Article Application of the NOx Reaction Model for Development of Low-NOx Combustion Technology for Pulverized Coals by Using the Gas Phase Stoichiometric Ratio Index

We previously proposed the gas phase stoichiometric ratio (SRgas) as an index to evaluate NOx concentration in fuel-rich flames. The SRgas index was defined as the amount of fuel required for stoichiometric combustion/amount of gasified fuel, where the amount of gasified fuel was the amount of fuel which had been released to the gas phase by pyrolysis, oxidation and gasification reactions. In the present study we found that SRgas was a good index to consider the gas phase reaction mechanism in fuel-rich pulverized coal flames. When SRgas < 1.0, NOx concentration was strongly influenced by the SRgas value. NOx concentration was also calculated by using a reaction model. The model was verified for various coals, particle diameters, reaction times, and initial oxygen concentrations. The most important reactions were gas phase NOx reduction reactions by hydrocarbons. The hydrocarbon concentration was estimated based on SRgas. We also investigated the ratio as an index to develop a new low-NOx combustion technology for pulverized coals. We examined the relation between local SRgas distribution in the fuel-rich region in the low-NOx flame and NOx emissions at the furnace exit, by varying burner structures. The relationship between local SRgas value and local NOx concentration was also examined. When a low-NOx type burner was used, the value of SRgas in the flame was readily decreased. When the local SRgas value was the same, it was difficult to influence the local NOx concentration by changing the burner structure. For staged combustion, the most important item was to design the burner structure and arrangement so that SRgas could be lowered as much as possible just before mixing with staged air.


Introduction
The implementation of low-NO x -combustion technologies for pulverized coals began in the 1980s [1].Low NO x combustion technologies have included two-stage combustion [2], re-burning [2,3], and in-flame NO x reduction by using a low-NO x burner [1,4].A feature which is common to all these technologies is that the NO x reduction region was formed in the flame where the oxygen concentration was very low [1].To predict NO x emissions well, it is important to be able to reliably predict NO x concentrations in the NO x reduction region.However, for combustion of coal, NO x reduction performance changes easily with the burning conditions, such as coal properties [3] and coal particle diameter.However, boiler design and development cannot be done efficiently if it is necessary to change the method of NO x reduction for each coal property.In a previous study, we proposed an index to estimate NO x reduction performance [5].This index has hardly any influence from the burning conditions.In the present study, we introduce a case which utilizes the proposed index to develop a new low-NO x -combustion technology.
As computers have become even more sophisticated, their use to perform the numerical analyses needed in designing pulverized coal firing boilers is increasing [6][7][8][9][10][11].The reduction of the computational load is important for engineering design of actual boilers, and this can be done by using appropriate models based on our proposed index.
Several studies have been made to investigate the reaction of volatile matter [12,13].These studies focused on the first half of the coal combustion phenomenon.Other studies have focused on the second half of the coal combustion, the char combustion region [14,15].While studies on NO x reduction by char [16][17][18][19] have been made, the contribution for total NO x emission from the pulverized coal combustion was not clear.In previous studies, we proposed a NO x reaction model which focused on the gas phase reaction in the char combustion region [5,15].A key reaction was reduction of NO x by hydrocarbons.In the present study, the model was verified for various coals, particle diameters, reaction times, and initial oxygen concentrations.

