Effect of Coal Blending Methods with Different Excess Oxygen on

Article pubs.acs.org/EF

Effect of Coal Blending Methods with Different Excess Oxygen on Unburned Carbon and NOx Emissions in an Entrained Flow Reactor Byoung-Hwa Lee,† Eric G. Eddings,‡ and Chung-Hwan Jeon*,† †

School of Mechanical Engineering, Pusan Clean Coal Center, Pusan National University, Republic of Korea Department of Chemical Engineering and Institute for Clean and Secure Energy, The University of Utah, Salt Lake City, Utah 84112, United States

‡

ABSTRACT: The influence of coal blending methods, such as out-furnace (external or premixed) blending and in-furnace (initially nonmixed) blending, with different excess oxygen (highest, medium, and lowest stoichiometric conditions) on unburned carbon and NOx emissions of blend combustion in an entrained flow reactor has been analyzed, using experimental and numerical approaches for binary coals used by Korean power plants. The results confirm that, under the medium condition, contrasting processes, such as reactive and unreactive effects, occur with SBRs in the out-furnace blending method. The infurnace blending method results in an improvement in the efficiency of unburned carbon fractions and a further reduction in the NOx emission. Under the highest condition, the unburned carbon fraction in both the out-furnace and the in-furnace blending methods corresponds with the tendency under the medium condition with contrasting processes of lower magnitude, whereas the NOx emission in the highest condition increases slightly. Under the lowest conditions, the unburned carbon fraction in the out-furnace blending method gradually decreases as SBR decreases, without a competition effect. The reduction of NOx emission under the lowest conditions is more effective than those under other conditions for the two blending methods because of homogeneous and heterogeneous NOx reduction mechanisms. These results show that the phenomenon that occurs with coal blending methods under different excess oxygen conditions has been demonstrated and the in-furnace blending method below medium conditions would be an effective method to improve combustibility and NOx emission due to penalty of NOx under the highest condition. In general, the numerical results are in agreement with the measured values and give insight into the phenomena that affect the blending methods under different excess oxygen conditions. of blended coal. The effects of firing blended coal on ignition behavior, char burnout, slagging, fouling, and NOx emissions cannot not be easily inferred by the additive method, which is calculated averagely from the properties of each component. Several laboratory and bench-scale devices, such as thermogravimetric analyzers, drop-tube furnaces, and fuel evaluation test facilities, have been used to make a realistic assessment of the combustion behavior of blends. Thermogravimetric analysis has been extensively used in studying the combustion behavior of coal and char and has also been used to assess the relative burning properties of coal blends, blends of coal and sludge, etc.6,7 Also, DTF studies on characteristics of reactivity and emission of coals and their blends are wellreported.8−10 However, most studies on blending have focused on the combustion characteristics of the out-furnace blending method. Regarding blending methods, two different coal blending methods can be used in a power plant: blending outside of the furnace (blending out in the coal yard) and blending inside the furnace. In blending outside of the furnace, the different types of coal are simultaneously injected into a boiler after being mixed together prior to injection. For blending inside of the furnace, each type of coal is injected into the boiler from a separate burner with no prior mixing.

1. INTRODUCTION Utilization of coal blends is becoming increasingly common in pulverized fuel (PF) and fluidized bed combustion (FBC) power plants. It would be reasonable to improve the usability of coal blends by reducing fuel costs, controlling emission limits and flame instability possibly due to differences in ignition characteristics of each coal, enhancing fuel flexibility, extending the range of acceptable coals, providing a uniform product from coals of varying quality, improving boiler performance, and solving existing problems, such as poor carbon burnout, slagging, and fouling. The possibility of improving performance in all of these areas has made the combustion of coal blends the subject of many recent studies.1−3 Coal is a complex substance with divergent properties with respect to rank, maceral composition, and associated impurities. Therefore, it is rather difficult to formulate a perfect methodology to predict combustion behavior of coals and coal blends. Some of the properties of a blend, such as proximate and ultimate analysis data, as well as heating value, can be determined by a weighted average of the properties of the individual coals in the blend.4 Similarly, some aspects of the combustion behavior of blended coals in power stations are known and can be determined reasonably well from knowledge about the properties of the component coals in the blend and their respective mass fraction (i.e., proximate and ultimate analysis, SO2 emissions),5 which is known as additive phenomena. However, it is nonadditive phenomena of some properties that cause concern about predicting the performance © 2012 American Chemical Society

Received: May 9, 2012 Revised: October 22, 2012 Published: October 23, 2012 6803

dx.doi.org/10.1021/ef300562t | Energy Fuels 2012, 26, 6803−6814

Energy & Fuels

Article

Ikeda11 experimentally investigated the combustion and emission of coal blending using out-furnace and in-furnace blending methods using pilot-scale three-stage burners. He successfully burned coal blends using the in-furnace blending method based on operational experience and achieved reduction in both unburned carbon and NOx emissions; however, the mechanisms and fundamental combustion behaviors of unburned carbon and NOx emission related to out-furnace and in-furnace blending methods were not accurately represented in their experiments. Subsequently, Lee et al.12 carried out a fundamental examination of the nonadditive phenomena that occur during the combustion of coal blends when using a simulated outfurnace blending method, along with a detailed look at the mechanisms that affect the in-furnace blending method when using an entrained flow reactor (EFR). The feeding system of EFR was modified, which can inject binary coal simultaneously under the high-temperature conditions contrary to the previous coal feeding system. In other words, two types of coals were injected through separate injectors into the furnace, and their blending time was controlled by varying the distance between the two injectors to simulate out-furnace and in-furnace blending methods of a real power plant. However, since Lee’s study was conducted under only one condition, and explained the behavior of unburned carbon and NOx emissions under just normal combustion operating conditions, there is still a need to investigate coal-blending phenomena, and the underlying mechanisms related to the effect of the initially in-furnace blending method in particular. One of the major requirements for the operation of PF and FBC power plants is an optimal amount of excess combustion air to provide the highest possible thermal efficiency with a minimal impact on the environment. For PF power plants in particular, it is one of the key operating variables, affecting simultaneously the boiler’s thermal efficiency, operational reliability, and environmental performance (emissions from the unit).13 Accordingly, this study was conducted to fundamentally investigate the effect of excess oxygen on unburned carbon and NOx emissions for different coal blending methods, such as out-furnace blending and in-furnace blending, using EFR. This paper has extended the investigations into the effect of coal blending methods with different excess oxygen, particularly, compared to the previous studies.12

Figure 1. Schematic diagram of the modified EFR apparatus.

