Numerical Research of Flue Gas Denitrification Using the SNCR Method in an OP 650 Boiler

Abstract

The presence of Poland in the European Union obliges the domestic economy and the professional energy sector to improve the condition of the natural environment by reducing the emissions of harmful substances into the environment. One of the substances that have a negative impact on the environment is nitrogen oxides. The results of numerical calculations of flue gas denitrification using the SNCR method in an OP 650 boiler are presented in the paper. The method of verifying the combustion of the numerical model, in terms of measurement and calculations with a zero-dimensional model, is presented. Then, the results of numerical tests of flue gas denitrification using the SNCR method with the use of urea solution injection in a specific temperature window for various nozzle positions are presented. In this paper, three variants of the reagent’s injection into the furnace chamber were carried out, depending on the height of the position of the nozzle. It is shown in this paper that thanks to combined NO_x reduction systems, it is possible to adjust the emission of nitrogen oxides to a level below 200 mg/m³_n with an oxygen content of 6% in the dry flue gas, with relatively low investment costs.

1. Introduction

The presence of Poland in the European Union obliges the domestic economy and the professional energy sector to improve the condition of the natural environment by reducing the emissions of harmful substances into the environment. There are increasing restrictions on the presence of compounds considered harmful in exhaust gases, forcing manufacturing and industrial plants to invest in equipment and installations to reduce emissions.

One of the substances that have a negative impact on the environment is nitrogen oxides. The NOx emission standards and legal requirements for the emission of nitrogen oxides resulting from the combustion of fuels in energy are presented in the Industrial Emissions Directive (IED) [1]. The benchmark for the values of the emission standards is the Best Available Technique (BAT) reference document [2]. For the combustion of coal fuel in units with a rated thermal input greater than 300 MW, this limit is 200 mg/m³_n, with a 6% oxygen content in the flue gas.

Nitrogen oxides are formed during the combustion of fuel in boilers. NO_X is one of the major pollutants of the environment. In order to reduce the emissions of nitrogen oxides, primary and secondary methods are used. Secondary methods for reducing nitrogen oxides include selective non-catalytic reduction (SNCR) and selective catalytic reduction (SCR) [3]. Primary methods for reducing nitrogen oxides include air staging and fuel staging. Compared to secondary methods, they are cheaper to build and operate.

Air staging is used in low-emissions combustion. As a rule, it is related to the division of the furnace chamber according to its height into zones that differ in air concentration—the sub-stoichiometric combustion zone (first zone) and the above-stoichiometric combustion zone (second zone). In the second zone, fuel particles are burnt with air from Overfire Air (OFA) nozzles [4]. Research on air staging in fossil fuel- and biomass-fired boilers has been conducted for many years. The reduction of thermal nitrogen oxides is related to the temperature drop in the combustion zone for willow and miscanthus [5]. Numerous model studies have also been conducted related to the method of coal combustion in boilers and the reduction in pollutant emissions from these boilers [6]. The effectiveness of NO_X emission reduction using air staging is dependent in particular on the air-fuel mixing intensity and the excess air ratio as well as the residence time of the coal particles in the combustion area.

More effective, however, is a combination of air and fuel staging [7]. This reduces NO_X emissions by reducing the nitrogen oxides formed in the first phase of combustion to N₂. However, a requirement is a reducing atmosphere and the presence of radicals such as CH_X and NH_X. The reducing radicals needed in the second combustion zone (reburning) come from either the coke particles left behind after the combustion of the coal volatile fractions or from additional fuel fed upstream of the first combustion zone. It is important to allow time for it to burn out completely before the gases reach the combustion chamber outlet. Yang et al. [8] conducted research on the effects of air and fuel staging and fuel characteristics on the reduction efficiency of nitrogen oxides. The research results show that that as the residence time of the fuel particles in the furnace chamber increases, the nitrogen oxide reduction index increases. For coal with a high volatile content and with residence times ranging from 0.8 s to 2 s, the reduction rate of nitrogen oxides rises from 20% to almost 75%. Numerical multi-injection and multi-stage tests for a utility boiler with a capacity of 350 MWe were carried out in [9]. The results reveal that the use of the new solution eliminates asymmetric combustion and significantly reduces NO_x (close to 50%) for all considered settings. The effect of throttle opening for the coal/air lean (FLD) flow on the coal dust combustion process in a 350 MW supercritical boiler was researched by Li et al. in [10]. However, the results show that the FLD regulation had only a minor impact on the reduction in NO_x emissions.

