Numerical investigation of low NOx combustion strategies in tangentially-fired coal boilers

Fuel 142 (2015) 215–221

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Numerical investigation of low NOx combustion strategies in tangentially-fired coal boilers Xiaohui Zhang a,⇑, Jue Zhou b, Shaozeng Sun c, Rui Sun c, Ming Qin c a

The Science and Technology Research Institute, China Power Investment Corporation, Beijing 100033, PR China China Huadian Science and Technology Institute, Beijing 100077, PR China c The Institute of Combustion Engineering, Harbin Institute of Technology, Harbin 150001, PR China b

h i g h l i g h t s  The prediction shows a good agreement with the measurement on-site.  HBC can reduce the NOx in the primary combustion zone when there is no air staging.  The predicted results have shown that OFA has a remarkable reduction of NOx emission.  Air staging dominates the integral contribution when combined with HBC burner.  The lower stoichiometry and unburned char in PCZ contribute to the final NOx reduction.

a r t i c l e

i n f o

Article history: Received 22 January 2013 Received in revised form 2 November 2014 Accepted 7 November 2014 Available online 18 November 2014 Keywords: NOx emission Air staging Horizontal bias combustion (HBC) Combustion simulation Tangentially-fired pulverised-coal boiler

a b s t r a c t A numerical model is developed to investigate the effects of horizontal bias combustion (HBC) and air staging combustion (over-fire air, OFA) technologies on the performance of a 200 MWe tangentially-fired pulverized-coal boiler. The devolatilization rate and volatiles amount are determined by a kinetic devolatilization model, which predicts the coal devolatilization prior to the volatile matter and char combustion. The characteristics of the devolatilization, combustion, heat transfer and NOx emission are studied and compared to achieve a comprehensive understanding of the low NOx combustion. The prediction shows a good agreement with the on-site measurement results, which confirms that the model is capable of predicting the characteristics of the investigated boiler. The predicted results have shown that the OFA has a remarkable effect on the reduction of NOx emission. The HBC makes a significant NOx reduction in the primary combustion zone (PCZ) when there is no air staging. In terms of the NOx reduction, the air staging plays a dominant role in comparison with HBC burners. The application of OFA tends to lead to slagging in the PCZ, which can be avoided using HBC due to the higher stoichiometry close to the furnace wall. The details of this study improve the understanding of combustion and NOx emissions in tangentially-fired pulverized-coal boilers with low NOx combustion technologies, especially for boilers adopting HBC burners and OFA. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Pulverized coal will still be a main fossil fuel used by power stations to generate electricity and heat in the coming years, though it is difficult to control pollutants emitted from pulverized-coal boilers. Nitrogen oxide (NOx) is one of the main pollutants formed during the coal combustion process. Selective non-catalytic reduction (SNCR) and selective catalytic reduction (SCR) have become the principal methods employed in controlling NOx emissions. Meanwhile, low NOx combustion technologies play an important role ⇑ Corresponding author. http://dx.doi.org/10.1016/j.fuel.2014.11.026 0016-2361/Ó 2014 Elsevier Ltd. All rights reserved.

in reducing NOx emission on account of their lower capital and maintenance costs. Normally, low NOx combustion technologies can be classified into three categories: low NOx burner, air staging, and fuel staging (reburning). The former two technologies are usually used together to reduce the generation of NOx during the coal combustion process. Computational Fluid Dynamics (CFD) is a useful tool to optimize the low NOx technologies. In addition, proven numerical models are regarded as an important way to improve the understanding of coal combustion and boiler operation, which can otherwise only be achieved through expensive experiment.