Figure 2. Detailed schematic diagram of in-furnace blending feeder apparatus.

2. EXPERIMENTAL SECTION

The feeding system was mounted above the injection probe, and the length between the feeders was adjustable from 0 to 7 cm along the axial direction of the injection probe, which means that the adjustment of the tube length controlled the mixing time of the binary coals. The maximum tube length was limited to 7 cm due to the position where the volatile combustion actively occurs, as well as by geometrical limitation. The results of Lee et al.15 are provided in the right half of Figure 2, where the experimental results for the variation in mass fraction of the coals have been compared with the numerical results of 2-competing and CPD models for devolatilization. The figure shows that the characteristics of a typical coal combustion process in which devolatilization occurs with rapid mass reduction in a short time within 7 cm in this EFR condition and char oxidation follows it with mass reduction over a relatively longer time. The devolatilization processes with the two models seem to differ somewhat from each other in the initial region. The comparison with the experimental results shows that the results obtained with from the CPD submodel are more accurate. The distance between the injection tubes in the out-furnace blending method was 0 cm, and the distance between the tips of the injection

Figure 1 presents a schematic diagram of the modified EFR apparatus used to conduct this study; the EFR was an entrained flow reactor (60.0 cm long with an internal diameter of 7.0 cm) designed at the Pusan Clean Coal Center (Korea). A detailed description of the experimental setup for the EFR has been provided in ref 14. This EFR was modified to examine the effect of coal blending using out-furnace and in-furnace blending methods, as shown in Figure 1. Contrary to the previous coal feeding system of the EFR, the modified feeding system used here could inject binary coals under high-temperature conditions by using a water-cooled injection probe. The mixing time between binary coals could be varied, as will be described below, and two feeders were installed in the injecting section to feed different types of coal. Figure 2 shows a schematic diagram of the in-furnace blending feeder apparatus, and it describes the area within the red line of Figure 1 in more detail. One coal was supplied directly into the side of the injection probe, and another coal was fed into a center tube installed with a cooling system. The detailed dimensions for the infurnace blending feeder are described in Figure 2. 6804


Energy & Fuels

Article

Table 1-1. Proximate and Ultimate Analyses for Coals Used in This Study proximate (wt %, as rec'd)

a

ultimate (wt %, DAF)

coals

Ma

VMa

FCa

ash

C

H

O

N

S

FR (FC/VM)

Yakutugol (bituminous coal) Adaro (sub-bituminous coal)

1.67 5.22

17.94 50.67

68.68 39.92

11.71 4.19

88.46 74.08

4.5 5.91

6.06 18.67

0.75 1.27

0.14 0.07

3.83 0.8

M, moisture; VM, volatile matter; FC, fixed carbon.

Table 1-2. Mineral Compounds in Ashes for Coals coals

SiO2

Al2O3

Fe2O3

CaO

MgO

Na2O

K2O

TiO2

others

sum

Yakutugol (bituminous coal) Adaro (sub-bituminous coal)

60.21 49.07

24.17 22.18

8.31 9.78

0.26 9.51

1.63 4.17

0.10 0.38

1.08 0.93

0.63 0.57

3.60 3.41

100.0 100.0

tubes in the in-furnace blending method was adjusted to 3, 5, and 7 cm, which corresponds to mixing times of 0.28, 0.45, and 0.63 s, respectively. The local temperature of the reacting coal was dependent upon the reactor wall temperature, which was set to a value of 1300 °C for these tests. In the experiments, the total feeding rate for coal blending was held constant at 0.24 g/min, and the blending ratios of sub-bituminous coal on the basis of total weight percentage (SBR) were 0, 25, 50, 75, and 100%, where 0% refers to feed of bituminous coal only and 100% refers to feed of sub-bituminous coal only. Coal particle size was 90− 150 μm (based on the Rosin−Rammler distribution), and the particle residence time was about 2.21 s, which was calculated from the drag force and the gravitational force of coal particles.16 The total gas flow rate mixed with O2 and N2 was 5 L/min (at 273 K, 1 atm), and the O2 concentration prior to the reaction was controlled at 9, 12, and 13.5% to represent lowest, medium, and highest stoichiometric conditions, respectively. The oxygen concentration at the exit section of the EFR under the medium stoichiometric condition was kept constant at 4− 5% (by volume, on a dry basis), because this is a typical value when including overfired air at the furnace exit in large PF boilers.17 The lowest and highest stoichiometric conditions represented a lower oxygen level and enhanced excess air levels than that of the medium condition. At this time, the excess oxygen coefficient (λ) for the medium stoichiometric condition was 1.3−1.5; for the lowest stoichiometric condition, 1.0−1.2; and for the highest stoichiometric condition, 1.5−1.7 with SBR due to coal elements. Here, the stoichiometric coefficient (λ) was defined as the ratio of the actual O2/coal ratio to the stoichiometric O2/coal ratio for a given mixture. The stoichiometric O2/coal ratio was determined by considering all combustible elements in the coal. The stoichiometric requirement of O2 volume flow for the combustion of 1 kg of coal was calculated using the elements C, H, N, S, and O, which were obtained as weight percent of carbon, hydrogen, nitrogen, sulfur, and oxygen, respectively, in the coal on an as-received basis and were then used as input parameters.18 A thermogravimetric analyzer (TGA) and portable gas analyzer (Eurotron Gas analyzer) were used to calculate the unburned carbon fraction with the ash tracer method and to measure the concentration of NOx. The unburned carbon fraction of ash tracer method19 can be expressed as