One of the secondary methods of reducing the emissions of nitrogen oxides to the atmosphere mentioned above is the selective non catalytic reduction method (SNCR). This is a flue gas denitrification method based on the reduction of nitrogen oxides formed by the injection of ammonia or urea. Ammonia or urea must be entered in the correct temperature window. The test results show that the temperature window is strongly dependent on the type of agent. For ammonia, it is in the temperature range of 800–1000 °C (optimal temperature: 870 °C) [11], and for urea it is 850–1100 °C (optimal temperature: 1000 °C) [12]. The most important factor for the process is the optimal selection of the place of injection of the reacting substance into the boiler. Above the temperature range, the problem of NO production in the oxidation of ammonia arises. Below the temperature window, the problem is the low degree of conversion of the reactant, which results in the so-called ammonia slip. The injected solution chemically reacts with the nitrogen oxide NO and reduces it to N₂ and H₂O. The more frequently used reagent is the aqueous urea solution due to the safer and less demanding conditions of transport and storage. The influence of the height and intensity of ammonia injection on the efficiency of the SNCR method was numerically investigated in [13]. By using the SNCR method, NO_x emissions can be reduced by up to 63%. The efficiency of the SNCR process is influenced by many items, where the reduction level of nitrogen oxides ranges from 25% to over 75%, as stated by Świeboda et al. in [14]. Meanwhile, Hu et al. [15] showed that ammonia should be fed in the appropriate temperature window, at several levels and as far as possible from the front. In addition, the SNCR injection had an optimum velocity of 13.80 m/s, and the efficiency of nitrogen oxide removal was 51.50% and the slip of NH₃ was 11.08 mg/m³_n. The enhancement of the nitrogen oxide removal efficiency was influenced by the increased importance of the first row of SNCR injection. The most desirable injection ratio of 6:3:1 (row importance ratio SNCR1:SNCR2:SNCR3) was obtained, resulting in an efficiency of nitrogen oxide removal of 53.10% and a slip of NH₃ of 8.43 mg/m³_n. A modern type of furnace with spatial ammonia feeding, developed on the basis of the gradual use of high-temperature waste heat and the SNCR method, is presented in [16]. In the research, the most favorable temperature window was found in the range of 911~984 °C, while the residence time was estimated at 0.21~0.38 s. A well-carried out process of mixing the reducer with flue gas, and thus the full contact of NH₃ with NO, contributes to the high effectiveness of nitrogen oxide reduction. The efficiency of nitrogen oxide removal was appreciably upgraded due to the increased depth of the reducing element infiltration into the central area of the power boiler, as was found in [17]. By using the mixing air along with enhancing the injection speed and the size of the droplet, respectively, the infiltration distance was enhanced and thus the mixing efficiency increased. Kang et al. [18] used CHMKIN software to optimize the mechanism of the SNCR reaction in a cyclone separator of a supercritical CFB boiler. As a result of the research, they came to the conclusion that the appropriate setting for the normalized stoichiometric ratio NSR should be 1.5.

The use of flue gas denitrification using the SNCR method together with other primary methods such as low-emission burners or air staging with OFA nozzles is sufficient to obtain the expected NO_x emission level. The effectiveness of these methods of reducing nitrogen oxides allows for reductions of about 40–70% [19].

Currently, therefore, a combination of primary and secondary methods is used to meet nitrogen oxide emission standards. This article presents the results of numerical tests of the OP 650 boiler using a combination of air and fuel staging with selective non-catalytic reduction. There are no examples of numerical models of pulverized coal boilers in the literature that include a combination of primary and secondary methods of reducing nitrogen oxides. In addition, for this type of boiler, no numerical calculations were proposed to search for the location of SNCR nozzles and the temperature window in order to effectively reduce nitrogen oxides. Therefore, this paper presents the procedure of searching for the temperature window and searching for the location of the SNCR nozzle set. The method of selecting the reagent stream required to obtain the required reduction of nitrogen oxides was also presented, based on the NSR index. A method was proposed based on a validated numerical model with a zero-dimensional model.

2. Boundary Conditions and Modelling Description

In order to carry out numerical studies of the flue gas denitrification using the SNCR method, it is important to first develop the batch data for modeling and to prepare the geometry and numerical grid. The next step is to validate the model based on the measurements and the zero-dimensional model. Next, a temperature window should be established for the proper placement of SNCR nozzles. The main objective of the research is to reduce nitrogen oxide emissions below the value of 200 mg/m³_n with a 6% oxygen content in the flue gas, as required by European legislation. To do this, numerical calculations will be carried out for three variants of the proposal to place SNCR nozzles in two rows. After analyzing the results, the most optimal option will be indicated in terms of the objective function, i.e., the maximum reduction of NOx nitrogen oxides. A block diagram of the basic approach to carrying out validated numerical calculations of the combustion process is shown in Figure 1, while a block diagram of the determination of the required reduction in nitrogen oxide emissions by the SNCR method is presented in Figure 2.

The OP 650 boiler is a natural circulation boiler fired with pulverized coal. The combustion chamber has the shape of a cuboid with a cross-section of 19.2 × 9 m. The walls of the chamber are covered with screen tubes. In the upper part of the furnace chamber, a part of the rear screen is bent towards the inside of the chamber, thanks to which, by creating a narrowing, the desired direction of the flue gas flow is given. The third stage of the primary steam superheater is located at the outlet of the furnace in the form of overhanging platens. In the lower part of the combustion chamber, there is an ash hopper with narrowings used to direct the slag falling from the combustion chamber to the slag remover tank. The combustion system used in the OP 650 boiler is based on the differentiation of the air-pulverized coal mixture concentration at different levels of the burners. This system was developed by the IPW Polin. The air-pulverized coal mixture was divided into concentrated (main burners) and thinned (additional burners) mixtures by the use of a mechanical shutter dust distributor [20]. It is located in the air-pulverized coal mixture path downstream of the mill separator. Low NO_x swirl burners are located in the two lower rows on the front wall (main burners). In addition to these burners, there are also additional burners located between the first and second stage of the OFA nozzles. A diagram of the combustion system is shown in Figure 3. Swirl vanes are only available for the air-pulverized coal mixture—see Figure 4. Burners in the first row have a larger outlet area for secondary air III than burners in the second row. In the burners, the fuel and primary air were swirled through the vanes at an angle of 40°—Figure 3. OFA nozzles are placed in two rows on the front wall and in one row on the back wall. In OFA nozzles, the air is fed without swirl.

