Volatile Releasing Characteristics of Pulverized Coals under Moderate or Intense Low-Oxygen Dilution Oxy-Combustion Conditions in a Flat-Flame Assisted Entrained Flow Reactor

: There has been little research on volatile releasing characteristics of pulverized coals under moderate or intense low-oxygen dilution (MILD) oxy-combustion (MO) conditions. For the ﬁrst time, volatile releasing characteristics of bituminous coal and semi-anthracite under both MILD air-combustion (MA) and MO conditions were investigated using a ﬂat-ﬂame assisted entrained ﬂow reactor. Both heating rate (~10 5 K/s) and residence time (65 ms) were carefully selected to mimic the conditions in typical industrial boilers. The combustion processes and properties of the volatiles were characterized through direct observation and char analysis. The results showed that the lower diffusion rate of the volatile in CO 2 resulted in the decreasing of the volatile envelope ﬂame size and a longer volatile burnout time (more than 20%). For bituminous coal (volatile content of 25%), the lower amount of apparent volatile yield under MO conditions reduced the heating value of the volatile. For semi-anthracite coal (volatile content of 7%), the short devolatilization time led to char-CO 2 gasiﬁcation reaction, which increased the apparent volatile yield and the heating value of the volatile by 47% and the volatile-N by 19%. This paper indeed provided new insight into the MILD oxy-combustion of solid fuels.