⎛ Ash uc − Ash rc ⎜ Unburned carbon fraction (%)=⎜1 − Ash ⎜ Ash uc 1 − 100rc ⎝

(

)

confidence intervals (CIs) by the mean values. The uncertainties of NOx emission factors for different conditions varied from 10% to 25%. The coals used in this study were Yakutugol (Russia) for bituminous coal and Adaro (Indonesia) for sub-bituminous coal, which are commonly utilized in Korean power plants. The results of the proximate and ultimate analyses for each coal are shown in Table 1-1, and the mineral compounds in ashes for coals are shown in Table 1-2. The input parameters considered in this study are given in Table 2.

Table 2. Experimental Conditions Considered in This Study input parameters coal feeding rate (g/min) total flow rate (lpm) coal size (μm) bituminous coal sub-bituminous coal SBR (%) injection distance from inlet (cm) EFR setting temperature (°C) stoichiometric coefficient (λ)

0.24 5 90−150 (based on Rosin−Rammler distribution) Yakutugol Adaro 0, 25, 50, 75, 100 out-furnace blending method: 0 in-furnace blending method: 3, 5, 7 1300 lowest condition 1−1.2 medium condition 1.3−1.5 highest condition 1.5−1.7

3. NUMERICAL SECTION The furnace geometry was modeled with a two-dimensional (2D) mesh composed of 41 000 cells. Figure 3 presents the gas flow and injection points of binary coals showing relative geometries and comparison of temperature between numerical modeling and experiment of the EFR used during simulation. The injection of the binary coals with carrier gas was downward, and the internal gas temperature was constant. The experimental conditions used in this study are applied as boundary conditions of the numerical simulation. Numerical modeling was conducted using the commercial CFD software FLUENT, code version 6.20 The model was based on the Navier−Stokes equations for gas and particle phases. The gas phase was modeled in an Eulerian domain, whereas the particles were tracked in a Lagrangian fashion. The model consisted of submodels for the turbulent fluid mechanics, gaseous combustion, particle dispersion, reactions (i.e., moisture evaporation, coal devolatilization, and char burnout), and radiation. A discrete ordinates (DO) model was used to model the radiation heat transfer in the furnace, as it is suitable for a furnace with a short optical length. Two separate groups of coal particle injections were used, enabling the

⎞ ⎟ ⎟ × 100 ⎟ ⎠

where Ashuc and Ashrc represent the ash contents in the unburned and raw char, respectively. Generally, the ash content analyses of coal and unburned char were carried out two or three times in this study, and the average difference between the values obtained from the repeated analyses was less than 0.15%. In addition, the uncertainty of the NOx measurement result was ±10%, which was judged from the difference in NOx, which was measured twice at the same burning conditions. The quantitative uncertainties of the NOx emission were calculated by dividing the differences between the upper and lower limits of the 95% 6805


Energy & Fuels

Article

Figure 3. Gas flow and injection points of binary coals showing relative geometries (left side) and comparison of temperature between numerical modeling and experiment of the EFR.

Table 3. Calculated 13C NMR and the Fuel N Partitioning from CPD Submodel, and the Apparent Kinetic Parameters for Char Oxidation input parameters for devolatilization rate (calcd 13C NMR) Yakutugol (bitu.) Adaro (sub-bitu.)

apparent kinetic parameters for char oxidation

p0

c0

σ+1

Mw,1

Mw,δ

volatile N

char N

A (g/cm2·s Pa)

E (J/kmol)

0.725 0.402

0.304 0

4.333 4.974

240.99 457.95

20.31 39.68

0.43 0.69

0.57 0.31

0.0014 0.2

8.3 × 107 8.2 × 107

⎛ E ⎞ RArr = A sρ2 YFYox exp⎜ − s ⎟ ⎝ RT ⎠

tracking of the two component coals during the simulation. The material properties, such as coal density, thermal heat capacity and conductivity, radiation characteristics, devolatilization kinetics, char swelling ratio, and char combustion kinetics, were specified for the two component coals. The initial conditions (including the velocities, temperature, and particle size distribution) of the two injection groups were defined separately. The coal components (moisture, volatile matter, fixed carbon, and ash) were assumed to behave in the following manner: the moisture vaporizes after entering the furnace; the volatile matter comprises elements C, H, O, N, and S and enters the gas phase through devolatilization; and the char, that is, fixed carbon, enters the gas phase through the char combustion process. Modeling the coal particle reactions, especially the devolatilization and char oxidation, is crucial for describing both the dispersed particle phase and the continuous gas phase. The equations for these phases are coupled through the particle-mass loss rate by these reactions. 3.1. Gas-Phase Combustion Model. The gas-phase combustion model was used to simulate the generation of volatile matter from coal and to simulate CO combustion. During combustion, a global single-step reaction was assumed in which the volatile matter is converted into CO2, CO, and water vapor. In the species mass fraction transport equations, equal effective turbulent mass diffusion coefficients are set for the fuel, O2, and products. The Magnussen−Hjertager21 finiterate/eddy-dissipation model is used to quantify the turbulent combustion rates of the volatile matter and CO. The net reaction rate is determined as R = min(REBU, RArr), where REBU = CRρ