Table 1. Distribution of air and fuel to the boiler.

Data	Unit	Burners Row 1	Burners Row 2	Burners Row 3
Core air	kg/s	1.25	1.25	-
Primary air	kg/s	18.30	18.30	37.7
Coal	kg/s	9.45	9.45	8.12
Temperature of primary air	°C	108	108	108
Temperature of secondary air	°C	250	250	250
Secondary air III	kg/s	29.1	23.3	-
Secondary air II	kg/s	12.5	14.3	-
OFA air 1	kg/s	12.7
OFA air 2	kg/s	42.3
OFA air 3	kg/s	25.2

shows individual values of coal and air mass flow for each burner row and the temperature factors.

The sieve residues for a given coal particle diameter for the main burners are shown in

Table 2. Coal granulation—main burner—1st and 2nd row.

Data	Unit	Value
Residue on 88 µm sieve	%	28.2
Residue on 132 µm sieve	%	10.3
Residue on 200 µm sieve	%	5.1
The average diameter	µm	70.1
Polydispersity number	-	1.04

, and for the additional burners in

Table 3. Coal granulation—additional burner (between level of OFA I and II).

Data	Unit	Value
Residue on 88 µm sieve	%	15.2
Residue on 149 µm sieve	%	3.9
Residue on 200 µm sieve	%	1.1
The average diameter	µm	48.5
Polydispersity number	-	1.06

. The mean particle diameter and polydispersity number are also given there. The parameters of the coal burnt in the boiler are presented in

Table 4. Coal parameters (as received).

Data	Unit	Value
Calorific value	kJ/kg	21,761
Ash content	%	27.1
Moisture content	%	7.4
Carbon	%	56.4
Hydrogen	%	3.57
Sulphur	%	0.63
Oxygen	%	3.92
Nitrogen	%	0.98

.

Not all quantities required as boundary conditions in the numerical model can be determined by measuring a real object. An example may be the emissivity of the combustion chamber walls, the amount of heat received by individual heat exchange surfaces from the flue gas, and the temperature of the flue gas behind individual heat exchange surfaces. Therefore, balance calculations of the entire boiler should be performed with a fully balanced zero-dimensional model based on the measurement and operational data of the OP 650 boiler. The description of the zero-dimensional model is given in [21].

The numerical model along the flue gas path took into account the third stage of the superheater (platen), the fifth stage of the superheater, and total heat exchangers located in the second pass of the boiler. In the second pass of the boiler, the collective area consists of the fourth stage of the live steam superheater, the first and second stages of the steam superheater, and the second stage of the water heater. The boundary conditions on the walls of the model are temperature, emissivity, and wall thickness. On the basis of the zero-dimensional model, the temperature and the emissivity of the walls were calculated. In the model, the walls were regarded as flat areas exchanging heat with the flue gas, whereas the superheaters were regarded as a porous element taking the shape of a plate exchanging heat with flue gas. The porous body model implemented in the calculations is described in [22].

The values of the heat flux received by the second stage of the superheater, the fifth stage of the superheater, and the total heat exchangers located in the second pass of the boiler, calculated in the zero-dimensional model, are presented below in

Table 5. Amounts of heat flux received.

Place of Heat Reception	W/m3
Platen superheater	133,355
5th stage of live steam superheater	90,310
2nd pass	62,294

. They were implemented in the numerical model on the given surfaces of the model as negative heat sources.

The numerical mesh of the OP 650 boiler consists of 512,000 elements—see Figure 5. The mesh density decreases along with the boiler in relation to the direction of the flue gas flow. It is denser in the areas of burners and OFA nozzles where fuel and air are injected. Most of the mesh is structural, consisting of cubes. The numerical mesh must meet the quality requirements so that calculations can then be performed:

• Minimum Orthogonal Quality = 1.37872 × 10⁻² (values from 0 to 1)
• Maximum Ortho Skew = 9.76954 × 10⁻¹ (values from 0 to 1).

The above criteria are within the acceptable ranges.

Figure 5. View of the boiler model geometry with description and the numerical mesh.

Figure 5. View of the boiler model geometry with description and the numerical mesh.