216

X. Zhang et al. / Fuel 142 (2015) 215–221

In recent decades, numerous simulation studies related to the flow field, coal combustion and pollutant emissions of boilers have been conducted by many researchers. Gómez et al. [1] and Huang et al. [2] investigated the fluid flow and heat transfer in different boilers. Their predictions in both flow field and heat transfer agreed well with experimental results. In terms of mathematical simulation for pulverized-coal-fired boilers, pyrolysis and combustion are the two main investigated processes. Chaos et al. [3] and Papadikis et al. [4] adopted the rapid pyrolysis process to validate the CFD devolatilization model. Backreedy et al. [5] improved a package for co-firing pulverized coal and biomass. A modeling study carried out by Modlinski [6] was applied to investigate a tangentially-fired furnace with swirl burners. Eddings et al. [7] and Chen et al. [8] predicted the mineral matter coalescence and char burnout in boilers with different capacities, respectively. The reaction rates of NO and N2 were determined considering HCN and ammonia as intermediates in the fuel NO mechanism [9]. Homogenous reactions related with volatile nitrogen were studied [10]. It is suggested that the volatile nitrogen yield and the NH3/HCN ratio strongly affected the NOx formation. A NOx index to predict NOx levels was proposed taking into account the effects of function groups related with coal nitrogen and nitrogen-containing species. The heterogeneous reactions between NO and char were studied over the temperature range of 1250– 1750 K by Levy et al. [11], who also investigated the effects of H2O and CO on the NO emission rate. As an important part of the post-processing of CFD modeling, the NOx models studied by De Soete [9], Levy et al. [11] and Smoot and Smith [12] were further developed by Leeds University, UK, and were mentioned in the FLUENT 6.3 User’s guide. The NOx models include three principal NOx mechanisms: thermal, prompt, and fuel NOx. The NOx post-processing was used not only for the general combustion prediction but also for the advanced reburning [13] and oxy-fuel combustion [14]. Based on the validated sub-models, combustion process in a series of boilers has been simulated to predict and optimize the boiler operation and NOx emission characteristics [15–21]. Choi and Kim [20] and Park et al. [21] investigated the combustion and the NOx emission performance of a tangential-fired (T-fired) boiler. More studies have been focused on the effect of the wallfired burner [22–24] rather than the T-fired burner on the reduction of NOx emissions. A numerical investigation on HBC burners, however, is still lacking. In this paper, T-fired HBC burners, a type of fuel-staging technology, were studied using mathematical modeling. The T-fired HBC burner was originally developed for the anthracite-fired boiler in the late 1980s in China [25], and has been successfully applied to a wide range of boilers with varying boiler capacities and coal characteristics. The aims of this study were to achieve a deeper understanding of combustion and NOx emissions in T-fired pulverized-coal boilers with low NOx combustion technologies, and furthermore, to evaluate the performance of different low NOx control technologies on the reduction of NOx emissions.

2. Boiler specification The considered supercritical T-fired boiler with steam conditions of 16.8 MPa/540 °C/540 °C possesses a 670 t/h unit, single reheat steam power cycle and balanced draft. In order to reduce the NOx emission at the exit, the HBC combustor system in the primary combustion zone (PCZ) consisted of 20 HBC burners in five rows. Only 12 burners in the bottom three rows were taken into service and the cooling air was injected to protect idle burners. The application of the HBC divided the horizontal primary air/coal stream into two substreams with a large difference in the fuel concentration. The two substreams were injected into the furnace