fuel N partitioning

Here, REBU and RArr are the reaction rates (kg/m3·s) for the EBU turbulent combustion model and Arrhenius reaction model, respectively; CR is an empirical constant of the gasphase k − ε turbulence model, where k is the turbulent kinetic energy, ε is the dissipation rate, and, accordingly, k/ε is the turbulent time scale; YF and Yox are the mass fractions of fuel and oxidizer, respectively; As is the pre-exponential factor (m3/ kg·s); and Es is the activation energy for the gas phase (J/ kmol). ρ is the density (kg/m3), β is the stoichiometric coefficient, and R is the universal gas constant (J/kmol·K). 3.2. Particle Devolatilization Model. The chemical percolation devolatilization (CPD) model22 was developed to describe coal devolatilization on the basis of the chemical structure of the parent coal. This model employs percolation statistics to describe the generation of light gas/tar precursors of finite size on the basis of the number of cleaved labile bonds in the infinite coal lattice. The model includes treatment of the vapor−liquid equilibrium and a cross-linking mechanism. Coalindependent kinetic parameters of the reaction rate are employed, and the coal-dependent chemical structure coefficients are set on the basis of correlations developed from 13C NMR measurements of several coals.22 In this correlation, five parameters describe the chemical structure of each coal: (1) initial fraction of bridges in the coal lattice, p0; (2) initial fraction of char bridges, c0; (3) lattice coordination number, σ + 1; (4) cluster molecular weight, Mw,1; and (5) side chain molecular weight, Mw,δ. Each parameter is used as an input for the CPD submodel, and 13C NMR parameters for the coals used in this study are shown in Table 3. From the fuel N partitioning shown in the table, it is evident that the amount of released volatile N is almost proportional to the volatile matter content and also that it is closely related to the number of side chains (Mw,δ) in particular. In other words, it is indicated that

⎛ Y ⎞ ε min⎜YF , ox ⎟ κ β ⎠ ⎝ 6806


Energy & Fuels

Article

Table 4. Models and Submodels Used in Numerical Modeling models and submodels governing equation turbulence model radiation model gases combustion model particle (components; moisture, volatile matter, char, ash)

devolatilization model char reaction model

NO emission

[(Tp + T∞)/2]0.75 dp

R 2 = A exp−(E / RTp) dm p dt

= −A p

diffusion-kinetic model (application of char oxidation kinetics for each coal) fuel NO (application of VM-N, char N calculated from CPD submodel) thermal NO (extended Zeldovich mechanism) reburning mechanism

properties obtained from the proximate and ultimate analysis and then used to calculate the N partitioning data. This model developed by Genetti et al.28 reported that the model predictions of nitrogen release compared well with measured values for most coals. The calculated 13C NMR and the fuel N partitioning results obtained from the CPD submodel in this study are shown in Table 3. Thermal NOx is the NOx produced from the reaction of N2 and O2 from air at high temperatures. The amount of NOx generated by this process increases with the temperature. This mechanism is well-known and can be described by the extended Zeldovich mechanism.29 In addition, a reburning mechanism is applied that is based on the model proposed by Kandamby et al.30 The model adds a reduction path to the De Soete global model31 that describes the NOx formation/destruction mechanism in a pulverized coal flame. The additional reduction path accounts for the NOx destruction in the fuel-rich reburn zone by CH radicals. The models and submodels used in numerical modeling of this study are given in Table 4. Numerical modeling approaches were employed to gain insights into the mechanisms involved in the blending methods with different excess oxygen.

the chemical structure of coal in the vicinity of each N atom will largely determine how an N atom will react during devolatilization and char oxidation. 3.3. Particle Char Oxidation Model. The overall reaction rate on the surface of coal particles is simulated by a diffusionkinetic model.23,24 The heterogeneous char reaction rate is assumed to be first-order in terms of the O2 concentration and CO2 concentrations. The diffusion rate coefficient R1 and the kinetic rate R2 are calculated and weighted to obtain the char combustion rate R1 = C1

Navier−Stokes equation (gas phase → Eulerian, particle → Lagrangian) standard two-equation k−ε model discrete ordinate (DO) model finite-rate/eddy-dissipation model (VM from coal and CO combustion) CPD model (application of calculated 13C NMR for each coal)

ρRT∞Yox R1R 2 M w,ox R1 + R 2

where Ap = πd2p is the surface area of the coal particle; dp, Tp, and mp are the diameter, temperature, and mass of particles; and Yox and Mw,ox are the mass fraction and the molecular weight of oxidizer, respectively; and a pre-exponential factor A and activation energy E can be measured experimentally.14 The apparent kinetic parameters of coals for the char oxidation in this study are shown in Table 3. 3.4. NOx Modeling. The NOx postprocessing package used in the coal combustion simulation can be subdivided into three main sections representing NOx formation by thermal, prompt, and fuel pathways. Prompt-NOx is not applied in this study. In addition, the partitioning of the fuel N species between the char and volatile matter for each coal can be obtained for the different coals using the CPD submodel, which was used in the NOx submodels in a postprocessing mode. The estimated average N partitioning based on blending ratio was used as an input for the blends in the NOx postprocessing package of the coal combustion simulation. Fuel-NOx was predicted using global reaction rates and included both volatile and char N. The volatile N is assumed to convert first into the intermediate species, that is, HCN (90%) and NH3 (10%), and then into N2 or NOx according to the mechanism utilized within the model.25 It is assumed that all char N transform to HCN and subsequently either to NO or to N2 by a gas-phase reaction.26,27 The partitioning of the fuel N species between the char and volatiles can be obtained for the coals using the CPD submodel, which incorporates a model to predict the amount and distribution of fuel N between the volatile matter and char on the basis of 13C NMR parameters for coals.28 The 13C NMR data were calculated using the