The Navier–Stokes equations are used to solve turbulent flow. The Reynolds-averaged Navier–Stokes approach (RANS) was utilized in the model. In the RANS approach, equations are formulated for the time averages of temperature, pressure and velocity. This approach reveals a good compromise between the quality of the studied phenomena and the hardware requirements. For the pressure–velocity coupling, the SIMPLE scheme was chosen. In the numerical calculations, the k-epsilon realizable turbulence model was applied. Two additional equations closing the system are introduced by this model. This requires the introduction of additional boundary conditions. As new variables, the kinetic energy of turbulence (k) and energy dissipation (ε) appear in this model. This permits the designation of the turbulent viscosity (µ_t). It is responsible for raising the viscosity in turbulent flows. Two extra equations of the transport are calculated concurrently with the energy, mass and momentum balance equations. The realizable k-epsilon model in comparison to other models from the k-e group gives the best results [23]. The transport equations for k and ε are presented below:

$$\frac{\partial }{\partial t}\left(\rho k\right)+\frac{\partial }{\partial x_j}\left(\rho k{u}_j\right)=\frac{\partial }{\partial x_j}\left[\left(u+\frac{u_t}{{\sigma }_k}\right)\frac{\partial }{{\partial }_k}\right]+G_k+G_b-\rho \epsilon -Y_M+S_k$$

(1)

$$\frac{\partial }{\partial t}\left(\rho k\right)+\frac{\partial }{\partial x_j}\left(\rho \epsilon u_j\right)=\frac{\partial }{\partial x_j}\left[\left(u+\frac{u_t}{{\sigma }_{\epsilon }}\right)\frac{\partial \epsilon }{\partial x_j}\right]+\rho C_{1}S\epsilon -\rho C_2\frac{{\epsilon }^2}{k+\sqrt{v\epsilon }}+C_{1\epsilon }\frac{\epsilon }{k}{C}_{3\epsilon }{G}_b+S_{\epsilon },$$

(2)

where G_k expresses the generation of turbulence kinetic energy in relation to the velocity gradient, G_b represents the generation of turbulence kinetic energy provoked by buoyancy, Y_M expresses the impact of the dilatation fluctuations in compressible turbulence on the total dissipation coefficient, C₂ and C_1ε are constants, and σ_k and σ_ε are the corresponding Prandtl numbers for k and ε, respectively.

Therefore, it is necessary to solve both the equations expressing the fluid flow (Navier–Stokes equations) and the application of extra N equations resulting from the products of reactions occurring in the combustion process. These equations can be expressed as classical transport equations:

$$\frac{\partial \rho Y_k}{\partial t}+\frac{\partial \rho u_i{Y}_k}{\partial x_i}=\frac{\partial }{\partial x_i}\left(\rho D_k\frac{\partial Y_k}{\partial x_i}\right)+\dot{{\omega }_k} dla k=1,2\dots N,$$

(3)

where D_k is the diffusion coefficient, Y_k expresses the mass fraction of particular reaction ingredients, and ω_k represents the rate of the chemical reaction.

The last important equation for the mathematical description of the combustion process is the energy equation. It can be represented by the transport equation for temperature:

$$\frac{\partial \rho T}{\partial t}+\frac{\partial \rho u_{i}T}{\partial x_i}=\frac{\partial }{\partial x_i}\left(\frac{\lambda }{C_p}·\frac{\partial T}{\partial x_i}\right)+\dot{{\omega }_T},$$

(4)

where

$$\dot{{\omega }_T}={\sum }_{k=1}^N\Delta h_k\dot{{\omega }_k},$$

(5)

where Δh_k represents the enthalpy of k ingredient production and Y_k represents the thermal diffusion coefficient while C_p is the heat capacity.

In calculations of the combustion process, the flue gas density is the sum of the density of particular ingredients, while the dynamic viscosity is the temperature function computed from empirical formulas [24]. However, more than enough reactions take place during the combustion process, leading to the use of a simplified finite rate/eddy dissipation (FR/ED) combustion model [24]. In this model, the reaction rate is computed from the ED equations and furthermore from the Arrhenius equation for the kinetic constants of total reactions, which can be defined singly. This model is based on physics (vortex disappearance time), which makes it more reliable. The combustion of volatile matter, as well as the oxidation of carbon monoxide reactions, are presented below. Based on the coal composition, the coefficients m, n, l, k, and j were received.

$$C_m{H}_n{O}_l{N}_k{S}_j+\left(\frac{m-l}{2}+\frac{n}{4}+j\right){\mathrm{O}}_2\stackrel{k_1}{\to }m\mathrm{CO}+\frac{n}{2}{\mathrm{H}}_2\mathrm{O}+\frac{k}{2}{\mathrm{N}}_2+j{\mathrm{SO}}_2$$

(6)

$$\mathrm{CO}+0.5{\mathrm{O}}_2\stackrel{k_2}{\to }{\mathrm{CO}}_2$$

(7)

The coal particle degassing was calculated based on the single rate model. This model is based on the assumption that the rate of devolatilization depends primarily on the amount of volatile matter staying in the coal [25]. The combustion of the char was carried out according to the kinetic-diffusion model based on research by Baum et al. [26] and Field [27]. The rate of the surface reaction is assumed to be dependent on the kinetics or the diffusion rate. The reaction of char to CO₂ by the kinetic-diffusion model is presented below.

$$\mathrm{C}+{\mathrm{O}}_2\to {\mathrm{CO}}_2$$

(8)

The flow of coal particles was simulated using a discrete phase model (DPM). Single coal particle trajectories are estimated according to Lagrange’s theory. Changes in particle properties are tracked in the model. It is therefore possible to determine the temperature and velocity at any point in the movement of the particle. In the particle trajectory equation, the trajectories were calculated using the mean fluid velocity $\overline{u}$:

$$\frac{d{u}_p}{dt}=F_d\left(u-u_p\right)+\frac{g_x\left({\rho }_p-\rho \right)}{{\rho }_p}+F_x,$$

(9)

The friction force is expressed by the first term of the equation. The second term is responsible for the force of gravity, and the third term is for the additional forces.

$$F_d=\frac{18\mu }{{\rho }_p{d}_p^2} \frac{C_{D}Re}{24},$$

(10)

$$Re=\frac{\rho d_p\left(u_p-u\right)}{\mu },$$

(11)

where u—flow velocity, u_p—particle velocity, µ—fluid dynamic viscosity, ρ—fluid density, ρ_p—particle density, d_p—particle diameter, Re—relative Reynolds number, C_D—drag coefficient depending on the particle shape.

