,Ph.D.Thesis,M.I.T.,Dept.of Mechanical Engineering,

Cambridge,Massachusetts,1976.

[29]S.Badzioch,P.G.W.Hawksley,Ind.Eng.Chem.Process Des.Dev.9(1970)

521–530.

[30]C.Ulloa,A.L.Grodon,X.Garcia,J.Anal.Appl.Pyrolysis71(2004)465–483.

[31]A.Williams,M.Pourkashanian,J.M.Jones,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.28(2000)

2141–2162.

[32]M.M.Baum,P.J.Street,Combust.Sci.Technol.3(1971)231–243.

[33]W.Fu,Y.Zhang,H.Han,D.Wang,Fuel68(1989)505–510.

[34]H.Kobayashi,J.B.Howard,A.F.Saro?m,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.16(1977)

411–425.

[35]S.K.Ubhayakar,D.B.Stickler,C.W.Von Rosenberg,R.E.Gannon,Proc.

Combust.Inst.16(1977)427–436.

[36]D.B.Anthony,J.B.Howard,H.C.Hottel,H.P.Meissner,http://www.wendangku.net/doc/90fe65812e3f5727a4e96290.htmlbust.Inst.15

(1975)1303–1317.

[37]S.C.Saxena,Prog.Energy and Combust.Sci.16(1990)55–94.

[38]J.Lehto,Ph.D.Thesis,Tampere University of Technology,2007.

[39]K.K.Pillai,J.Inst.Energy54(1981)142–150.

[40]D.B.Anthony,J.B.Howard,J.B.Hottel,H.P.Meissner,Fuel55(1976)121–128.

[41]D.Menage,R.Lemaire,C.Bruhier,J.L.Harion,Récent Progrès en Génie des

Procédés,104,ed.SFGP,Paris,2013,ISBN978-2-910239-78-7.

[42]X.Ma,H.Nagaishi,T.Yoshida,G.Xu,M.Harada,Fuel Process.Technol.85

(2003)43–49.

[43]G.Scheffknecht,L.Al-Makhadmeh,U.Schnell,J.Maier,Int.J.Greenh.Gas

Control5S(2011)S16–S35.

[44]J.Riaza,R.Khatami,Y.A.Levendis,L.Alvarez,M.V.Gil,C.Pevida,F.Rubiera,J.J.

Pis,Combust.Flame161(2014)1096–1108.

[45]M.Zhong,Z.Zhang,Q.Zhou,J.Yue,S.Gao,G.Xu,J.Anal.Appl.Pyrolysis97

(2012)123–129.

[46]R.Lemaire,A.Faccinetto,E.Therssen,M.Ziskind,C.Focsa,P.Desgroux,Proc.

Combust.Inst.32(2009)737–744.

[47]P.Glarborg,A.D.Jensen,J.E.Johnsson,Prog.Energy Combust.Sci.29(2003)

89–113.

[48]M.J.Aho,K.M.Paakkinen,P.M.Pirkonen,P.Kilpinen,M.Hupa,Combust.Flame

102(1995)387–400.

[49]S.Scaccia,TG-FTIR and kinetics of evolatilization of Sulcis coal,J.Anal.Appl.

Pyrolysis104(2013)95–102.