from the same elevation with an angle (typically between 0° and 15°) between their axes. The fuel-rich substream formed a higher temperature flame core in the central zone of the furnace, which improved the ignition stability, while the fuel-lean substream generated an outer layer of a more oxidizing atmosphere, blanketing the high temperature flame core, which consequently reduced the risk of slagging in the area close to the furnace wall. The air staging technology was also employed to control the NOx concentration at a lower level. Partial combustion air was injected through eight OFA nozzles in two rows above the PCZ. Fig. 1 shows the 3-D structure of the simulated boiler. The proximate and ultimate analyses of the used coal are shown in Table 1. According to the Chinese Standard, GB/T 212–2008, the coal sample was heated to 900 °C ± 10 °C for 7 min to test the volatile yield. 3. Numerical analysis procedures 3.1. Domain and mesh system The calculation domain is considered to be from the hopper to the reheater, including all the primary air and secondary air nozzles. With structured mesh approaches the calculation domain is made of around 5 million hexahedron cells. Three million cells are generated in the hopper and primary combustion zones. The OFA zone contains around 1 million cells. The upper furnace zone, which includes a platen superheater, superheater, and two reheaters, has around 1 million cells. 3.2. Numerical models 3.2.1. Turbulent, radiation and DPM models The numerical investigation is supported by a commercial computational fluid dynamics code (Ansys FLUENT 6.3.26). The mathematical modeling has been carried out using different Ansys FLUENT versions (e.g. v14), and the modeling results are identical. Therefore, it is reasonable to use FLUENT 6.3.26 in this study. The species transport model is selected to solve the continuous phase equations. The standard k–e two-equation model is employed for the turbulent modeling and the Discrete Ordinates (DO) model is selected for the radiation modeling. The DO model first developed by Truelove [16] can solve problems ranging from the surface-tosurface radiation to the participating radiation in combustion problems. It is more suitable for calculating the radiation of the boiler furnace wall and the coal particle surface. Hence, the combination of k–e two-equation model and DO model is capable of simulating the turbulent flow and heat transfer in the industrial flow. Additionally, the Lagrangian discrete phase model is used to consider the pulverized-coal injection and the mass, momentum, and heat exchange between the discrete and continuous phases. The incompressible ideal gas, mass-weighted mixing law, and weighted-sum-of-gray-gases models are chosen to define the density, viscosity, and absorption coefficient of the gas phase mixture. The particle size distribution is simulated using the Rosin–Rammler equation. Table 2 lists the key parameters of the Rosin–Rammler function. The mean diameter and spread parameter are decided by testing the coal samples, which gave 82% passing and 98% passing on 75 lm and 150 lm sieves, respectively. 3.2.2. Devolatilization kinetic model The single-rate and kinetics/diffusion-limited models are selected to predict the devolatilization and combustion processes. In terms of the coal devolatilization and combustion, the proper volatile and fixed char fractions at the high heating rate (approximate 104 K/s) play an important role in the prediction of the temperature and flue gas components. The FG-DVC model is employed

217

X. Zhang et al. / Fuel 142 (2015) 215–221

Fuel Enricher

Inlet

OFA2 OFA1

th

5 PA row th 4 PA row 4.62m

Radial Planes

rd

3 PA row

3.6m

nd

2 PA row

2.58m

st

1 PArow 0m

Fig. 1. 3-D structure and combustion system of the investigated boiler.

Table 1 Ultimate and proximate analyses of coal. Ultimate (daf, wt%)

Proximate (As received, wt%)

C

H

O

N

S

Volatile

Fixed carbon

Moisture

Ash

82.5

4.6

10.3

1.4

1.2

24.82

37.12

8.6

29.46

Table 2 Coal particle parameters for the Rosin–rammler function. Variables Temperature Particle diameter (lm) Spread Number of (K) parameter diameters Minimum Maximum Mean Value

353

1

200

47.7

1.1898

10

to predict a reliable devolatilization yield and reaction rate. A heating rate of 104 K/s up to temperature of 1600 °C with a hold time of 1 s at the peak temperature was selected. The yields of char, volatile are predicted and listed in Table 3. The Coats–Redfern method [26] was selected to calculate the pre-exponential factor (A) and activation energy (Ea), which are 3.49e+04 s1 and 4.9e+07 kJ/kmol, respectively. 3.2.3. Combustion reaction model The Eddy–Dissipation model is used to simulate the volatile and intermediate combustion. It is assumed that the combustion reaction rate is controlled by the turbulent mixing rate only. The fuel and combustion air are separately injected into the furnace where they burn quickly. The volatile composition, C3.28H5.6O0.78, is assumed, based on the mass and energy conservations. It is assumed that the volatile and char combustion processes consist of two-step reactions, as below.

3.2.4. NOx model As one of the post-processing models, the NOx model includes the NOx generation (thermal NOx, prompt NOx and fuel NOx) and reduction approaches. The prompt NOx is minimal and can be ignored in this study. The thermal NOx is formed through three principal reactions (O + N2, N + O2, and N + OH). The partial equilibrium approach is chosen to count the effect of the radical concentration on the industrial turbulent flames. The relative O-atom [27] and OH [28] concentrations are predicted. The main purpose of air staging combustion is to reduce the formation of fuel NOx. A proper partition ratio of HCN and NH3 regarded as intermediates of fuel nitrogen can have a significant effect on the NOx prediction. Visona and Stanmore [29,30] studied the fuel-N fate in the pyrolysis and combustion processes in the flat-flame and laminar flow furnaces. It is found that 70% volatile-N is released as hydrogen cyanide (HCN) prior to mixing with the bulk gas [29]. The investigation also indicates all char-N is transformed into HCN and forms NO [30]. In the current study, the same assumption is used to calculate the intermediates of fuel nitrogen. 582 equally spaced radial planes, as shown in Fig. 1, within the zone of the second primary air row (the elevation is 3.595 m), were extracted to investigate the effect of the NOx control strategy on the stoichiometry in the PCZ. In order to evaluate the effect of the low NOx combustion technologies on the performance within the whole furnace, the modeling results at 211 horizontal planes along the furnace height were collected and analyzed.