The Relationship between Gas Phase Stoichiometric Ratio, SRgas, and NO x
Previously, we evaluated NO x generation characteristics for the fuel-rich condition in an experiment which used a drop-tube furnace [15].Under these conditions the NO x characteristics seemed to depend on the burning conditions.Then, we analyzed the experimental data in the relationship between inlet stoichiometric ratio (SRin) and NO x and this result depended on the coal properties, particle diameter, and burning temperature.SRin was defined as the stoichiometric ratio of inlet mixture of coal and air as shown in Equation (1) [5]: SRin = amount of fuel required for stoichiometric combustion/amount of fuel actually supplied (1) For pulverized coal combustion NO x is reduced by both gas phase and solid phase reactions [5,15].We tried to analyze the experimental data in reference [15] by separating them into gas phase and solid phase reactions; we focused especially on the gas phase reaction.
Figure 1 shows how to obtain the value of SRgas.Before combustion, all fuel components of the pulverized coals are in the solid phase.Pulverized coal particles are surrounded by a combustion supporting gas such as air.
After the coal particles are ignited, some of the fuel components move from the solid phase to the gas phase by pyrolysis, oxidation, and gasification reactions (Figure 1).The remaining fuel components stay in the solid phase.The gas phase stoichiometric ratio is the index which focuses on the amount of fuel components moving from the solid phase to the gas phase.We define the gas phase stoichiometric ratio (SRgas) by Equation (2): SRgas = amount of fuel required for stoichiometric combustion/amount of gasified fuel (2) Here, the amount of gasified fuel means both the amount of fuel that has moved from the solid phase to the gas phase by pyrolysis, oxidation, and gasification reactions, and, the amount of gas and liquid fuel supplied to the combustible mixture.We do not consider the fuel components which are left in the solid phase.Usually, the gas phase reaction rate exceeds the solid phase reaction rate.When the gas phase reaction is the focus, sometimes the effect of the solid phase reaction can be ignored.The stoichiometric ratio is a good index which shows the difference in the burning conditions.We thought a numerical analysis might become easy when solid is removed from the burning mixture and an equivalence ratio is defined to consider the gas phase reaction.
The SRgas index can be obtained by analyzing the mass balance of H, C and O in the burning gas.Sometimes, it is difficult to analyze the mass balance correctly, because the amount of water is difficult to measure.For such a situation, SRgas can be approximately by Equation (3): Here, X O 2 , X CO , X H 2 , and, X CH 4 are O 2 , CO, H 2 , and CH 4 mole fractions in the burning gas.X O 2 0 is the average mole fraction of O 2 in the combustion supporting gas.X TR is the concentration of tracer.N 2 , Ar and He are examples of suitable tracers.In the present study, we chose N 2 as a tracer because there is hardly any change in its amount due to chemical reactions.X TR 0 is the concentration of tracer in the combustion supporting gas.
At first, combustible materials in the coal are released to the gas phase as CO, hydrocarbons and H 2 .When the gas contains sufficient oxygen the CO, hydrocarbons, and H 2 are oxidized by O 2 to form CO 2 and H 2 O, respectively.When the gas contains little oxygen, a part of CO, hydrocarbon and H 2 remain in the gas.The quantity of the fuel that was moved from the solid phase to gas phase could be estimated by measuring the amount of O 2 , CO, hydrocarbon and H 2 in the gas.The concentration of CH 4 was mentioned specifically in Equation (3) because CH 4 was the major hydrocarbon remaining in flames.
Figure 2 plots the NO x concentration characteristics obtained for the drop-tube furnace [15].The results were obtained for a variety of coal properties (these are summarized later in Table 1), particle diameters, burning temperatures, and compositions of combustion supporting gas.Relationships between both SRin and NO x , and SRgas and NO x are shown in the figure.When the data were analyzed by SRin, NO x concentration varied with experimental conditions.However, when the data were analyzed by SRgas the difference between NO x concentrations became small and NO x was barely influenced by the burning conditions when SRgas was less than 1.0.We judged that SRgas was a good index which estimated NO x concentration in fuel-rich conditions.We further thought that the gas phase NO x reduction was the key reaction when SRgas was smaller than 1.0; that is, the mechanism of the gas reaction did not depend so much on properties of the coal and there were many common points.