Introduction
Oxy-fuel combustion technology has received considerable attention globally as one of the potential approaches to achieve sequestration-ready CO 2 gas streams from coal-fired power plants [1,2].In oxy-fuel combustion, the mixture of pure oxygen and recycled flue gas is employed as an oxidizer, replacing air [3,4].However, remarkable differences exist in the physical, chemical, and thermodynamic properties of CO 2 and N 2 , such as the heat capacity and transport properties [5][6][7].At the same time, the chemical properties of CO 2 on the char reaction are more pronounced than those of nitrogen [8,9].
Although oxy-fuel combustion possesses the above significant preponderances, not only its combustion stability needs to be further strengthened, but also the generation of soot and NO x needs to be further controlled [10,11], to meet the requirements of ultra-low emissions [12].The above issues are hopefully resolved by moderate or intense lowoxygen dilution (MILD or flameless combustion) [13][14][15].The MILD oxy-combustion, which combines oxy-fuel combustion and MILD combustion [9][10][11]16], can further realize low-carbon, efficient, and clean utilization of coal burning.The advantages of MILD combustion can not only enhance the stabilization of oxy-fuel combustion and further decrease the NO x generation [11,17], but also homogenize the furnace temperature [18], Processes 2022, 10, 358 2 of 16 strengthen the in-furnace heat exchange [19], and thus improve the efficiency of oxy-fuel combustion [20,21].
Previous research has studied MILD combustion or oxy-fuel combustion of pulverized coal, while experimental research on MILD oxy-combustion of pulverized coal is still in its early stage and relatively scarce.The existing MILD oxy-combustion studies of pulverized coal mostly focus on the implementation conditions [22][23][24], transformation law [21,25,26], and the NO x emission reduction mechanism [20,23,24], but do not involve the research on the volatile releasing characteristics and reactivity of coal char.
The existing research on char characteristics under conventional oxy-fuel combustion is of certain reference significance for MILD oxy-combustion and is essential for brief review.The high CO 2 concentration in oxy-fuel combustion shows a strong impact on the coal devolatilization process [27,28].Numerous experimental studies on volatile release and combustion characteristics of pulverized coals have been reported using different facilities in N 2 and CO 2 environments.These experiments mainly focused on evaluating the apparent volatile yield through the analysis of the char collected [29][30][31][32] and measuring the devolatilization duration through direct observation using optical equipment [33][34][35][36][37][38].Only a few researchers [31] have studied the composition of the volatile under oxy-fuel conditions.It is noted that these studies did not produce a consistent conclusion, primarily because of different experimental equipment and operating conditions being used.When an isothermal thermogravimetric analysis (TGA) was chosen to measure the weight loss of a pulverized bituminous coal in N 2 and CO 2 atmospheres [30], it was observed that, although coal devolatilization started at the same temperature in the two atmospheres, the weight loss in CO 2 was significantly higher than that in N 2 at T > 1030 K.This indicated that the char-CO 2 gasification reaction was prominent at temperatures greater than 1030 K.A large difference exists in temperature and heating rates between TGA and real industrial boilers, and thus this influences the devolatilization rates [39].
To augment the heating rate and the final temperature of the particles, a drop tube furnace (DTF) and/or entrained flow reactor (EFR) is usually employed to conduct the pyrolysis experiment.The char samples were collected to measure the weight loss, structure, and reactivity of the fuel and char.Rathnam et al. [30] examined the volatile yields of four different coals in a DTF under O 2 /N 2 and O 2 /CO 2 atmospheres at 1670 K.They found that the apparent volatile yields measured in CO 2 atmosphere were greater than those in N 2 for all the coals studied with a residence time of 620 ms.The fact that the presence of CO 2 increased the volatile yield was also obtained by Al-Makhadmeh et al. [31] from the EFR pyrolysis tests with a residence time of 1000 ms.In contrast, an opposite trend was found by Borrago and Alvarez [32] from the devolatilization experiment of bituminous coal using a DTF with a residence time of 300 ms at 1573 K.They concluded that CO 2 could be involved in the cross-linking at the surface of the resolidifying chars, thus reducing the swelling of the particles and inhibiting the volatile release.Brix et al. [29] studied the pyrolysis of bituminous coal using an EFR with a residence time of 297 ms.Their results showed that the volatile yields in O 2 /N 2 and O 2 /CO 2 atmospheres were not noticeably different.The discrepancies found in these DTF and EFR experiments suggested that the operating conditions, such as the particle heating rate and the total particle residence time, could inevitably affect the volatile yields [10].
Seeker et al. [40] observed directly the thermal decomposition of pulverized coal particles under a condition close to industrial boilers.It was found that the devolatilization occurred in the range of 10~75 ms.A longer residence time can be used to investigate the volatile release of pulverized coal in the N 2 atmosphere, then in the CO 2 atmosphere, due to the reaction of CO 2 with char at high temperatures.If the residence time is employed longer than necessary, then char reaction can inevitably take place.
With the development of optical measurement technology, direct observations or imaging technology have been employed to study combustion phenomena [16,41].These experiments have usually been conducted on flat-flame reactors [35,37,38,42] or hightemperature EFRs [33,34], which provide a high temperature of more than 1600 K and a high heating rate of more than 10 5 K/s.The optical intensity signal was captured by a high-speed camera or pyrometer.The volatile combustion process could be recognized from the optical intensity profiles captured.Shaddix et al. [38,42] measured the combustion time of the volatiles of bituminous coals under O 2 /N 2 and O 2 /CO 2 conditions by an intensified charge-coupled device (ICCD) camera and suggested that the presence of CO 2 could lead to a 20% longer volatile burnout time because of the relatively low diffusion rate of the volatile in CO 2 .Levendis and Khatami et al. [33,36] extended the research from bituminous coal to lignite.Additionally, longer volatile burnout times were observed for bituminous coals burning in the CO 2 atmosphere than in the N 2 atmosphere.However, the difference decreased for sub-bituminous coals and lignite.Riaza et al. [34] further added the semi-anthracite coals.They found that the volatile burnout times increased linearly with the increase in the volatile matter content of the coal in all gas environments.The volatile burnout times were found to be drastically shortened with the increase of oxygen concentration under O 2 /N 2 and O 2 /CO 2 atmospheres [33,34,37,38].In these studies, the solid sample was not collected for evaluating the apparent volatile yield because of the short devolatilization duration and very low particle feeding rate.
In summary, there are at least three remaining research gaps: (1) the particle residence times (e.g., 150-290 ms in Ref. [29] and 300 ms in Ref. [32]) in most previous studies were longer than the actual devolatilization times in typical industrial furnaces (less than 100 ms [43]).A longer residence time can lead to additional char reactions after completion of the devolatilization process [41], which may not reflect the real devolatilization process; (2) almost all existing studies have focused on highly volatile coals (lignite and bituminous coal), and only a few papers have reported the devolatilization characteristics of anthracite coal; and (3) the previous relevant studies have been primarily concerned about the conditions of high oxygen concentration, while the corresponding conclusions cannot refer to the conditions of low oxygen concentration of MILD oxy-combustion.Therefore, the present work was designed to address these gaps.For the first time, an investigation of the devolatilization characteristics of two different rank coals under MILD oxy-combustion (MO) and MILD air-combustion (MA) conditions was conducted.A high heating rate and a short residence time were used to simulate conditions similar to those in industrial boilers.A high-speed camera was employed to capture the image of the individual particles to evaluate the combustion process of the volatile, such as volatile burnout time and the size of the volatile flame.Other properties of the volatiles, such as apparent volatile yield, element compositions, and heating values, were studied based on the analysis of the collected chars to reveal the reaction mechanisms and effects of CO 2 on the devolatilization process of pulverized coals.

Coal Samples
Two different ranks of coals were investigated, i.e., DT bituminous coal and JC semianthracite coal.The coals were sieved to obtain a narrow particle size range of approximately 105~125 µm and then dried in an oven at a temperature of less than 378 K.The properties of the two coals were shown in Table 1.

Devolatilization Experiments
The devolatilization experiments were conducted on a flat-flame assisted entrained flow reactor (FF-EFR) at 1670 K.The reactor (shown in Figure 1) was operated at 1 atm.A modified McKenna flat-flame burner was used to provide different gas compositions through adjustment of fuel (the mixture of CH 4 /CO/H 2 ) and oxidizer (the mixture of O 2 /N 2 /CO 2 ) flow rates.The high-temperature burnt gas acted as the coal oxidation medium and heated the injected coal particles.The particle heating rate was approximately 10 5 K/s, which was close to that found in industrial furnaces [44].The feeding rate of the coal particles was kept at 2.5 g/h to weaken particle-particle interactions and diminish the influence of coal combustion on the reactor gas temperature and composition.