4. RESULTS AND DISCUSSION 4.1. Unburned Carbon Fraction and NOx Emissions for Out-Furnace Blending with Different Excess Oxygen Levels. Figure 4 shows predictions of the char reaction rate (kg/s) and O2 concentration at different SBRs and excess oxygen conditions in the out-furnace blending method. The variation of char reaction behavior for the two coals with different SBRs can be readily identified from this figure. Recall that the coal injected in the center tube is sub-bituminous and the coal injected in the annular tube is bituminous. Here, the excess oxygen level is represented by three stoichiometric conditions: highest, medium, and lowest, as described in Table 2. The exit oxygen concentration of the medium condition was adjusted to a condition similar to a real boiler condition. For a fixed coal feeding rate, is apparent that the highest condition describes an oxygen-rich condition, and the lowest condition describes a leaner oxygen scenario relative to the medium condition. The figure also shows that the reaction time for bituminous coal is longer than that of sub-bituminous coal, which is consistent with experimental results shown in Figure 5. A detailed description of the experimental results obtained for the unburned carbon fraction is explained in the next paragraph. Furthermore, the distributions of the char reaction rate and O2 concentration vary significantly with the different excess oxygen levels. The char reaction time under the lowest conditions is even longer than that under the medium 6807


Energy & Fuels

Article

Figure 4. Predictions of char reaction rate (left, kg/s) and O2 concentration (right, mole fraction) at different SBRs and excess oxygen conditions in out-furnace blending.

conditions, whereas the char reaction time under the highest conditions is shorter, as would be expected from the higher

surrounding oxygen concentration. Figure 5 shows the comparison of unburned carbon fractions between experimen6808


Energy & Fuels

Article

there exists only a reactive effect by sub-bituminous coal. This behavior can be confirmed from the simulation of the lowest condition in Figure 4a, which shows that, because subbituminous coal reacts over a longer time, an oxygen-deficient environment occurs less frequently than in the other conditions. Some of these results are consistent with those of previous researchers. Moron et al.32 investigated the relationship between unburned carbon concentration in fly ash and excess air ratio for blending coal. They indicated that, as the excess air ratio increases, the unburned carbon fraction decreases due to combustibility by O2 concentration, but they did not describe the competition between the reactive and the nonreactive effect. Figure 5 also shows that, in all conditions, unburned carbon at SBR 100% seldom exists because all of the sub-bituminous coal is oxidized. Numerical simulations under the highest and lowest conditions reveal the elaborate reaction mechanism of blending coals, and the magnitude and trends observed for the unburned carbon fraction in the numerical results are in reasonable agreement with the experimental values. Figure 6 shows a comparison of NOx emission between experimental and numerical results as a function of SBR at

Figure 5. Comparison of unburned carbon fractions as a function of SBR at three different excess oxygen conditions in out-furnace blending.

tal and numerical results as a function of SBR at three different excess oxygen conditions in out-furnace blending. The experimental results show that, as the SBRs for the medium condition increase up to 50%, the unburned carbon fraction gradually decreases. However, the unburned carbon fraction first increases again and then decreases with a peak at an SBR of 75%, which indicates that there may be competition between two opposite effects on the char-burning rate under the medium condition. In other words, the reactive effect is explained by the fact that the high volatile coal, which is sub-bituminous, releases more volatile matter, forming a higher gas temperature field, which, in turn, heats the low volatile coal and promotes its devolatilization and combustion. Simultaneously, an unreactive effect is expected where a high oxygen-deficient environment is formed in the initial stage due to rapid combustion of sub-bituminous coal, and this leads to inefficient combustion of the bituminous coal. This effect can be confirmed from the simulation of the medium condition, as shown in Figure 4b. The figure shows the higher reactivity in char burning and the faster depletion of oxygen concentration with increasing SBRs. In other words, coal particles with faster devolatilization and combustion rates consume more oxygen and leave a depleted oxygen concentration for the combustion of less reactive particles. The numerical results in Figure 5 under a medium condition also show that there is competition between the reactive and the unreactive effect. In the case of the highest condition, the unburned carbon fraction also gradually decreases up to SBR 50%, and it is shown that the best condition for the unburned carbon fraction is likely at SBR 50%, as there appears to be a slightly oxygendeficient environment at SBR 75%, analogous to the more clearly demonstrated case of the medium condition. However, the magnitude of the unburned carbon fraction is lower than that under medium conditions, with a value that is 53% of the medium condition value at SBR 75%. This result would imply that combustibility in coal blending can be improved and oxygen deficiency can be reduced by an oxygen-rich condition. In the case of the lowest condition, the results show that the magnitude of the unburned carbon fraction is much higher than that of the other conditions and the competition between the reactive and unreactive effect is not apparent, unlike other conditions. The unburned carbon fraction decreases gradually from 14.8% at SBR 0% to 7.3 at SBR 75%, which implies that

Figure 6. Comparison of NOx emission as a function of SBR at three different excess oxygen conditions in out-furnace blending.

three different excess oxygen conditions in out-furnace blending. The NOx emission at the medium condition of experimental results is proportional to SBR due to the increase of fuel N contained in a coal with increasing SBR. Table 1-1 clearly shows that the fuel N content of the sub-bituminous coal is 60% higher than that of bituminous coal. The NOx emission at the highest condition has a similar value to that of the medium condition within the error range, which implies that the medium and highest conditions are favorable environments for the formation of NOx emission, as would be expected due to fuel N conversion. However, the NOx emission for the lowest condition is about 25% lower than those of the medium and highest conditions, which indicates that there is effective reduction in NOx for the lowest condition tested. From the literature, the overall formation of NOx in pulverized coal is known to be affected significantly by the fuel NOx mechanism, which is strongly dependent on the environmental conditions (temperature and O2) around the coal.33 The results of the lowest condition are consistent with the previous findings of Harding et al.,34 who indicated that higher amounts of HCN and NH3 at the lowest condition would lead to less net formation of NOx emission through 6809