Reaction Model
Previously, we proposed a NO x reaction model which can estimate NO x concentrations under fuel-rich conditions [5].The main reaction scheme is shown in Figure 3.The main feature of the model is the gas phase reaction of NO x reduction by hydrocarbons.Ox(OH) is an important oxidative species.We thought that concentrations of hydrocarbons and Ox(OH) strongly depend on SRgas.According to the scheme, NO x is mainly reduced by hydrocarbons to form XN (NH 3 , HCN).Most of the XN (NH 3 , HCN) forms NO x again.However, the remaining XN reacts with NO x to form N 2 .Reactions used for this study are listed in Table 1.(H/C) VM in Table 1 is molar ratio of H and C in volatile matter.(H/C) char in Table 1 is molar ratio of H and C in char.Example of (H/C) VM and (H/C) char values were described elsewhere [15].
We estimated total hydrocarbons THC in the following way: hydrocarbons are mainly formed from the part of the fuel which was released from the solid phase to the gas phase by the pyrolysis, oxidation, and gasification reactions.We assumed that the amount of THC formation increases with the rates of these reactions.Hydrogen is necessary for THC formation.When the hydrogen content in a solid fuel is increased, the THC formation rate increases.It is easy to form THC in the volatile combustion region, because more hydrogen is contained in the volatile matter than in the char.Finally, we concluded that SRgas was the most important factor.
The steady state assumption was used for reactions of XNradical.For R3 and R4, the ratio of the reaction rates (R3/R4) is important.A pyrolysis model was proposed by Yamamoto et al. [20].The concentration of OH was calculated by using equilibrium concentration and gas temperature.Equation ( 4) proposed by Bose and Wendt [14] was used for the calculations: Reaction parameters are listed in Tables 2 and 3. Table 2 shows parameters of solid phase reactions of coal K (hv-bituminous coal).These parameters vary with coal properties.Usually, the rate constants of R5, R10, R11, and R12 are decreased with fuel-ratio.The rate constant of R5 was estimated from char combustion experiments [15].Char was burnt in N 2 -O 2 -NO mixtures.O 2 in the surrounding gas was varied from 2 to 9 vol %.NO in the surrounding gas was varied from 0 to 512 ppm.The rate constants were estimated by difference of the NO concentration between inlet and exit value.For some coals, the values of k R5 were varied with surrounding oxygen concentration.Values of γ and η in Figure 3 also varied with coal properties.For coal K, γ was 0.2 and η was 0.07.
Table 3 shows the gas phase reaction parameters.These parameters were the same for all coals.O 2 , H 2 , CO and CO 2 concentration in the drop-tube furnace were measured under various temperatures and stoichiometric ratios.Concentrations of these species could be similar by equilibrium values.

Calculated Results
Figures 4, 5 and 6 show verifications of the calculation model using the experimental results obtained with the drop-tube furnace.Figure 4 shows the effect of coal properties (see Table 4 below) and burning temperature on NO x concentration.We checked the results for both the relationships between SRin and NO x , and SRgas and NO x .The calculated results reproduced both characteristics.When SRgas was less than 1.0, NO x was influenced by SRgas, while it was hardly influenced at all by the other experimental conditions.
We checked the model for around 200 other experimental conditions.We changed oxygen and NO x concentrations in the combustion supporting gas, particle diameters, and burning times (L = 400-800 mm).The results are shown in Figure 5.The difference between experimental and calculated results was approximately 20%.When particle size was very large, calculation precision decreased.The calculation did not predict the value of SRgas well, because the calculation error of the coal burnout became large.At first, it is important to predict the value of SRgas well, for predicting NO x concentration well.Staged combustion properties were also verified for coal K. Typical results are shown in Figure 6.Structures of the drop-tube furnace for staged combustion are shown later (Figure 16).Burning temperatures of these experiments were higher than those shown in Figure 4 and 5. NO x concentration and carbon in ash were measured at the furnace exit.The proposed model was also verified for staged combustion conditions.Carbon in ash (w t%, dry)