Devolatilization Experiments
The devolatilization experiments were conducted on a flat-flame assisted entrained flow reactor (FF-EFR) at 1670 K.The reactor (shown in Figure 1) was operated at 1 atm.A modified McKenna flat-flame burner was used to provide different gas compositions through adjustment of fuel (the mixture of CH4/CO/H2) and oxidizer (the mixture of O2/N2/CO2) flow rates.The high-temperature burnt gas acted as the coal oxidation medium and heated the injected coal particles.The particle heating rate was approximately 10 5 K/s, which was close to that found in industrial furnaces [44].The feeding rate of the coal particles was kept at 2.5 g/h to weaken particle-particle interactions and diminish the influence of coal combustion on the reactor gas temperature and composition.The effect of CO2 on the coal devolatilization process was assessed by entraining coal particles into the mixture gas with N2 or CO2 as the diluent gas.Table 2 displayed the composition of the gas mixtures examined.Before the experiment, the EQUIL module in CHEMKIN was used to calculate chemical equilibrium based on the inlet compositions, and the compositions of the gas mixtures shown in Table 2 were obtained.The high-temperature and low-oxygen conditions could mimic the typical MILD oxy-combustion conditions [12,16].To avoid problems with tar clogging the sampling probe and the pore of the char particles, the devolatilization experiments were performed with approximately 2% O2 in the gas mixture for the burnout of the volatile.This could not influence the coal mass losses [29] and the 2% O2 during the devolatilization process was not found to affect the char reactivity [45].The gas temperature profiles along the centerline of the reactor for two blank cases were measured with a type B beaded thermocouple after correction for radiative losses The effect of CO 2 on the coal devolatilization process was assessed by entraining coal particles into the mixture gas with N 2 or CO 2 as the diluent gas.Table 2 displayed the composition of the gas mixtures examined.Before the experiment, the EQUIL module in CHEMKIN was used to calculate chemical equilibrium based on the inlet compositions, and the compositions of the gas mixtures shown in Table 2 were obtained.The hightemperature and low-oxygen conditions could mimic the typical MILD oxy-combustion conditions [12,16].To avoid problems with tar clogging the sampling probe and the pore of the char particles, the devolatilization experiments were performed with approximately 2% O 2 in the gas mixture for the burnout of the volatile.This could not influence the coal mass losses [29] and the 2% O 2 during the devolatilization process was not found to affect the char reactivity [45].The gas temperature profiles along the centerline of the reactor for two blank cases were measured with a type B beaded thermocouple after correction for radiative losses [46].As exhibited in Figure 2, the gas temperature was similar for the two cases.The highest gas temperatures were equal to the adiabatic flame temperatures and appeared at the location of 5 mm from the burner surface, where it was close to the flat flame front.The gas temperatures decreased with the distance from the burner owing to the heat loss through the quartz walls.
Processes 2022, 10, x FOR PEER REVIEW 6 of 16 [46].As exhibited in Figure 2, the gas temperature was similar for the two cases.The highest gas temperatures were equal to the adiabatic flame temperatures and appeared at the location of 5 mm from the burner surface, where it was close to the flat flame front.The gas temperatures decreased with the distance from the burner owing to the heat loss through the quartz walls.The solid sample was iso-kinetically collected for further analysis through an oilcooled sampling probe and then separated by a cyclone.The recycled oil of the sampling probe was heated to 400 K to avoid water condensation in the probe, and an additional nitrogen stream was added to the exhaust gases to quench the reaction and to improve the collection efficiency of the cyclone separator.To ensure consistency, the oil-cooled sampling probe was adjusted to a position where the estimated residence time of the particles in the reactor was approximately 65 ms for the investigated cases.The sampling times lasted 7.5 h to collect sufficient partially reacted chars, guaranteeing the representativeness of the samples.
A high-speed camera (Photron FASTCAM SA4) was used to take images of the individual particles for capturing the transition process of the coal particles.This camera was mounted with a 105 mm Micro lens, an aperture of 2.8, and an anti-blooming CMOS sensor.The camera focus was set on the centerline of the reactor, the shutter speed was set to 5000 fps, and 500 pictures were recorded for each case.

Direct Observations of the Volatile Releasing Processes
The typical photographs (when the flat flame was formed and the powder was stabilized) of the DT bituminous coal and JC semi-anthracite coal devolatilizing in MA and MO atmospheres were shown in Figure 3.These photographs only captured the envelope flame without the unburned char particles, due to the absence of a high-brightness background light source [35,41,47].It was clear that both the spot brightness and size of DT bituminous coal were larger than those of JC semi-anthracite coal.To obtain more detailed particle information, a series of image processing procedures for the optical intensities and sizes of the discernible spots were conducted using the Matlab software.More than 500 consecutive images were chosen to obtain sufficient particles to reduce the statistical error.The solid sample was iso-kinetically collected for further analysis through an oilcooled sampling probe and then separated by a cyclone.The recycled oil of the sampling probe was heated to 400 K to avoid water condensation in the probe, and an additional nitrogen stream was added to the exhaust gases to quench the reaction and to improve the collection efficiency of the cyclone separator.To ensure consistency, the oil-cooled sampling probe was adjusted to a position where the estimated residence time of the particles in the reactor was approximately 65 ms for the investigated cases.The sampling times lasted 7.5 h to collect sufficient partially reacted chars, guaranteeing the representativeness of the samples.
A high-speed camera (Photron FASTCAM SA4) was used to take images of the individual particles for capturing the transition process of the coal particles.This camera was mounted with a 105 mm Micro lens, an aperture of 2.8, and an anti-blooming CMOS sensor.The camera focus was set on the centerline of the reactor, the shutter speed was set to 5000 fps, and 500 pictures were recorded for each case.