Particle sphericity was adopted for the coal particles. The drag coefficient C_D is given by the equation:

$$C_D=a_1+\frac{a_2}{Re}+\frac{a_3}{R{e}^2},$$

(12)

where a₁, a₂, and a₃ are constants defined by Morsi and Alexander [28].

Between coal particles and flue gas, two-way interaction was observed. Thus, the coal particles act on the flue gas through momentum, mass, and energy. On the other hand, the flue gas affects the coal particles through friction and turbulence. For coal particles, the Rosin–Rammler–Sperling distribution was utilized in the simulations.

Radiation heat transfer was taken into account by applying the discrete ordinates (DO) radiation model. The heat transport equation by radiation for a finite number of discrete fixed angles was solved. Equation (1) is transformed by the DO model into the radiation intensity transport equation in spatial coordinates (x, y, z). The solution is carried out using the same method as for the fluid flow and energy flux equations. The isotropic phase dispersion function has been applied.

The exchange of radiation between the flue gas and the coal particles as well as between the boiler walls and the pulverized coal particles was taken into account in the computations. The absorption coefficient was computed by the weighted sum of the gray gas model (WSGGM).

Combustion of fuel produces three basic ingredients of oxygen with nitrogen—nitrous oxide N₂O, nitrogen oxide NO, and nitrogen dioxide NO₂. The temperature and the oxygen concentration affect the rate of nitrogen oxide creation. From nitrogen included in the air, the thermal and prompt oxides are created. On the other hand, fuel oxides are formed from nitrogen compounds included in the fuel. The relationships between temperature and concentrations of the ingredients as well as the NO_x formation rate are highly nonlinear. Thus, in order to include these fluctuations in temperature and composition, the probability density function (PDF) was utilized in simulations. The numerical model includes fuel and thermal nitrogen oxides. To compute thermal nitrogen oxide formation, the Zeldowicz approach [29] was applied:

$$\mathrm{O}+{\mathrm{N}}_2\rightleftharpoons \mathrm{NO}+\mathrm{N}$$

(13)

$$\mathrm{N}+{\mathrm{O}}_2\rightleftharpoons \mathrm{NO}+\mathrm{O}$$

(14)

The third reaction providing the thermal nitrogen oxides formation was proposed in environmental conditions comparable to stoichiometric conditions, and in the case of fuel-rich mixtures [30]:

$$\mathrm{N}+\mathrm{OH}\rightleftharpoons \mathrm{NO}+\mathrm{H}$$

(15)

Later, the combustion process is finished, and most of the thermal nitrogen oxides are generated. Assuming thermal equilibrium and the equilibrium of stable ingredients, O atoms, and OH free radicals [27], it is possible to isolate the process of thermal NO_x production from the main combustion process. The partial equilibrium method was used to establish the concentration of O [31] and OH radicals [32]. It was assumed in the simulations that nitrogen contained in the fuel is allocated between volatile matter and char components. It was assumed that the nitrogen included in volatile matter forms HCN and NH₃, while nitrogen included in char takes part directly in NO formation [33]. As shown by the research of Winter et al. [34], the adoption of a division factor for HCN/NH₃ equal to 9:1 for hard coal allows one to obtain better NO_x predictions in comparison with the measurements than the indication of the only one indirect product.

Table 6. Assumptions of the numerical model.

Data	Model
Two-phase model	Euler–Lagrange
Turbulence model	Realizable k-ε
Combustion model	Finite-Rate/Eddy-Dissipation
Radiation model	Discrete Ordinates (DO)
For the fuel particle
Devolatilization	Single-Rate model
Combustion of char	The kinetic-diffusion model
NOx model
Type	Thermal and fuel
Concentration of OH, O	Partial equilibrium
Turbulence interaction for PDF	Temperature path
Intermediates of volatile components	HCN/NH3
Intermediates of char	NO

presents the conclusion of the numerical model. The SNCR model uses the seven-step reduced kinetic mechanism proposed by Brouwer et al. to predict NO concentration. [35]:

$${\mathrm{NH}}_3+\mathrm{NO}\to {\mathrm{N}}_2+{\mathrm{H}}_2\mathrm{O}+\mathrm{H}$$

(16)

$${\mathrm{NH}}_3+{\mathrm{O}}_2\to \mathrm{NO}+{\mathrm{H}}_2\mathrm{O}+\mathrm{H}$$