4. Results and discussion

C3:28 H5:6 O0:78 þ O2 ! CO þ H2 O

ð1Þ

4.1. Validation of the combustion simulation results

C þ O2 ! CO ðfor charÞ

ð2Þ

CO þ O2 ! CO2

ð3Þ

A measurement was carried out when both the HBC and OFA were in service. A baseline case (‘HBC + OFA’) identical to the practical operation was built and compared with the experimental results. The air and coal feeding rates and their temperatures in the baseline case were measured and are listed in Table 4. Table 5 summarizes the measurement (the first column) and modeling results (the second column) based on the selected models. According to the comparison between the measurement and the simulation, a lower carbon in ash is predicted because of the ignoring of the effect of the thermal annealing on the char combustion in the modeling. It should be noted that the exposure of the

Table 3 Composition of devolatilization products (daf). Product

Char Volatile

Yield mass fraction (%)

50.2 49.8

Mass fraction (%) C

H

O

N

S

97.43 70.81

0.13 9.0

0 17.74

1.67 1.07

0.77 1.39

X. Zhang et al. / Fuel 142 (2015) 215–221

Table 4 The operation conditions of the considered cases. HBC & OFA Fuel rate (kg/s, dry)

OFA only

HBC only

None

54.1 110.2 55.0

54.1 165.2 0

54.1 165.2 0

28.33

Combustion air flow rate (kg/s) Primary air 54.1 Secondary air 110.2 OFA 55.0 Temperature of combustion air (K) Primary air 353 Secondary air 581 OFA 581

fuel to high temperatures in fact causes a thermal annealing of the particles, characterized by the loss of surface area, loss of active sites and/or loss of active site reactivity [31]. Additionally, the development of an ash layer can affect the oxygen penetration and consequently reduce the rate of oxidation of the partially burned char [32]. However, a constant reaction rate is adopted in this modeling work, which overestimates the char combustion rate at high conversions and finally predicts a better char burnout performance. Two measuring points were arranged on the left and right sides of the superheater entrance. The measured temperatures at these two points were 767 °C and 847 °C, respectively. The predicted average temperature at the monitored plane is 793 °C, which matches reasonably with the measured data. The difference between the measurement and the prediction of the CO emission is due to the simplified volatile combustion progress. The comparison between the experiment and the prediction shows an error of 8% for NOx, 4% for O2, 10ppmv for CO, which confirms that the selected models are capable of predicting the performance of the T-fired boiler adopting HBC burners and OFA. Apart from the baseline case with a combination of HBC and OFA (referred as ‘HBC + OFA’), three more cases with different NOx control strategies, ‘OFA Only’, ‘HBC Only’, and ‘None’ (conventional boiler furnace without the low NOx burner and air staging), were then calculated using the same validated models. All calculation results are summarized in Table 5.

flow. The difference is not obvious in the area close to the furnace wall. The stoichiometries are 0.63 and 0.7 in the cases of ‘HBC + OFA’ and ‘OFA Only ’, respectively. Compared with the case of ‘OFA Only ’, the stoichiometry in the area close to the furnace wall is higher in the case of ‘HBC + OFA’. The radii with the lowest stoichiometry are also different in both cases. The rich fuel flows created by HBC generate an inner layer of the tangential circle and form a tangential circle smaller than that in the case of OFA Only. By comparing the cases with and without HBC, the radius difference of the tangential circle is almost 0.8 m. This means that HBC results in a different manner of fuel injection and consequently a different flow field in the PCZ. On the positive side, the HBC can reduce the risk of slagging, although a lower stoichiometry in the PCZ has been generated by the OFA. On the negative side, the smaller tangential circle produced by the HBC would reduce the residence time of the coal particles and affect the gas flow and mixing performance, which may eventually cause a higher carbon mass fraction in ash. 4.2.2. Temperature distribution The temperature distributions are different between the cases with and without OFA, as shown in Fig. 3. The application of OFA generates a lower stoichiometry in the PCZ, which leads to a lower PCZ temperature than the case without OFA. In addition, the flame center is moved up, which consequently increases the flue gas temperature at the OFA zone entrance. Table 6 summarizes the heat flux of the furnace water tube wall and Fig. 4 shows the contours of the heat flux on the left furnace wallof selected cases. The comparison of the heat flux (‘HBC + OFA’ and ‘HBC Only’) explains the change of the flame center. More heat released from the coal