Energy & Fuels

Article

Figure 7. Predictions of char reaction rate (left, kg/s) and O2 concentration (right, mole fraction) at different feeder distances and excess oxygen conditions in the in-furnace blending (SBR 75%).

greater NOx reduction, resulting from these intermediate fixed N species. The numerical results for NOx emission in the

highest and medium conditions are in reasonable agreement with a similar tendency within the error range. However, that of 6810


Energy & Fuels

Article

numerical modeling are greater in all of the conditions than those obtained from experiments. Figure 9 shows the

the lowest condition has a further difference between experimental and numerical compared to other conditions, for it may not be considered the catalytic effect of NOx reduction caused by mineral matter in the numerical models. Zhong et al.35 reported that mineral matter in ashes is a very effective catalyst for NO reduction with chars, which can increase the NO reduction efficiency under oxygen-lean conditions in particular. 4.2. Unburned Carbon Fraction and NOx Emission of In-Furnace Blending with Different Excess Oxygen. Figure 7 shows the predictions of char reaction rates (kg/s) and O2 concentration at different feeder distances and excess oxygen conditions for the in-furnace blending method. The infurnace blending method in this study was suggested to mitigate the adverse effect of unburned carbon and to reduce NOx emissions at SBR 75% where the oxygen-deficient environment is prevalent. Dynamic motion of the binary coal in the furnace with the in-furnace blending method with different excess oxygen can be confirmed in Figure 7. The different distances between the feeders (3, 5, and 7 cm) are varied by the changeable center tube, through which the subbituminous coal is injected. Figure 8 shows the comparison of

Figure 9. Comparison of NOx emission as a function of feeder distance at three different excess oxygen conditions in the in-furnace blending.

comparison of NOx emission between experimental and numerical results at different feeder distances and excess oxygen conditions for the in-furnace blending. NOx emission under the medium conditions decreases with increasing feeder distance for the in-furnace blending method. One of the reasons for the reduction of NOx emissions is that NO formed from bituminous coal is reduced by the intermediate species (HCN and NH3) formed from the volatile matter of the subbituminous coal. This phenomenon can be explained partially with the predicted distribution of HCN concentration for 0 and 5 cm feeder distances at three different excess oxygen conditions, as shown in Figure 10. Contrary to the medium condition, it is seen that the NOx emission in Figure 9 under the highest condition is actually slightly increasing with distance between feeder locations. This result indicates that NOx formed from bituminous coal is not as readily reduced, due to the lower net concentrations of intermediate species (HCN and NH3) formed from the sub-bituminous coal under the highest conditions, which results in greater NOx emissions in the higher oxygen environment than those of the medium condition. In the case of the lowest condition, the results in Figure 10 show that a lowest condition persists for a greater distance, which leads to higher concentrations of intermediate species over longer periods of time, and thus the NOx emission continually decreases due to the NOx reduction mechanisms. Glarborg et al.36 demonstrated that reactions of NOx with intermediate fixed N species lead to reduced amounts of NOx by decomposition and that fuel N is oxidized to NO with a lower efficiency under the lowest conditions. The other reason for NOx reduction in the lowest conditions is the role played by sub-bituminous coal injected as reburning fuel during burning of bituminous coal. As demonstrated by the results of Spliethoff et al.,37 the hydrocarbons (CH, CH2, and CH4) coming from fuels can reduce NOx effectively in a lean-oxygen environment. In addition, Zhong et al.38 reported that heterogeneous reduction mechanisms also contribute to the NOx reduction more than the homogeneous mechanisms, such as nitrogencontaining species (HCN, NH3) and hydrocarbons, when the sub-bituminous coal is used as the reburning fuel. In other words, further NOx reduction can be promoted by the catalytic

Figure 8. Comparison of unburned carbon fractions as a function of feeder distance at three different excess oxygen conditions in the infurnace blending.

unburned carbon fractions between experimental and numerical results as a function of feeder distance at three different excess oxygen conditions in the in-furnace blending. The results under the medium condition show that, as the distance between the two feeders increases, the unburned carbon fraction gradually decreases, which means that each coal has more abundant oxygen to burn and the oxygen-deficient environment is improved. Furthermore, the discrepancy in the unburned carbon fractions becomes smaller with an increasing feeder distance, and this keeps the magnitude similar beyond a distance of 5 cm. In the case of the highest condition, all of the unburned carbon fractions are lower than those for the medium conditions, due to higher combustibility. The unburned carbon fraction decreases similarly to that of the medium condition, but the position that the magnitude of the unburned carbon fraction becomes similar was put forward from 5 to 3 cm than that of the medium condition. It means that the oxygendeficient environment is improved by an atmosphere of more oxygen. In the case of the lowest condition, lower combustibility leads to a higher unburned carbon fraction. The magnitudes of unburned carbon fraction obtained by 6811


Energy & Fuels

Article

Figure 10. Distribution of HCN concentration (mole fraction) for 0 cm (left) and 5 cm (right) feeder distances at three different excess oxygen conditions (SBR 75%).