Direct Observations of the Volatile Releasing Processes
The typical photographs (when the flat flame was formed and the powder was stabilized) of the DT bituminous coal and JC semi-anthracite coal devolatilizing in MA and MO atmospheres were shown in Figure 3.These photographs only captured the envelope flame without the unburned char particles, due to the absence of a high-brightness background light source [35,41,47].It was clear that both the spot brightness and size of DT bituminous coal were larger than those of JC semi-anthracite coal.To obtain more detailed particle information, a series of image processing procedures for the optical intensities and sizes of the discernible spots were conducted using the Matlab software.More than 500 consecutive images were chosen to obtain sufficient particles to reduce the statistical error.
Figure 4a,b showed the mean optical intensity curves and mean particle size distributions of DT bituminous coal and JC semi-anthracite coal in two atmospheres.As shown in Figure 4, the positions of the critical points of optical intensity curves and particle size distributions were similar.The ignition and volatile combustion stages could be identified through the critical point in the optical intensity curves [33,42].Clearly, JC semi-anthracite coal was ignited later than DT bituminous coal and the volatile burnout times of JC semianthracite coal were significantly shorter than those of DT bituminous coal.Both the ignition delay times and volatile burnout times of the two coals under MO conditions were longer than those under MA conditions.Figure 4a,b showed the mean optical intensity curves and mean particle size distributions of DT bituminous coal and JC semi-anthracite coal in two atmospheres.As shown in Figure 4, the positions of the critical points of optical intensity curves and particle size distributions were similar.The ignition and volatile combustion stages could be identified through the critical point in the optical intensity curves [33,42].Clearly, JC semi-anthracite coal was ignited later than DT bituminous coal and the volatile burnout times of JC semianthracite coal were significantly shorter than those of DT bituminous coal.Both the ignition delay times and volatile burnout times of the two coals under MO conditions were longer than those under MA conditions.
(a) To reveal the observable changes in the ignition and volatile combustion behaviors, it was important to obtain the particle temperature.The particle temperature was obtained through CPDCP and CPDCP-CO 2 codes, which were described in Appendix A. The particle energy conservation equation during the heat-up and devolatilization stages could be described by the following equation: where T p , T w , and T g were the particle, wall, and gas temperatures, respectively; v p , m p , c p , A p , and ε p were the velocity, mass, heat capacity, surface area, and emissivity of particle, respectively; σ was the Stefan-Boltzmann constant and B was the blowing factor.B = (c pg (−dm p /dt)/2πd p k g ), where d p was the particle diameter and k g was the gas thermal conductivity.h was the coefficient for convection heat transfer.For the present conditions, with low particle slip velocity, h could be calculated to assume a Nusselt number (Nu = hd p k g −1 ) of 2 [48].∆H pyr was the heat of pyrolysis.Figure 5 showed the particle temperature histories of the two studied coals under MA and MO conditions at the gas temperatures in Figure 2. Negligible differences existed in the temperature rises of an inert particle under MA and MO conditions, owing to the ratio of the thermal conductivities for CO 2 and N 2 being close to one [10].Hence, the start times of the devolatilization process of pulverized coal particles under MA and MO conditions were consistent.However, the high specific heat of CO 2 and its tendency to suppress radical formation increased the ignition delay of pulverized coal particles under the MO atmosphere [38].
distributions were similar.The ignition and volatile combustion stages could be identified through the critical point in the optical intensity curves [33,42].Clearly, JC semi-anthracite coal was ignited later than DT bituminous coal and the volatile burnout times of JC semianthracite coal were significantly shorter than those of DT bituminous coal.Both the ignition delay times and volatile burnout times of the two coals under MO conditions were longer than those under MA conditions.To reveal the observable changes in the ignition and volatile combustion behaviors, it was important to obtain the particle temperature.The particle temperature was obtained through CPDCP and CPDCP-CO2 codes, which were described in Appendix A. The particle energy conservation equation during the heat-up and devolatilization stages could be described by the following equation: where Tp, Tw, and Tg were the particle, wall, and gas temperatures, respectively; vp, mp, cp, Ap, and εp were the velocity, mass, heat capacity, surface area, and emissivity of particle, respectively; σ was the Stefan-Boltzmann constant and B was the blowing factor.B = (cpg(−dmp/dt)/2πdpkg), where dp was the particle diameter and kg was the gas thermal conductivity.h was the coefficient for convection heat transfer.For the present conditions, with low particle slip velocity, h could be calculated to assume a Nusselt number (Nu = hdpkg −1 ) of 2 [48].∆Hpyr was the heat of pyrolysis.Figure 5 showed the particle temperature histories of the two studied coals under MA and MO conditions at the gas temperatures in Figure 2. Negligible differences existed in the temperature rises of an inert particle under MA and MO conditions, owing to the ratio of the thermal conductivities for CO2 and N2 being close to one [10].Hence, the start times of the devolatilization process of pulverized coal particles under MA and MO conditions were consistent.However, the high specific heat of CO2 and its tendency to suppress radical formation increased the ignition delay of pulverized coal particles under the MO atmosphere [38].Once the volatile cloud was ignited, a diffusion flame was formed to consume the evolved volatiles and provide additional heating to the particles.The combustion process of the volatile could be described using the quasi-steady droplet combustion theory [39].The mass consumption rate was expressed as follows: where rs was the droplet diameter, ρs was the fuel vapor density at the droplet surface, D Once the volatile cloud was ignited, a diffusion flame was formed to consume the evolved volatiles and provide additional heating to the particles.The combustion process of the volatile could be described using the quasi-steady droplet combustion theory [39].The mass consumption rate was expressed as follows: where r s was the droplet diameter, ρ s was the fuel vapor density at the droplet surface, D was the mass diffusivity of the fuel vapor, and the Spalding transfer number, SP, was given by where T ∞ and T s were the bulk gas temperature and the droplet surface temperature; Y O,∞ was the mass fraction of oxygen in the bulk gas; OF was the stoichiometric mass oxygenfuel ratio; h c and h v were the heat of combustion of the fuel vapor and vaporization (or devolatilization, in the case of coal) of the drop; and the relevant c v here was heat capacity of the volatilized fuel.
The change in the volatile consumption rate was mainly dependent on the diffusion rate of the volatile when N 2 was replaced by CO 2 .The diffusivity of CH 4 (representing the volatiles) in CO 2 was 20% lower than in N 2 at 1200 K [38].However, as shown in Figure 4, the increase in volatile consumption time was more than 20% for DT bituminous coal and approximately 20% for JC semi-anthracite coal.This suggested that the effect of CO 2 was dependent on the coal rank because more tar was contained in the volatile of bituminous coal.A higher molecular weight means a lower diffusion rate.
In addition to the optical intensity of the coal particles, the particle size distributions under MA and MO conditions also showed significant differences.As mentioned above, the spot image captured by the high-speed camera was actually the envelope flame around the pulverized coal particles.The change of spot size represented the moving of the flame front.The flame front appeared in the position of maximum burning rate, where the reactant mass fraction was stoichiometric.According to the moving flame front (MFF) model [49], the position of the flame front could be calculated using the following formula: where b was the radius of the flame front, a was the radius of the particle, ρ s was the density of the volatile, D was the diffusion coefficient of the volatile, Y O,∞ was ambient oxygen concentration, η was the stoichiometric of the volatile combustion, and dm v /dt was the release rate of the volatile.According to Equation (4), the position of the flame front was mainly affected by the diffusion coefficient and the releasing rate of the volatile.All particle size distributions showed a single peak trend, which were consistent with the variation of the releasing rate of the volatile.When N 2 was replaced by CO 2 , the reducing diffusion rate of the volatile resulted in the expansion of the diffusion flame front.Further, the reducing releasing rate of the volatile, caused by the lower diffusion rate of the volatile and oxygen, may result in the shrinkage of the flame front, which is consistent with the observed phenomenon in Figure 4.In general, the reduction in size of the envelope flame was mainly caused by the reducing diffusion rate when N 2 was replaced by CO 2 .