1.2

Stoichiometry

218

HBC+OFA

OFA Only

HBC Only

None

1

0.8

0.6

0.8m

4.2. Results analysis

0.4

4.2.1. Stoichiometry in the PCZ Fig. 2 presents the stoichiometry distribution along the radial direction. When considering the cases with OFA, the ‘HBC + OFA’ lowers the stoichiometry in the central furnace due to the rich fuel

0

1

2

3

4

5

6

Radius (m) Fig. 2. Stoichiometry distributions along radial planes in the second burner row (elevation = 3.6 m).

Table 5 Description of modeling results. NOx control strategy

Test

Prediction

HBC OFA

HBC OFA

OFA only

HBC only

None

Arch temperature (°C) Monitored temperature (°C) Exit temperature (°C) Exit O2 (vol% dry)

– 767/847 – 3.33

1339 793 678 3.21

1342 786 674 3.18

1299 788 678 3.07

1300 778 670 3.05

Heat account (MW) Heat to water walls Heat to platens Heat to SH Heat to RH

– – – –

236 112 21 25

238 115 19 24

244 105 21 26

248 108 19 24

Emissions NOx (mg/N m3 @ 6% O2) CO (ppmv, dry) Carbon in ash (%)

287 20 3.16

311.0 10 2.17

317 7 1.97

384 5 0.72

374 4 0.56

219

X. Zhang et al. / Fuel 142 (2015) 215–221

6

O2 (%vol, dry)

Temperature (ºC)

1600

1500

1400

4

2

1300 HBC+OFA

OFA Only

HBC Only

None

0 5

10

15

OFA Only

HBC Only

None

0

1200 0

HBC+OFA

5

10

20

15

20

Furnace Height (m)

Furnace Height (m) Fig. 5. Average O2 concentration along the furnace height. Fig. 3. Average temperature along the furnace height.

4.2.3. O2 distribution Fig. 5 shows the comparison of O2 concentration along the furnace height. For the case without OFA, a higher O2 concentration is observed in the PCZ due to the supply of all combustion air. The first row of the secondary air increases the O2 concentration, which is then decreased by feeding three rows of the primary air. In the domain above the third primary air row, the O2 concentration increases gradually and achieves the highest level at the top secondary air row caused by the multiple secondary air rows. A lower O2 concentration is observed at the furnace exit compared to the case without OFA. 4.2.4. Devolatilization and combustible species distribution Fig. 6 compares the summed devolatilization rate (Fig. 6a) and volatile volume fraction (Fig. 6b) at horizontal planes along the furnace height. By comparing the devolatilization performance in the

Table 6 Furnace wall heat flux PCZ, OFA zone and top furnace. Heat flux (MW)

HBC + OFA

OFA only

HBC only

None

Heat to PCZ Heat to OFA Heat to top furnace

100.2 80.4 55.8

100.0 80.9 56.7

120.7 77.6 45.7

120.5 78.7 49.1

PCZ, the first primary air row shows a similar devolatilization rate among the investigated cases. However, the devolatilization performance is different for the cases with and without OFA, which can be observed at the second and third primary air rows. The employment of OFA seems to contribute to a better devolatilization performance than in cases without OFA, as more coal particles are considered when applying the OFA. For cases with OFA, less combustion air is fed into the PCZ, which, subsequently, reduces the amount and velocity of the flue gas, and increases the residence time of coal particles. This suggests that more particles exist in the PCZ, compared to cases without OFA. The summed devolatilization rate shown in Fig. 6 adds the devolatilization rate of all coal particles together on a certain plane and shows a higher value for the cases with OFA. Fig. 6(b) presents the volatile fraction at the horizontal planes along the furnace height. The combination of lower flue gas volume and higher devolatilization rate gives a higher volatile concentration in the PCZ. For the cases with OFA, the volatile combustion lasts longer due to the lower stoichiometry in the primary combustion. The released volatile is burned out by two OFA rows.