involvement of O2−char, NO−char, and CO−NO reactions in the reaction system, and these reactions would further influence each other. Therefore, more detailed studies are needed for heterogeneous reactions in this system. Figure 11 shows the sensitivity analysis of numerical modeling with respect to fuel N distribution under medium conditions in out-furnace and in-furnace blending methods. The fuel N distribution (NR) is defined as the volatile N/char N ratio, and in this study, NR for Adaro (sub-bituminous coal) was varied among 0.6, 2.2, and 4.0 with respect to the fixed fuel N content and the fuel N distribution of Yakutugol (bituminous coal). The results show that NOx emission for NR 4.0 in out-furnace blending is 6.4% lower and NR 0.6 is 21% higher than that of NR 2.2, which means that, as the NR increases, the NOx reduction increases because of the greater homogeneous NO mechanism caused by volatile N. The NOx emission in the in-furnace blending method exhibits a similar trend with respect to NR, which indicates that the blending ratio and blending time in the in-furnace blending method is more significant than the fuel N distribution. A comparison of numerical and experimental results shows that NR 2.2 is able to provide better prediction of the NOx emission among others.

effect of more alkali species (Ca, Fe, Mg, K) contained in subbituminous coals (see Table 1-2). To estimate the characteristics of NOx emission more precisely, the NOx emission model, which takes the heterogeneous mechanism into account, should be incorporated in numerical modeling as well as in the homogeneous mechanism. The heterogeneous mechanisms of the catalytic effect that can be incorporated in numerical modeling are as follows:39−41 First, NO is absorbed at the active sites of the char surface; the metal ions remove the O atom from the NO molecule, resulting in the formation of metal oxide and an N radical. The metal oxides are subsequently reduced into metal ions by the char. Thus, the metal ions are recovered, and they continue taking part in the chemical reaction cycles. N radicals in the char surface will combine together to form N2 molecules, and thus, NO can be reduced continuously with the help of metal ions as catalysts. The chemical reaction cycles using one such metal ion, namely, K are listed below K+ + NO → KO− + N KO− + C → K+ + CO N + N → N2

5. CONCLUSIONS This study was conducted to investigate the behavior that occurs with coal blends and to examine the mechanisms that affect the coal blending methods at different excess oxygen conditions using a drop-tube furnace with both experimental and numerical methods. The conclusions are as follows: 1. Under the medium conditions, which represent typical levels of excess oxygen in air, it was confirmed that, in the out-furnace blending method, contrasting processes, such as reactive and unreactive effects, occur with SBRs and the worst condition for burning was found at an SBR of 75%. In addition, the in-furnace blending method was suggested in order to mitigate the adverse effect of unburned carbon and to reduce NOx emissions. The infurnace blending method resulted in an improvement in the efficiency of unburned carbon fractions and a further reduction in the NOx emission.

2NO + 2C → 2CO + N2

Second, the excess O2 can contribute to CO production through O2−char gasification in the pores of the char. The aromatic ring in the char structure is cracked during O2−char gasification, and the surface carbon−oxygen complex (C(O)), which acts as the active sites for NO reduction, is produced. Mineral matters are good catalysts for char gasification as they promote the O2−char gasification and increase the concentration of C(O) on the char surface. CO produced by the O2− char gasification plays an important role in the NO reduction for chars. O2 + 2C → 2C(O)

C(O) + NO → 1/2N2 + CO2

However, Zhao et al.41 stated that NO−char reactions in the presence of O2 are very complicated because of the 6812


Energy & Fuels

Article

predicted experimental results on unburned carbon and NO emission in the out-furnace and in-furnace blending methods with different excess oxygen and suggested that, to estimate the characteristics of NOx emission more precisely, the NOx emission model, which takes the heterogeneous mechanism (catalytic effect of NO reduction on the char N oxidation) into account, should be incorporated in numerical modeling as well as the homogeneous mechanism. However, the practical applications of the in-furnace blending could be very limited because of the complex arrangement of the pulverized coal burner and a number of complicated phenomena, including a real boiler. Accordingly, these findings will be evaluated and verified using a boiler CFD model of the two scenarios in the next paper.

■

AUTHOR INFORMATION

Corresponding Author

* Tel: 82-51-510-3051. Fax: 82-51-582-9818. E-mail: chjeon@ pusan.ac.kr. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Pusan Clean Coal Center (PC3), Pusan National University, and by a primary project of the Korea Institute of Energy Research (KIER) in Republic of Korea.

■

REFERENCES

(1) Biswas, S.; Choudhury, N.; Sarkar, P.; Mukherjee, A.; Sahu, S. G.; Boral, P.; Choudhury, A. Fuel Process. Technol. 2006, 87, 191−199. (2) Du, S. W.; Chen, W. H.; Lucas, J. A. Energy 2010, 35, 576−581. (3) Shen, Y. S.; Guo, B. Y.; Yu, A. B.; Zulli, P. Fuel 2009, 88, 255− 263. (4) Moroń, W. Combustion Characteristics of Blended Coals; Wroclaw University of Technology: Wroclaw, Poland, 2005. (5) Smart, J. P.; Nakamura, T. J. Inst. Energy 1993, 66, 99. (6) Rubiera, F.; Arenillas, A.; Arias, B.; Pis, J. J. Fuel Process. Technol. 2002, 77, 111−117. (7) Peralta, D.; Paterson, N. P.; Dugwell, D. R.; Kandiyoti, R. Energy Fuels 2002, 16, 404. (8) Biswas, S.; Choudhury, N.; Sarkar, P.; Mukherjee, A.; Sahu, S. G.; Boral, P.; Choudhury, A. Fuel Process. Technol. 2006, 87, 191−199. (9) Ulloaa, C.; Borregob, A. G.; Hellec, S.; Gordona, A. L.; García, X. Fuel 2005, 84, 247−257. (10) Tsuji, H.; Shirai, H.; Matsuda, H.; Rajoo, P. Fuel 2011, 90, 850− 853. (11) Ikeda, M. Development of Technology for Reducing NOx Emission and Unburned Carbon Concentration in Fly Ash Using In-Furnace Blending Method; CRIEPI Report; CRIEPI: Tokyo, Japan, 2008. (12) Lee, B. H.; Kim, S. G.; Song, J. H.; Chang, Y. J.; Jeon, C. H. Energy Fuels 2011, 25, 5055−5062. (13) Kuprianov, V. I. Renewable Sustainable Energy Rev. 2005, 9, 474−498. (14) Lee, B. H.; Song, J. H.; Kang, K. T.; Chang, Y. J.; Jeon, C. H. Korean Soc. Mech. Eng. 2009, 33, 876−885. (15) Lee, B. H.; Song, J. H.; Kim, R. G.; Kim, S. G.; Kim, Y. G.; Chang, Y. J.; Jeon, C. H. Energy Fuels 2010, 24, 4333−4340. (16) Shearer, S. A. Fluid Mechanics: Stokes’ Law and Viscosity. Measurement Laboratory No. 3 EGR 101, University of Kentucky: Lexington, KY, USA, Available online, 2010. (17) Barroso, J.; Ballester, J.; Ferrer, J. M.; Jimenez, S. Fuel Process. Technol. 2006, 87, 737−752. (18) Liu, H.; Shao, Y. Appl. Energy 2010, 87, 3162−3170.