Apparent Volatile Yields
The apparent volatile yield (V daf ) was calculated by the ash tracer, using the following equation: where x a,0 and x a were weight fractions of ash in the raw coal and in the partially reacted chars, respectively.It was assumed that the ashes did not suffer any further transformation in the reactor than that undergone during the ISO ashing test (ISO 17246; 2010).Figure 6 showed the apparent volatile yields for two coals devolatilized under MA and MO atmospheres.The experimental apparent volatile yields of both coals were higher than their volatile yields obtained with the proximate analysis (ASTM).In addition, Figure 6 displayed a decrease in the volatile yield for DT bituminous coal and an increase in the volatile yield for JC semi-anthracite coal when N 2 was replaced by CO 2 .The result for DT bituminous coal was different from those results reported in Refs.[30,31] for high volatile coals devolatilized with a long residence time that could lead to an additional Boudouard reaction (C+CO 2 ↔ 2CO) after the completion of the devolatilization process.chars, respectively.It was assumed that the ashes did not suffer any further transformation in the reactor than that undergone during the ISO ashing test (ISO 17246;2010).
Figure 6 showed the apparent volatile yields for two coals devolatilized under MA and MO atmospheres.The experimental apparent volatile yields of both coals were higher than their volatile yields obtained with the proximate analysis (ASTM).In addition, Figure 6 displayed a decrease in the volatile yield for DT bituminous coal and an increase in the volatile yield for JC semi-anthracite coal when N2 was replaced by CO2.The result for DT bituminous coal was different from those results reported in Ref. [30,31] for high volatile coals devolatilized with a long residence time that could lead to an additional Boudouard reaction (C+CO2 ↔ 2CO) after the completion of the devolatilization process.
Combined with previous studies [29][30][31][32], the apparent volatile yield was highly dependent on the particle residence time.For DT bituminous coal, the selected particle residence time of 65 ms in this study only covered the devolatilization process under MA condition and was much shorter than the devolatilization time under MO condition.The incomplete devolatilization of DT bituminous coal in an MO atmosphere resulted in a lower apparent volatile yield.For JC semi-anthracite coal, the selected particle residence times were longer than the devolatilization times under the two atmospheres.This was similar to previous studies [30,31] with longer residence times, in which the gasification reaction resulted in more weight loss.Combined with previous studies [29][30][31][32], the apparent volatile yield was highly dependent on the particle residence time.For DT bituminous coal, the selected particle residence time of 65 ms in this study only covered the devolatilization process under MA condition and was much shorter than the devolatilization time under MO condition.The incomplete devolatilization of DT bituminous coal in an MO atmosphere resulted in a lower apparent volatile yield.For JC semi-anthracite coal, the selected particle residence times were longer than the devolatilization times under the two atmospheres.This was similar to previous studies [30,31] with longer residence times, in which the gasification reaction resulted in more weight loss.