Summed Coal Devolatilisation Rate (kg/s)

combustion is absorbed by the furnace wall when there is no OFA. With regard to the heat flux, the air staging gives approximately 20% lower heat flux in the PCZ compared to the heat flux in the case without OFA.

0.24 HBC+OFA

HBC Only

None

0.18

0.12

0.06

(a) 0 0.5

0

2.5

4.5

6.5

Furnace Height (m) Volatile Mole Fraction (%vol)

-3e5

OFA Only

0.5 HBC+OFA

OFA Only

HBC Only

None

0.4 0.3

0.2 0.1

(b)

0 0

5

10

15

20

Furnace Height (m)

HBC+OFA

None

Fig. 4. Contours of heat flux (W/m2) in the side wall.

Fig. 6. Comparison of (a) summed devolatilization rate and (b) volatile concentration along the furnace height.

220

X. Zhang et al. / Fuel 142 (2015) 215–221

Compared with the cases with OFA, the lower stoichiometry in the PCZ caused by OFA also generates a zone with more CO. According to the Eddy–Dissipation model employed in this study, CO is a product of both volatile matter and char combustion. Therefore, both the devolatilization and char burnout rates can affect the CO concentration. Fig. 7 expresses the CO concentration along the furnace height. A better devolatilization performance and a lower stoichiometry result in more CO in the PCZ, which is converted into CO2 with the O2 from the OFA injection.

2e-6

0

HBC+OFA

HBC Only

Fig. 8. Contour of char burnout colored by burnout rate (kg/s) for ‘HBC + OFA’ vs. HBC Only.

500 HBC+OFA

Total NOx (mg/m3)

4.2.5. Char burnout performance Fig. 8 shows the comparison of char combustion rates between the case ‘HBC + OFA’ and the case ‘HBC Only’. For the case with OFA (‘HBC + OFA’), it is found that less char combusts in the central area of the PCZ and the remaining char is burned out in the OFA zone. The coal properties calibrated by the FG-DVC model give almost the same dry base mass fraction between volatile (30.85 wt%) and char (31.09 wt%). The limited primary and secondary air can only combust volatiles and a part of char in the PCZ. The rest of unburned char is burned out by the additional air injected through eight OFA nozzles. In contrast, the case without OFA (‘HBC Only’) can burn out all volatiles and more char in the PCZ. A higher char combustion rate is observed in the lower furnace because all the combustion air is fed into the PCZ. The char is consumed to the lowest level quickly once the coal particles are injected into the furnace. The better char combustion performance in the PCZ, consequently, contributes to a lower unburned carbon mass fraction at the furnace exit.

OFA Only

HBC Only

None

400 300 200 100 0

4.2.6. NOx distribution Fig. 9 presents the NOx concentration at the horizontal planes along the furnace height. By comparing the cases of ‘HBC Only’ and ‘None’, it is found that the application of HBC burners can decrease the NOx concentration in the PCZ. For the cases with OFA, the average NOx concentration shows a similar value, indicating that the effect of the HBC on the NOx reduction turns out to be insignificant when the OFA is used. In the central area of the furnace, the application of OFA lowers the stoichiometry to around 0.63 for the case ‘HBC + OFA’ and around 0.7 for ‘OFA Only’, as shown in Fig. 2. It reveals that there is not much possibility of combustion between N and O2 in either case. Fig. 9 shows a similar NOx concentration when the OFA has been employed. Although the application of HBC burners alone has a good effect on inhibiting NOx formation in the PCZ, as shown in cases without OFA, the air staging (i.e. OFA) makes a major contribution to NOx reduction once in combination with HBC burners. The fact that the lower stoichiometry in the PCZ leads to lower NOx emission from the boiler can be explained as follows. On the one hand, the lower stoichiometry suppresses the conversion of

CO (ppmv)

30000 HBC+OFA

OFA Only

HBC Only

None

20000

10000

0 0

5

10

15

Furnace Height (m) Fig. 7. Average CO concentration along the furnace height.