Application of SRgas to Low-NO x -Combustion Technology Development in the Air Combustion Experiment
Structures of the laboratory scale furnace for the air combustion experiment are shown in Figure 7 [1].The furnace was the down flow type combustor, in which pulverized coal particles and air were supplied from the upper part of the apparatus.The combustion air was divided into the primary (1ry), secondary (2ry) and ternary (3ry) air.Coal feed rate was approximately 25 kg/h.Pulverized coals were supplied through the primary air nozzle.Burner Nozzles A and B structures are shown in Figure 8.The structure of the third Nozzle C was the same as that of Nozzle B, but the angles of their swirler vanes were different.The nozzles used the swirler vane to form a swirled flow, and plate-like members to obtain a straight flow.The effects of the swirler vane and the plate-like members were described previously [21].In the present study, we tested NO x emission by varying the diameter of the primary nozzle, the size of the swirler vane and the plate-like members, and the angle of the swirler vane.
Figure 9 shows the relationship between burner stoichiometric ratio and NO x emission obtained for the three nozzles.Coal E (see Table 1) was used for these experiments.NO x emission using Nozzle B was the lowest, followed by Nozzle A and Nozzle C.  We measured local SRgas distributions in the flames and the results are shown in Figure 10.Local SRgas distributions were measured at two different axial positions.One was obtained in the burner neighborhood (x = 513 mm) and the other was obtained just before the staged air nozzle (x = 1488 mm).SRgas had the minimum value at the central axis, and it increased near the side wall of the furnace.Then we investigated the effect of nozzle structure on SRgas.When we used the low-NO x type Nozzle B, the value of SRgas was easily decreased.The same characteristics were observed at both X = 513 mm and X = 1488 mm.SRgas was also a good index for investigating the reason why NO x emissions were reduced.When the NO x concentration was compared for the same axial position, the NO x concentration was strongly influenced by SRgas, but hardly influenced by the nozzle structure.The key to low NO x combustion technology development is reducing the value of SRgas in the burner zone as much as possible.The relationship between SRgas and NO x concentration obtained in the drop-tube furnace is shown in Figure 11.We compared NO x concentration obtained just before the staged air nozzle (x = 1488 mm).When SRgas was less than 0.9, NO x concentrations obtained by the laboratory scale furnace and drop-tube furnace were almost the same.NO x concentration was made stable by SRgas.NO x emission was strongly influenced by how much the SRgas value could be reduced just before staged air injection.The NO x concentration in the vicinity of the burner (x = 513 mm) was higher than that obtained downstream (x = 1488 mm).The SRgas was a good index to consider NO x performance, but, in the burner neighborhood, it did not predict the actual value.We thought that one of the reasons was the concentration fluctuation.Flow turbulence could be strong in the burner neighborhood.This might cause concentration fluctuation.In the burner neighborhood, both reductive gas (SRgas < 1.0) and oxidative gas (SRgas > 1.0) conditions were observed at the same position.NO x in oxidative gas was high, so that NO x in the burner neighborhood became high, even if SRgas was low.G as phase stoichiom etric ratio: S R gas (-) NO x (ppm , dry) We examined why NO x emissions varied with nozzle structures.Figure 12 shows differences in flow and coal concentration distributions near the burner exit between Nozzles A, B and C. Swirl flow was formed inside the nozzle.The local coal concentration of nozzle circumference rose by the centrifugal force of the swirl flow.The plate-like members installed at the nozzle exit changed swirl flow into straight flows.The resulting phenomenon was that local coal concentrations rose as the nozzle circumference side was maintained.The coals ignited from the circumference side of the nozzle.When the coal concentration rose, ignition was accelerated, then, the flame temperature rose rapidly.Oxidation and gasification reactions were accelerated because of high flame temperature, then, coal burnout in the burner zone easily increased.SRgas value was easy to decrease, so NO x concentration decreased.
The swirl strength of Nozzle B was larger than that of Nozzle A. Consequently the local coal concentration in the nozzle circumference of Nozzle B was larger than that of Nozzle A, and coal ignition in Nozzle B became easy.
The swirl strength of Nozzle C was very large, however, swirl ingredients were left at the nozzle exit.The coal particles which gathered in the nozzle circumference spread more on the circumference side.For Nozzle C, the local coal concentration at the nozzle exit could not kept high.Coal ignition was not accelerated.

Application of SRgas to Low-NO x -Combustion Technology Development in the Oxyfuel Combustion Calculation
We applied the model to the oxyfuel combustion calculation for a boiler.Two kinds of NO x are formed from pulverized coal combustion, thermal NO x and fuel NO x .The SRgas mainly affects prediction of the fuel NO x as most of the NO x generated in oxyfuel combustion is fuel NO x , because the combustion supporting gas only contains a little nitrogen.
We calculated NO x emissions for a pulverized coal firing boiler.We used a boiler of average size and structure based on publicly available information [22][23][24][25][26] about recent pulverized coal firing power plants.The electric output was around 1000 MW. Figure 13 shows calculation conditions and example results for the calculated concentration distributions of O 2 and NO x , and gas temperature distribution.The calculation method was shown previously [7,8].NO x near the burner neighborhood was high, where oxygen concentration and SRgas were high.The oxygen concentration and SRgas decreased at the center of the furnace where NO x was reduced and the NO x concentration increased a little by mixing with staged air.
The variations in SRgas in the furnace are shown in Figure 14.The SRgas distributions are expressed as an average for a horizontal section of the furnace.The SRgas depended on the burner arrangement.We compared the SRgas values just before mixing with staged air.The SRgas of Case 2 was lower than that of Case 1. NO x emissions, O 2 concentration, and unburned carbon in fly-ash at the exit of the furnace are shown in the figure.Oxygen concentration and unburned carbon in fly-ash were equal.However, the NO x emission of Case 2 was lower than that of Case 1.The SRgas became lower just before mixing with staged air, and NO x emission became lower.The results were the same as obtained from the experiments.In development of the low NO x combustion technology, the SRgas was a good index for investigating the mechanism of NO x reduction.The burner arrangements for CASE 1 and CASE 2 are described in Figure 15.Areas shown by dotted lines represent the main combustion region.Much of the supplied coal burned in this area.Much of the supplied heat was released in the area too.Total heat input of CASE 1 and CASE 2 were the same, however, the main combustion region of CASE 1 was larger than that of CASE 2. Flame temperature in the main combustion region of CASE 2 was easy to increase, because heat was released in a small area.Oxidation and gasification reactions were accelerated for CASE 2, so that the corresponding SRgas value was easy to decrease.