Element Releases during the Devolatilization Process
In this study, elemental compositions of the volatiles were characterized using the rates of the major elements entering the gas phase during the devolatilization process.The fraction of the element released was calculated by the ash tracer based on the ultimate analysis of the particular raw coal and char.It was assumed that all the ash remained in the char and all moisture released into the volatile.The fraction of the initial amount of element i released in the gas phase (volatile), f i , was given as follows: where x i and x i,0 were the fractions of the element i in the partially reacted char and in the raw coal, respectively.Figure 7 showed the extent of the release of carbon, hydrogen, nitrogen, and sulfur relative to the apparent volatile yields during the devolatilization processes of DT and JC coals under MA and MO conditions, labeled as DT-MA, DT-MO, JC-MA, and JC-MCO, respectively.The release of oxygen was not discussed because of the large accumulated error.The apparent volatile yield was the total mass release rate on a dry and ash-free basis for the coal particles in the devolatilization process.The line in Figure 7 represented the total mass release of the combustible matter.Points below the line indicated the release rates being slower than the total mass release rate, and points above the line indicated the release rates being faster than the total mass release rate.i where x i and x i,0 were the fractions of the element i in the partially reacted char and in the raw coal, respectively.
Figure 7 showed the extent of the release of carbon, hydrogen, nitrogen, and sulfur relative to the apparent volatile yields during the devolatilization processes of DT and JC coals under MA and MO conditions, labeled as DT-MA, DT-MO, JC-MA, and JC-MCO, respectively.The release of oxygen was not discussed because of the large accumulated error.The apparent volatile yield was the total mass release rate on a dry and ash-free basis for the coal particles in the devolatilization process.The line in Figure 7 represented the total mass release of the combustible matter.Points below the line indicated the release rates being slower than the total mass release rate, and points above the line indicated the release rates being faster than the total mass release rate.In Figure 7, for DT bituminous coal, every element was released at a rate proportional to the total mass release rate on an ash-free basis.This phenomenon was consistent with the incomplete release of the volatile.The release rate of carbon was lower than that of the volatile because carbon was the principal component of the coal.Other elements were released faster than the volatile, as a result of the bituminous coal being present primarily in labile peripheral groups that were labile and more readily released than aromatic carbon.Compared with the DT bituminous coal, both the release rates of hydrogen and oxygen of JC semi-anthracite coal were close to the release rate of the volatile, which was due to the less labile peripheral groups in the semi-anthracite coal.In addition, it was noteworthy that for JC coal the increase in the release rate of carbon was slightly larger than that of the volatile when N2 was replaced by CO2.This further confirmed the occurrence of the char-CO2 gasification reaction under MO conditions, which led to an increase in apparent volatile yield.Simultaneously, the gasification reaction also led to the release of more char-N, which was also reported in previous study [50].This phenomenon suggested that the staged combustion mode could be used in the MILD oxy-combustion of anthracite coal to further reduce NOx emissions.In Figure 7, for DT bituminous coal, every element was released at a rate proportional to the total mass release rate on an ash-free basis.This phenomenon was consistent with the incomplete release of the volatile.The release rate of carbon was lower than that of the volatile because carbon was the principal component of the coal.Other elements were released faster than the volatile, as a result of the bituminous coal being present primarily in labile peripheral groups that were labile and more readily released than aromatic carbon.Compared with the DT bituminous coal, both the release rates of hydrogen and oxygen of JC semi-anthracite coal were close to the release rate of the volatile, which was due to the less labile peripheral groups in the semi-anthracite coal.In addition, it was noteworthy that for JC coal the increase in the release rate of carbon was slightly larger than that of the volatile when N 2 was replaced by CO 2 .This further confirmed the occurrence of the char-CO 2 gasification reaction under MO conditions, which led to an increase in apparent volatile yield.Simultaneously, the gasification reaction also led to the release of more char-N, which was also reported in previous study [50].This phenomenon suggested that the staged combustion mode could be used in the MILD oxy-combustion of anthracite coal to further reduce NO x emissions.