20

0

5

10

15

20

Furnace Height (m) Fig. 9. NOx concentration along the furnace height.

the volatile N to NOx. On the other hand, the NOx in the flue gas is reduced by the unburned char. Hence, in this study, the combination of HBC and air staging contributes approximately 20% of the NOx reduction. 5. Conclusions The characteristics of the combustion, stoichiometry, temperature, species concentration and NOx emissions in a 200 MWe tangentially-fired pulverized-coal boiler have been numerically investigated using comprehensive models for the combustion processes and NOx formation. Four cases (‘HBC + OFA’, ‘OFA Only’, ‘HBC Only’ and ‘None’) have been studied to obtain a deep understanding of a large boiler employing various low NOx technologies. The flue gas temperature, O2 concentration and NOx emission of the baseline case have been obtained and compared with the test results on-site. The good agreement implies that the employed turbulent flow, chemical reaction, radiation and NOx formation models are capable of predicting the performance of the selected boiler. The relationships between the stoichiometry, O2 concentration, combustible species, fuel particle properties and NOx emissions have been clearly demonstrated. Comparison of the investigated four cases has shown that although the HBC is capable of reducing NOx emissions, its effect is incomparable to that of the air staging technology (i.e. OFA). For the tangentially-fired pulverized-coal boiler, when both the HBC and OFA are applied, the OFA makes a major contribution to the NOx reduction. The decrease in the formation of NOx is due to the lower level of contact between the nitrogen from the fuel and the oxygen in the combustion air, especially in the central area of the furnace. It is worth highlighting that the HBC leads to a higher stoichiometry in the area close to the

X. Zhang et al. / Fuel 142 (2015) 215–221

furnace wall when combined with the OFA, which would reduce the risk of slagging when the boiler adopts the low NOx technology. The study presented here has developed a methodology that combines a devolatilization model and CFD codes for the prediction of combustion and NOx emission in a large utility boiler. The results improve the understanding of combustion, particle burnout and NOx formation when various low NOx control technologies are adopted. References [1] Gómez A, Fueyo N, Díez LI. Modeling and simulation of fluid flow and heat transfer in the convective zone of a power-generation boiler. Appl Therm Eng 2008;28(5–6):532–46. [2] Huang LY, Wen JX, Karayiannis TG, Matthews RD. Numerical prediction of high efficiency boiler heat exchanger performance. Appl Therm Eng 1998;18(11):1089–99. [3] Chaos M, Khan MM, Krishnamoorthy N, Ris JL, Dorofeev SB. Evaluation of optimization schemes and determination of solid fuel properties for CFD fire models using bench-scale pyrolysis tests. Proc Combust Inst 2011;33(2): 2599–606. [4] Papadikis K, Gu S, Bridgwater AV, Gerhauser H. Application of CFD to model fast pyrolysis of biomass. Fuel Process Technol 2009;90(4):504–12. [5] Backreedy RI, Fletcher LM, Jones JM, Ma L, Pourkashanian M, Williams A. Cofiring pulverised coal and biomass: a modeling approach. Proc Combust Inst 2005;30(2):2955–64. [6] Modlinski N. Computational modeling of a utility boiler tangentially-fired furnace retrofitted with swirl burners. Fuel Process Technol 2010;91(11): 1601–8. [7] Eddings EG, Sarofim AF, Lee CM, Davis KA, Valentine JR. Trends in predicting and controlling ash vaporization in coal-fired utility boilers. Fuel Process Technol 2001;71(1–3):39–51. [8] Chen JY, Mann AP, Kent JH. Computational modeling of pulverised fuel burnout in tangentially fired furnaces. Symp Int Combust Proc 1992;24(1):1381–9. [9] De Soete GG. Overall reaction rates of NO and N2 formation from fuel nitrogen. In: 15th Symp Int Combust Proc 1975;15(1):1093–102. [10] Kambara S, Takarada T, Toyoshima M, Kato K. Relation between functional forms of coal nitrogen and NOx emissions from pulverized coal combustion. Fuel 1995;74(9):1247–53. [11] Levy JM, Chan LK, Sarofim AF, Beer JM. NO/char reactions at pulverized coal flame conditions. In: 18th Symp Int Combust Proc 1981;18(1):111–20. [12] Smoot LD, Smith PJ. NOx pollutant formation in a turbulent coal system. In: Coal combustion and gasification. New York: Springer; 1985. p. 373. [13] Frassoldati A, Frigerio S, Colombo E, Inzoli F, Faravelli T. Determination of NOx emissions from strong swirling confined flames with an integrated CFD-based procedure. Chem Eng Sci 2005;60(11):2851–69.