(17)

$$\mathrm{HNCO}+\mathrm{M}\to \mathrm{H}+\mathrm{NCO}+\mathrm{M}$$

(18)

$$\mathrm{NCO}+\mathrm{NO}\to {\mathrm{N}}_2\mathrm{O}+\mathrm{CO}$$

(19)

$$\mathrm{NCO}+\mathrm{OH}\to \mathrm{NO}+\mathrm{CO}+\mathrm{H}$$

(20)

$${\mathrm{N}}_2\mathrm{O}+\mathrm{OH}\to {\mathrm{N}}_2+{\mathrm{O}}_2+\mathrm{H}$$

(21)

$${\mathrm{N}}_2\mathrm{O}+\mathrm{M}\to {\mathrm{N}}_2+\mathrm{O}+\mathrm{M}$$

(22)

and the two-stage urea decomposition mechanism proposed by Rota et al. [36]:

$$\mathrm{CO}{\left({\mathrm{NH}}_2\right)}_2\to {\mathrm{NH}}_3+\mathrm{HNCO}$$

(23)

$$\mathrm{CO}{\left({\mathrm{NH}}_2\right)}_2+{\mathrm{H}}_2\mathrm{O}\to {\mathrm{NH}}_3+{\mathrm{CO}}_2$$

(24)

3. Results and Discussion

3.1. Verification of the Numerical Model

In the beginning, the flow part was verified as part of the so-called cold flow. The air velocities obtained in these tests at the outlet of individual burners and OFA nozzles (Figure 6) were compared with the velocities obtained in the zero-dimensional model. In this model, they are calculated from the stream continuity equation based on the knowledge of temperature, and thus the density and mass flow of air, as well as the geometry of the burners and OFA nozzles. In addition, the flow part was also verified as part of the appropriate calculations taking into account the combustion process. The average velocities as area-weighted averages behind the individual heat transfer surfaces obtained from these tests were compared with the average velocities from the zero-dimensional model—see

Table 7. Average velocities (m/s) as area-weighted average in plane behind the individual heat exchange surfaces.

Data	CFD	0D Model	Error%
Platen superheater	5.52	5.44	1.4
5th stage superheater	6.63	6.45	2.7
Outlet	4.8	4.70	2.0

. As shown by the results from Table 7, fairly good compliance with the verified velocity was obtained. The velocity error between numerical modeling and zero-dimensional model, depending on the tested area, ranges from 1.4 to 2.7%.

The parameters that were achieved by performing the calculations in the Ansys Fluent program include: the flue gas temperature at the outlet of the chamber T_OC, the flue gas temperature behind the platen superheater T_OPS, the flue gas temperature after the fifth stage of the live steam superheater T_O5THS, the flue gas temperature at the outlet of the model T_OM, the oxygen content at the outlet of the chamber O₂, the amount of carbon monoxide at the outlet of the model CO, and the amount of nitrogen oxides at the outlet of the model NO_x. The carbon monoxide and nitrogen oxides values were taken from the measurement—signature m. These values were measured on a real object using a Siemens Ultramat 23 analyzer. The nitrogen oxides were converted to 6% of O₂ in dry flue gas and shown in mg/m³_n. Presented values were obtained as area-weighted average in plane. The verification of the numerical model with the zero-dimensional model was presented in

Table 8. Summary of the parameters of the verification results.

Parameter	0D Model/Measurement	CFD Model	Difference
TOC	1371 K	1367 K	4 K
TOPS	1253 K	1235 K	18 K
TO5THS	1163 K	1171 K	8 K
TOM	773 K	792 K	19 K
O2	2.57%	2.025%	0.545 p.p.
CO—m	250 ppm	290 ppm	40 ppm
NOx—m	330 mg/m3n	310 mg/m3n	20 mg/m3n

.

As one can read from Table 8, the numerical model was verified with the zero-dimensional model and measurement to a satisfactory degree. Temperature differences do not exceed 3%. Below, the simulation results for temperature, oxygen, carbon monoxide, and the trajectory of coal dust particles are presented graphically.

Figure 7 shows the temperature distribution in the planes of the burner axis. It can be seen that the maximum temperature of the flue gas in the combustion chamber is roughly at the level of the low-emission burners fed with the concentrated mixture. Similarly, in [37], the temperature reaches the highest values in the area of the burners. The temperature then drops along the path of the flue gas. There is a clear loss of temperature behind the platen superheater and the fifth stage of the superheater. This is caused by the heat reception on their surfaces. The same effect is noticeable in [38]. The average temperature on the surface of the model outlet is approx. 790 K. It can also be noticed that the air injection through the OFA nozzles lowers the temperature in the combustion chamber. In [39], a decrease in the flue gas temperature above the over-fire air nozzles is also observed. You can see that the mixture is ignited at a short distance from the front wall of the boiler.

The oxygen concentration in the planes of the burner axis is shown in Figure 8. The point of ignition of the air-fuel mixture can be clearly observed. There is also a visible difference in the air flows supplied by the OFA nozzle. OFA nozzles on the second level deliver twice as much air to the combustion chamber than on the other levels. The amount of oxygen at the outlet of the model is approx. 4%, and in the outlet surface of the chamber it is approx. 2%. A high excess air ratio promotes the formation of nitrogen oxides, which are then eliminated by air staging by OFA nozzles. Air staging effectively reduces the amount of NOx in the flue gas. A noticeable difference in the oxygen concentration between the first and second levels of the burners was observed. This is due to the greater degree of opening of the nozzles (see Figure 4) and the greater amount of supplied air III from the first row of burners (see Table 1). Among the three levels of OFA nozzles, the first row supplied the least amount of air.