Heating Value of the Volatiles
If coal samples were first partly pyrolyzed and then the residual char and volatile products were burned separately to CO 2 and H 2 O, the total heat resulting from the thermal decomposition of the coal could be given by: ∆H C,coal = ∆H C,char + ∆H C,volatile + ∆H pyrolysis (7) where ∆H was the heat change (increment of enthalpy H) of individual reactions, the subscript 'pyrolysis' was the pyrolysis reaction, the subscript 'c,coal', 'c,char', and 'c,volatile' were the combustion reaction of coal, char, and volatile, respectively.The heat of pyrolysis was relativity small, less than 10% of the heat of coal combustion [51].Therefore, the heat of pyrolysis could be neglected.The total heat resulting from the thermal decomposition of the coal would be given as follows: On this basis, the heating values of the volatiles were calculated based on the heating values of the raw coals and chars, which were measured using a Parr automatic isoperibol calorimeter.The following two formulas were used for calculating the heating values of the volatiles: where Q v.daf was the heating value per kg of the volatile, Q v was the heating value of the volatile per kg of raw coal, and Q coal.daf and Q char.daf were the heating values per kg of raw coal and char, respectively.Figure 8 displayed the correlation between the heating values per kg of the volatile (Q v.daf ) and the atomic C/H ratio of the volatile.As expected, it was observed that the heating values per kg of the volatile showed a decreasing trend with an increase in the atomic C/H ratio of the volatile.As mentioned above, the chars obtained from DT bituminous coal under MO and MA conditions represented two different degrees of decomposition.The atomic C/H ratio of the volatile increased with the progress of devolatilization because the atomic C/H ratio of the aromatic rings was greater than that of the breakable labile peripheral groups.Therefore, the heating value per kg of the volatile released from DT coal under MA conditions was lower than that under MO conditions.In contrast, the heating value per kg of the volatile released from JC coal under an MO atmosphere was also lower than that under an MA atmosphere, and this was because the char-CO 2 gasification reaction led to more carbons being released and this further increased the atomic C/H ratio of the volatile.
script 'pyrolysis' was the pyrolysis reaction, the subscript 'c,coal', 'c,char', and 'c,volatile' were the combustion reaction of coal, char, and volatile, respectively.The heat of pyrolysis was relativity small, less than 10% of the heat of coal combustion [51].Therefore, the heat of pyrolysis could be neglected.The total heat resulting from the thermal decomposition of the coal would be given as follows: On this basis, the heating values of the volatiles were calculated based on the heating values of the raw coals and chars, which were measured using a Parr automatic isoperibol calorimeter.The following two formulas were used for calculating the heating values of the volatiles: (10) where Qv.daf was the heating value per kg of the volatile, Qv was the heating value of the volatile per kg of raw coal, and Qcoal.daf and Qchar.daf were the heating values per kg of raw coal and char, respectively.Figure 8 displayed the correlation between the heating values per kg of the volatile (Qv.daf) and the atomic C/H ratio of the volatile.As expected, it was observed that the heating values per kg of the volatile showed a decreasing trend with an increase in the atomic C/H ratio of the volatile.As mentioned above, the chars obtained from DT bituminous coal under MO and MA conditions represented two different degrees of decomposition.The atomic C/H ratio of the volatile increased with the progress of devolatilization because the atomic C/H ratio of the aromatic rings was greater than that of the breakable labile peripheral groups.Therefore, the heating value per kg of the volatile released from DT coal under MA conditions was lower than that under MO conditions.In contrast, the heating value per kg of the volatile released from JC coal under an MO atmosphere was also lower than that under an MA atmosphere, and this was because the char-CO2 gasification reaction led to more carbons being released and this further increased the atomic C/H ratio of the volatile.Compared to the heating values per kg of the volatile, the heating values of the volatiles per kg of the two raw coals (Q v ) were shown in Figure 9, which considered both the content and the calorific value of the coal volatile.It was found that the heating value of the volatile per kg of coal was highly linear correlated to the apparent volatile yield, as the difference in the heating values per kg of the volatile was smaller than the difference in the apparent volatile yields.Furthermore, the Q v of DT coal in an MO atmosphere was lower than that in an MA atmosphere.It was expected that the Q v of JC coal in an MO atmosphere would be greater than that in an MA atmosphere since more CO was generated from the char-CO 2 gasification reaction.This indicated that the increase in the heating value of the volatile per kg of coal might improve the flame stability of anthracite coal in the flue-rich stage.
the difference in the heating values per kg of the volatile was smaller than the difference in the apparent volatile yields.Furthermore, the Qv of DT coal in an MO atmosphere was lower than that in an MA atmosphere.It was expected that the Qv of JC coal in an MO atmosphere would be greater than that in an MA atmosphere since more CO was generated from the char-CO2 gasification reaction.This indicated that the increase in the heating value of the volatile per kg of coal might improve the flame stability of anthracite coal in the flue-rich stage.