Figure 11. Sensitivity analysis with respect to fuel N distribution under medium conditions in out-furnace and in-furnace blending methods.

2. Under the highest conditions, which represent enhanced levels of excess air, the best conditions for the unburned carbon fraction in the out-furnace blending method was observed at an SBR of 50%; a slightly oxygen-deficient environment is formed at SBR 75%, the same as those under the medium condition. The unburned carbon fraction in the in-furnace blending method lower than those for the medium conditions due to higher combustibility but the position that the magnitude of the unburned carbon fraction gets similar was put forward from 5 cm to 3 cm than that of medium condition. Meanwhile, NOx emission under the highest conditions increases slightly. 3. Under the lowest conditions, this represents an oxygenlean atmosphere compared with the medium conditions, and the unburned carbon fraction in the out-furnace blending method decreases gradually from SBR 0% to SBR 75% without any competition effect. The unburned carbon fraction in the in-furnace blending methods also gradually decreases, and the reduction of the NOx emission is more effective than those under other conditions because of homogeneous and heterogeneous NOx reduction mechanisms. These results imply that the in-furnace blending method under the medium conditions would be an effective method to improve combustibility and NOx emission. 4. Numerical modeling was conducted using the commercial CFD software FLUENT code and reasonably 6813


Energy & Fuels

Article

(19) Du, S. W.; Chen, W. H.; Lucas, J. A. Energy 2010, 35, 576−581. (20) FLUENT User’s Guide; Fluent Inc.: Lebanon, NH, 2001. (21) Magnussen, B. F.; Hjertager, B. H. Symp. (Int.) Combust., [Proc.] 1977, 16, 719−729. (22) Genetti, D. B.; Fletcher, T. H.; Pugmire, R. J. Energy Fuels 1999, 13, 60−68. (23) Baum, M. M. Combust. Sci. Tech. 1971, 3, 231−243. (24) Field, M. A. Combust. Flame 1969, 13, 237−252. (25) Hill, S. C.; Smoot, L. D. Prog. Energy Combust. Sci. 2000, 26, 417−458. (26) Visona, S. P.; Stanmore, B. R. Combust. Flame 1999, 118, 61− 75. (27) Molina, A.; Eddings, E. G.; Pershing, D. W.; Sarofim, A. F. Combust. Flame 2004, 136, 303−312. (28) Genetti, D. B.; Fletcher, T. H. Energy Fuels 1999, 13, 1082− 1091. (29) Baulch, D. L.; Drysdall, D. D.; Horne, D. G.; Lloyd, A. C. Evaluated Kinetic Data for High Temperature Reactions; Butterworth: Cambridge, U.K., 1973. (30) Kandamby, N.; Lazopoulos, G.; Lockwood, F. C.; Perera, A.; Vigevano, L. ASME International Joint Power Generation Conference and Exhibition, Houston, TX; ASME: New York, 1996. (31) De Soete, G. G. Symp. (Int.) Combust., [Proc.] 1975, 15, 1093− 1102. (32) Moroń, W. In Combustion Characteristics of Blended Coals, THIRD EUROPEAN COMBUSTION MEETING, Proceedings of the European Combustion Meeting: U.K., 2007. (33) Hill, S. C.; Smoot, L. D. Prog. Energy Combust. Sci. 2000, 26, 417−458. (34) Harding, N. S.; Smoot, L. D.; Hedman, P. O. AIChE J. 1982, 28, 573. (35) Zhong, B. J.; Tang, H. Combust. Flame 2007, 149, 234−243. (36) Glarborg, P.; Jensen, A. D.; Johnsson, J. E. Prog. Energy Combust. Sci. 2003, 29, 89−113. (37) Spliethoff, H.; Greul, U.; Rüdiger, H.; Hein, K. R. G. Fuel 1996, 75, 560−564. (38) Zhong, B. J.; Shi, W. W.; Fu, W. B. Fuel Process. Technol. 2002, 79, 93−106. (39) Wang, Z.; Zhou, J.; Wen, Z.; Liu, J.; Cen, K. Energy Fuels 2007, 21, 2038−2043. (40) Lee, Y. W.; Park, J. W. Environ. Sci. Technol. 2002, 36, 4928− 4935. (41) Zhao, Z.; Qiua, J.; Lib, W.; Chenb, H.; Lib, B. Fuel 2003, 82, 949−957.

6814


Effect of Coal Blending Methods with Different Excess Oxygen on

Recommend Documents