The distribution of carbon monoxide in the boiler is presented in Figure 9. At the boiler outlet, the amount of carbon monoxide formed is already small and amounts to approx. 290 ppm. Its highest concentration occurred in the area of the burners and in the area of the highest temperatures. A similar distribution of CO can be observed in [39]. Air staging with OFA nozzles had a positive effect on reducing the concentration of carbon monoxide in the chamber. Proper air staging also prevented the formation of a large amount of CO and an increase in the amount of unburned carbon in the fly ash.

The concentrations of nitrogen oxides in various sections of the boiler are shown in Figure 10. This model is the base model for the subsequent SNCR model calculations.

The concentration of nitrogen oxides at the outlet of the model is 310 mg/m³_n. It can be seen that the greatest amount of NO_x occurs in the burner regions and then decreases with successive OFA nozzles. In [39], a similar effect of OFA nozzles is also observed. The air staging with OFA nozzles had a beneficial effect in reducing the emissions of nitrogen oxides at the outlet, this effect is also visible in [37]. Due to the high temperature in the combustion chamber, most of the NO_x are generated at the rear wall of the boiler; these are thermal nitrogen oxides formed at high temperatures. The figures show a reduction atmosphere that helps to reduce the number of nitrogen oxides through the formation of reducing radicals due to the additional fuel.

3.2. Reduction of Nitrogen Oxides by the SNCR Method

For a coal-fired boiler containing a combination of primary and secondary methods of reducing nitrogen oxides, numerical calculations were proposed to search for the location of SNCR nozzles and the temperature window in order to effectively reduce nitrogen oxides. The method of selecting the reagent flux necessary to obtain the required reduction of nitrogen oxides based on the NSR index was also presented.

As part of the application of the SNCR method, a 32.5% urea solution was fed into the combustion chamber in an appropriate temperature window in the range of 850–1100 °C, where it reacts with NO to form N₂. The injection of urea at a higher temperature than the appropriate window can lead to an increased formation of nitrogen oxides, while at lower temperatures ammonia is formed. The effectiveness of the NO_x reduction by the SNCR method is determined by the reaction temperature, the ratio of the reducing substance to nitrogen oxides, the contact time of the reagent in the appropriate temperature window, and the chemical composition of the flue gas.

In order to find the temperature window in which urea injection should take place, planes from the level of 23.5 m to 34.5 m were created. The distance between the planes was 0.5 m. The temperature distribution on the created planes is presented in Figure 11.

Based on the analysis of the results of the flue gas temperature in the planes presented in Figure 11, the location of the urea injection was determined. For simulations using the SNCR method, three injection cases were adopted depending on the height of the nozzles. They were arranged at equal intervals in a row every 1.75 m with a distance of 2 m between the rows. In order to properly select the amount of urea injected into the SNCR, the normalized stoichiometric ratio (NSR) is used. The NSR determines the amount of reagent needed to achieve the target NOx reduction. In theory, two moles of NOx are removed with one mole of urea. In practice, more reagent is introduced to achieve NOx reduction. The reason for this is the complexity of the chemical reactions involving nitrogen oxides and urea injection. It is also influenced by the limitation of mixing between the reagent and the flue gas [40]. The NOx reduction as a function of NSR is shown in Figure 12. It should be noted that as NSR increases, NOX reduction increases.

There were 10 nozzles per level and a total mass flow of 32% urea solution of 0.1446 kg/s was divided evenly among all of the nozzles. Three cases with two rows of nozzles were calculated. The injections took place in the temperature window from approx. 930 °C to approx. 1050 °C. The considered cases of the location of the nozzles are presented in

Table 9. The cases of urea solution injections.

Case	1st Row of Nozzles	2nd Row of Nozzles	Mass Flow from 1 Nozzle (kg/s)
1	27.5 m—approx. 1050 °C	29.5 m—approx. 990 °C	0.00723
2	28.5 m—approx. 1020 °C	30.5 m—approx. 960 °C	0.00723
3	29.5 m—approx. 990 °C	31.5 m—approx. 930 °C	0.00723

.

The simulation results for various cases of urea solution injection are shown in Figure 13. The highest efficiency was obtained for the third case of calculations, where the injection took place in two rows of nozzles at the levels of 29.5 m and 31.5 m in the temperature window of approx. 930 °C and 990 °C. In relation to the original value of 310 mg/m³_n for the base case, NO_x reduced by 112 mg/m³_n, or 36.1%, in case 3. For this case, the NH₃ slip was obtained at the level of 9.8 mg/m³_n. An intermediate efficiency was obtained for case 2. The urea injection took place in the temperature window of 960 °C and 1020 °C at the levels of 28.5 m and 30.5 m in two rows of nozzles. The obtained value was 205 mg/m³_n, i.e., NO_x reduction by 105 mg/m³_n, i.e., by 33.9% in relation to the base case. For case 2, the NH₃ slip was obtained at the level of 9 mg/m³_n. The lowest efficiency was obtained in case 1. The injection took place in the temperature window of 990 °C and 1050 °C at the levels of 27.5 m and 29.5 m in two rows of nozzles. The lower efficiency could have resulted from the injection of the urea at a higher temperature than in the other cases. The received value was 222 mg/m³_n, which means a reduction of NO_x by 88 mg/m³_n, i.e., by 28.4% in relation to the base case. The NH₃ slip was obtained at the level of 7.6 mg/m³_n for case 1. The NO_x reduction efficiency was calculated as:

$${\eta }_{N{O}_x}=\frac{N{O}_{x\_i}-N{O}_{x\_out}}{N{O}_{x\_i}},$$

(25)

where NO_{x_i}—initial value NO_x (mg/m³_n), and NO_{x_out}—value after injection NO_x (mg/m³_n).

The results of the calculations for the reduction produced by SNCR for individual cases are presented in

Table 10. NOx reduction results for 3 cases.

Case	Initial Value NOx (mg/m3n)	Value after Injection NOx (mg/m3n)	Difference (mg/m3n)	The NOx Reduction Efficiency	Slip NH3 (mg/m3n)
1	310	222	88	28.4	7.6
2	310	205	105	33.9	9
3	310	198	112	36.1	9.8

.

In case 3, the reduction of nitrogen oxides was favorably achieved in relation to case 2 by 7 mg/m³_n, i.e., by about 3.5%. On the other hand, case 3 was more favorable than case 1 by 24 mg/m³_n, i.e., by about 11%. Comparing case 2 with case 1, the difference in favor of case 2 is 17 mg/m³_n, i.e., about 8%. It is therefore important to carry out a study aimed at finding the right temperature window in order to obtain the correct amount of urea injection.

Figure 13 clearly shows the reduction of NO_x in the area of the injected urea. As can be seen, a factor of great importance in the reduction caused by the SNCR method is locating the injections in the appropriate temperature window. If the nozzles were located in the higher temperature zone, the NO_x reduction was less effective.

As in [13], with the temperature increase in cases 3 to 1, the NO_x emission value increases, i.e., the NO_x reduction efficiency decreases. The increase in temperature results, of course, from a decrease in the level of urea injection, i.e., from approaching the region of higher temperatures above the furnace chamber of the boiler. In [15], the influence of the position of SNCR nozzles on NOx emission was also investigated. Similar to the studies presented in [36], NOx reduction decreases with increasing temperature in the interval considered in the study, which can be seen in Figure 14.

As can be seen from Figure 14, for the considered average temperatures in the range of 960 °C to 1020 °C, the NH₃ slip decreases with a decrease in NOx emissions reduction. The same trend can be observed in [19]. Figure 15 shows that with an increase in the average urea feed level, the flue gas temperature decreases, and thus the emissions of nitrogen oxides NOx decrease. The same trend was presented in [13].

4. Conclusions

This paper presented numerical research on flue gas denitrification using the SNCR method. The numerical model was verified with a zero-dimensional model as well as measurement. Numerical calculations were performed for three cases of injection of 32.5% urea solution into the combustion chamber of the OP 650 boiler depending on the height of the level of the nozzles level. Urea was administered in the temperature range of 930–1050 °C. Based on the results obtained, the following conclusions can be drawn:

• The velocity error between numerical modeling and the zero-dimensional model, depending on the tested area, ranges from 1.4 to 2.7%.
• The temperature differences between the numerical model, the zero-dimensional model and the measurement do not exceed 3%.
• The injection of urea at an average level of 30.5 m in the temperature window of approx. 960 °C results in a reduction of NOx by 112 mg/m³_n or 36.1% compared to the base case. The NH₃ slip was obtained at the level of 9.8 mg/m³_n.
• The injection of urea at an average level of 29.5 m in the temperature window of approx. 990 °C results in a reduction of NOx by 105 mg/m³_n or 33.9% compared to the base case. The NH₃ slip was obtained at the level of 9 mg/m³_n.
• The injection of urea at an average level of 28.5 m in the temperature window of approx. 1020 °C results in a reduction of NOx by 88 mg/m³_n or 28.4% compared to the base case. The NH₃ slip was obtained at the level of 7.6 mg/m³_n.
• The best efficiency in reducing NOx emissions was obtained for the highest position of the nozzles, which was associated with a lower reagent feed temperature of 930–990 °C. The emission level was 198 mg/m³_n, which is 36.1% lower than the initial value from the base case.
• Therefore, it was possible to achieve the assumed goal, i.e., the level of nitrogen oxide emissions required by European legislation.

The method of non-catalytic reduction of nitrogen oxides is much cheaper in terms of investment and operation than catalytic methods, but it is less effective. As shown in this paper, the SNCR method can be combined with other methods of reducing nitrogen oxide emissions, such as low-emission burners, air and fuel staging, and OFA nozzles systems. Thanks to the combined NO_x reduction systems, it is possible to adjust the emission of nitrogen oxides to a level lower than 200 mg/m³_n with an oxygen content of 6% in the dry flue gas with relatively low investment costs. For boilers in which combustion systems have already been modernized to the primary methods and have already obtained lower NO_x emissions, it is enough to use the SNCR method to meet the emission standards.