221

[14] Xu H, Smoot LD, Hill SC. Computational model for NOx reduction by advanced reburning. Energy Fuels 1999;13(2):411–20. [15] Cao HL, Sun SZ, Liu YH, Wall TF. Computational fluid dynamics modeling of NOx reduction mechanism in oxy-fuel combustion. Energy Fuels 2010;24(1):131–5. [16] Truelove JS. Discrete-ordinate solutions of the radiation transport equation. ASME J Heat Transf 1987;109(4):1048–51. [17] Man CK, Gibbins JR, Witkamp JG, Zhang J. Coal characterisation for NOx prediction in air-staged combustion of pulverised coals. Fuel 2005;84(17): 2190–5. [18] Vuthaluru R, Vuthaluru HB. Modeling of a wall fired furnace for different operating conditions using FLUENT. Fuel Process Technol 2006;87(7):633–9. [19] Xu MH, Azevedo JLT, Carvalho MG. Modeling of a front wall fired utility boiler for different operating conditions. Comput Method Appl Mech 2001;190(28):3581–90. [20] Choi CR, Kim CN. Numerical investigation on the flow, combustion and NOx emission characteristics in a 500 MWe tangentially fired pulverized-coal boiler. Fuel 2009;88(9):1720–31. [21] Park HY, Baek SH, Kim YJ, Kim TH, Kang DS, Kim DW. Numerical and experimental investigations on the gas temperature deviation in a large scale, advanced low NOx, tangentially fired pulverized coal boiler. Fuel 2013;104:641–6. [22] Xue S, Hui SE, Liu TS, Zhou QL, Xu TM, Hu HL. Experimental investigation on NOx emission and carbon burnout from a radially biased pulverized coal whirl burner. Fuel Process Technol 2009;90(9):1142–7. [23] Jing JP, Li ZQ, Zhu QY, Chen ZC, Ren F. Influence of primary air ratio on flow and combustion characteristics and NOx emissions of a new swirl coal burner. Energy 2011;36(2):1206–13. [24] Khanafer K, Aithal SM. Fluid-dynamic and NOx computation in swirl burners. Int J Heat Mass Transf 2011;54(23–24):5030–8. [25] Sun S, Li ZQ, Wang ZJ, Sun R, Wu SH, Qin YK, et al. Horizontal bias combustion burner for low rank coal of China. In: Joint 7th APEC coal flow seminar & the 8th APEC clean fossil energy technical seminar, Bangkok, Thailand; 2000. p. 9. [26] Coats AW, Redfern JP. Kinetic parameters from thermogravimetric data. Nature 1964;201:68–9. [27] Warnatz J. NOx formation in high-temperature processes. Proc Eurogas. Trondheim, Norway; 1990. p. 303–20. [28] Baulch DL, Cobos CJ, Cox RA, Esser C, Frank P, Just T, et al. Evaluated kinetic data for combustion modeling. J Phys Chem Ref Data 1992;21(3):411–734. [29] Visona SP, Stanmore BR. Modeling NO, release from a single coal particle I. Formation of NO from volatile nitrogen. Combust Flame 1996;105:92–103. [30] Visnona SP, Stanmore BR. Modeling release from a single coal particle II. Formation of NO from char-nitrogen. Combust Flame 1996;106:207–18. [31] Hurt R, Sun JK, Lunden M. A kinetic model of carbon burnout in pulverized coal combustion. Combust Flame 1998;113:181–97. [32] Zolin A, Jensen AD, Jensen PA, Dam-Johansen K. Experimental study of char thermal deactivation. Fuel 2002:1065–75.