Conclusions
The devolatilization experiments on bituminous coal and semi-anthracite (volatile contents of 25% and 7%) under both MA and MO atmospheres were conducted on a flatflame assisted entrained flow reactor.The experimental conditions, including the heating rate (~10 5 K/s) and residence time (65 ms), were close to those observed with typical industrial boilers.A direct observation method and an indirect analysis method were both used to quantitatively evaluate the volatile releasing and burnout characteristics of pulverized coal particles, such as the volatile burnout time, the apparent volatile yields, element releases, and heating values of the volatile.Additionally, the physical and chemical reaction mechanisms and the effect of CO2 on the devolatilization process of pulverized coals under MA and MO conditions were discussed.The following conclusions could be drawn: (i) The presence of CO2 prolonged the burnout time of the volatile (more than 20%) and decreased the size of the volatile envelope flame.The lower diffusion rate of the volatile in the CO2 mixture was considered to be the main reason for these observations.Further, the effect of CO2 was found to be dependent on the rank of the coal.(ii) The presence of CO2 reduced the release rate of the volatile, and this resulted in the apparent volatile yield (27%) and the heating value of the volatile (2%) of bituminous coal under MO conditions being lower than those under MA conditions.This indicated that the flame stability of bituminous coal under MIO conditions was not as good as that under MA combustion conditions.

Conclusions
The devolatilization experiments on bituminous coal and semi-anthracite (volatile contents of 25% and 7%) under both MA and MO atmospheres were conducted on a flatflame assisted entrained flow reactor.The experimental conditions, including the heating rate (~10 5 K/s) and residence time (65 ms), were close to those observed with typical industrial boilers.A direct observation method and an indirect analysis method were both used to quantitatively evaluate the volatile releasing and burnout characteristics of pulverized coal particles, such as the volatile burnout time, the apparent volatile yields, element releases, and heating values of the volatile.Additionally, the physical and chemical reaction mechanisms and the effect of CO 2 on the devolatilization process of pulverized coals under MA and MO conditions were discussed.The following conclusions could be drawn: (i) The presence of CO 2 prolonged the burnout time of the volatile (more than 20%) and decreased the size of the volatile envelope flame.The lower diffusion rate of the volatile in the CO 2 mixture was considered to be the main reason for these observations.Further, the effect of CO 2 was found to be dependent on the rank of the coal.(ii) The presence of CO 2 reduced the release rate of the volatile, and this resulted in the apparent volatile yield (27%) and the heating value of the volatile (2%) of bituminous coal under MO conditions being lower than those under MA conditions.This indicated that the flame stability of bituminous coal under MIO conditions was not as good as that under MA combustion conditions.(iii) Although the presence of CO 2 could affect the volatile flame size of semi-anthracite coal, the effect on volatile burnout time was less clear.The char-CO 2 gasification increased the apparent volatile yield and the heating value of the volatile by 47%.
The char gasification reaction also led to more char-N by 19% being released into volatile.The higher heating value of the volatile improved the flame stability in the fuel-rich stage.In this paper, the chemical structure parameters were obtained by the nonlinear correlation developed by Genetti and Fletcher [53].The kinetic parameters that controlled the rate of devolatilization used in the model were assumed to be coal independent [54].The structural parameters and kinetic parameters were listed in Tables A2 and A3, respectively.

Figure 1 .
Figure 1.Schematic diagram of the flat-flame assisted entrained flow reactor.

Figure 1 .
Figure 1.Schematic diagram of the flat-flame assisted entrained flow reactor.

Figure 2 .
Figure 2. Measured gas temperature profiles along the centerline of the reactor.

Figure 2 .
Figure 2. Measured gas temperature profiles along the centerline of the reactor.

Figure 3 .
Figure 3.Typical photographs of (a) DT bituminous coal, and (b) JC semi-anthracite coal, devolatilizing in MA and MO atmospheres by FASTCAM SA4.

Figure 3 .
Figure 3.Typical photographs of (a) DT bituminous coal, and (b) JC semi-anthracite coal, devolatilizing in MA and MO atmospheres by FASTCAM SA4.

Figure 4 .
Figure 4. Optical intensity curves and particle size distributions of (a) DT bituminous coal and (b) JC semi-anthracite coal.

Figure 4 . 16 Figure 5 .
Figure 4. Optical intensity curves and particle size distributions of (a) DT bituminous coal and (b) JC semi-anthracite coal.Processes 2022, 10, x FOR PEER REVIEW 9 of 16

Figure 5 .
Figure 5. Particle temperature histories calculated by the CPDCP code.

Figure 6 .Figure 6 .
Figure 6.Apparent volatile yields of two coals under MA and MO conditions.

Figure 7 .
Figure 7. Element releases during the devolatilization processes of the two coals investigated.

Figure 7 .
Figure 7. Element releases during the devolatilization processes of the two coals investigated.

Figure 8 .
Figure 8. Heating values per kg of the volatiles (Q v.daf ) as a function of the atomic C/H ratios of the volatiles.

Figure 9 .
Figure 9. Heating values of the volatiles per kg of raw coal (Qv) as a function of the apparent volatile yields.

Figure 9 .
Figure 9. Heating values of the volatiles per kg of raw coal (Q v ) as a function of the apparent volatile yields.

Table 1 .
Properties of DT and JC coals.

Table 2 .
Calculated composition of the combustion products (mol %).

Table 2 .
Calculated composition of the combustion products (mol %).

Table A2 .
Chemical Structure Parameters obtained from a NMR correlation.

Table A3 .
Rate Parameters Used in the CPDCP code.