Experimental Section
The drop-tube furnace we used was described elsewhere [15].The reaction zone (maximum length, 1200 mm) was made of an alumina tube which had an inner diameter of 50 mm.The heating rate was 15,000 K/s.The flow rate of combustion supporting gas (O 2 /N 2 ) was 0.96 m 3 N/h and the supply rated of pulverized coals was 0.02-0.5 kg/h.A nozzle supplied the pulverized coals and combustion supporting gas to the furnace.The burning temperature (measured as the wall temperature) was 1173-1773 K.We changed the length (L) from the nozzle to a sampling port between 200, 400, and 800 mm.The reaction time was changed between approximately 0.25-1.0s.Gas composition and ash content in unburned char were measured.
Staged combustion properties were also examined.The structure of the drop-tube furnace for staged combustion is shown in Figure 16.Two high-temperature electric furnaces were connected in series.The basic structure of the each drop-tube furnace was the same as described previously [15].The maximum temperature was risen to 2100 K by using a carbon electric heater.Gasification reaction rates have been studied using a pressurized high-temperature drop-tube furnace [27].We modified that the furnace to allow use at atmospheric pressure, and we used this design for the basic construction of our two electric furnaces.Coal and air (air for the burner zone) were supplied from the upper part of the electric furnace in the upper section.Coal was burnt under fuel-rich conditions in the upper electric furnace.Staged air was supplied at the connection of the two electric furnaces.The total amount of air (air for burner zone + staged air) was 0.96 m 3 N/h.Analyses of eleven coal samples are listed in Table 4. Five types were studied: sub-bituminous, hvbituminous, mv-bituminous, and anthracite, and petroleum coke.

Conclusions
We applied a gas phase stoichiometric ratio index (SRgas) to predict NO x concentrations, to develop the NO x reaction model and a new low-NO x -combustion technology.
(1) Gas phase stoichiometric ratio (SRgas) was defined as: SRgas = amount of fuel required for stoichiometric combustion/amount of gasified fuel where the amount of gasified fuel was defined as the amount of fuel which had been released to the gas phase by pyrolysis, oxidation and gasification reactions.SRgas was determined to be a good index to consider the gas phase reaction mechanism in fuel-rich pulverized coal flames.
(2) We applied the SRgas index to develop a NO x reaction model.The key feature was that the concentration of hydrocarbons was estimated as a function of SRgas.Calculated NO x

Figure 2 .
Figure 2. Relationship between gas phase stoichiometric ratio and NO x concentration.

Figure 3 .
Figure 3. Reaction model of NO x formation and reduction.

Figure 4 .Figure 5 .
Figure 4. Effect of coal properties and burning temperature on NO x concentration, symbols are experimental results and lines are calculated results.

Figure 6 .
Figure 6.Staged combustion properties of coal K. Exit stoichiometric ratio was 1.2.Burning temperature was 1873 K for the burner zone and 1673 K for the staged combustion zone.Residence time was around 1.4 s for the burner zone and around 1.0 s for the staged combustion zone.Symbols are experimental and lines are calculated results.

Figure 9 .
Figure 9.Comparison of NO x emissions for three burner structures at different stoichiometric ratios.Exit stoichiometric ratio was 1.15 for all experiments (exit O 2 was 2.7 vol%, dry).

Figure 10 .
Figure 10.Distribution of gas phase stoichiometric ratio for the laboratory scale furnace.

Figure 11 . 2 G
Figure 11.Relationship between gas phase stoichiometric ratio and NO x concentration for the laboratory scale furnace.

Figure 12 .
Figure 12.Images of flow and coal concentration distributions near the burner exit for Nozzles A, B and C.

Figure 15 .
Figure 15.Arrangement of Burners for CASE 1 and CASE 2.

Figure 16 .
Figure 16.Structure of the drop-tube furnace for staged combustion.

Table 1 .
Reactions used in this study.

Table 2 .
Parameters of solid phase reactions.

Table 3 .
Parameters of gas phase reactions.

Table 4 .
Analysis of studied coals.