Optimization of a premixed cylindrical burner for low pollutant emission

Energy Conversion and Management 99 (2015) 151–160

Contents lists available at ScienceDirect

Energy Conversion and Management journal homepage: www.elsevier.com/locate/enconman

Optimization of a premixed cylindrical burner for low pollutant emission Dong-Fang Zhao a, Feng-Guo Liu a,⇑, Xue-Yi You b,⇑, Rui Zhang a, Bin-Long Zhang a, Gui-Long He c a

School of Energy and Safety Engineering, Tianjin Chengjian University, Tianjin 300384, China School of Environmental Science and Engineering, Tianjin University, Tianjin 300072, China c China Quality Supervising Engineering and Test Center for Gas Appliances (CGAC), Tianjin 300384, China b

a r t i c l e

i n f o

Article history: Received 9 December 2014 Accepted 13 April 2015 Available online 28 April 2015 Keywords: Optimal design CFD Premixed burner Low emission NOx and CO

a b s t r a c t A premixed cylindrical burner is numerically and experimentally investigated to realize low pollutant emission. The geometrical parameters of nozzle exit position and nozzle diameter are optimized by using a validated Computational Fluid Dynamics model. The natural gas-air mixing in the mix chamber indicates that the uniformity of methane concentration increases with the increase of distance from ejector outlet. It is found that the nozzle exit position at 3.0 mm improves the overall performance of premixed cylindrical burner, when nozzle diameter is not less than 1.6 mm. The emission characteristics of nitrogen oxides and carbon monoxide are also examined by experimental approach. It is found that load factor has a great influence on nitrogen oxides and carbon monoxide emissions, but the effect is gradually disappeared when air coefficient is not less than 1.4. When nozzle exit position is 3.0 mm, nozzle diameter is not less than 1.6 mm and air coefficient is not less than 1.4, the emissions of nitrogen oxides and carbon monoxide are less than 20 ppm and 50 ppm, respectively. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction In recent years, environmental problems are becoming more and more attention. Air pollution is one of the major problems of modern world. The low-pollutant emission technology has received increasing attentions. One of the main pollutants is NOx formed by the oxidization of nitrogen during combustion process. Wei et al. [1] investigated the effects of operating parameters on the NOx emission and found out the optimal operating conditions to reduce the NOx emission for high-temperature air combustion furnaces. The concentration of NOx emission reduces by 25.45% in their optimal operating condition. Gonca [2] conducted a steam injected method to supply ethanol–diesel blend fuel for a diesel engine. The results showed that the NO emission is reduced to 34% compared with conventional diesel engine (D) and steam injected diesel engine (D + S20). Kesgin [3] researched the relationship between design and operational parameters and the NOx emission of a turbocharged natural gas engine. By reducing the charge temperature, it was found that there is an increase in the excess air ratio, which causes a significant decrease in NO emission and the concentration of emissions is decreased to a level to meet the international standards. Coelho described a numerical simulation of Eulerian particle flamelet mode in mild combustors to calculate NO emission in a post-processing stage. The combustor was ⇑ Corresponding authors. E-mail addresses: [email protected] (F.-G. Liu), [email protected] (X.-Y. You). http://dx.doi.org/10.1016/j.enconman.2015.04.039 0196-8904/Ó 2015 Elsevier Ltd. All rights reserved.

characterized by relatively lower NOx emission [4]. The further investigation [5] pointed out that N,N0 -diphenyl-1, 4-phenylenediamine (DPPD) antioxidant additive can reduce NOx emission significantly. In this case of the addition of 0.15% (m) DPPD additive in JB5, JB10, JB15 and JB20 (Jatropha methyl ester is blended with diesel at 5%, 10%, 15% and 20% by volume), the reduction in the NOx emission is 8.03%, 3.503%, 13.56% and 16.54%, respectively, compared to biodiesel blends without the additive under the full throttle condition. Although an increase in CO emission with the addition of DPPD antioxidant to all Jatropha biodiesel blends was observed, the value is low in comparison to diesel emission. However, the low-NOx emission approaches are based on non-premixed combustion in the above investigations, which focus mainly on engine and biodiesel fields. Premixed combustion can notably reduce NOx emission because it does not produce Fuel NO (F-NO) and Prompt NO (P-NO), which decrease the chance of Themal NO (T-NO) generation. Authors’ research group have designed and developed a plate-type premixed burner, which have the advantages of high efficiency and low emissions. After increasing the mixing effect in mixing chamber, NOx and CO emissions was found to be less than the conventional burners in present market [6,7]. It is characterized by high temperature, relatively short flame and excellent ductility. To improve premixed combustion behavior, some of research devoted to study the geometry parameters of burning system. Zhang et al. [8] optimized the geometry parameters of diversion plate to improve the uniformity at the outlet of gas mixing system. The

152

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

hole diameters and the arrangement of a cylindrical multi-hole premixed burner were investigated by Lee et al. [9]. The results showed that the NOx and CO emissions are less than 40 ppm and 30 ppm for a 0% O2 basis, respectively. Panwar et al. [10] evaluated the performance of premixed type industrial burner with swirl vane for mixing the air and gas. However, the premixed cylindrical burner (PCB) with ejectors of mixing the air and natural gas has not been studied. In China, it has experienced a process to establish NOx and CO emissions standards for gas appliance. In the early 1990s, China formulated her own standards to the European standards gradually. As environmental awareness increasing, the national standards of domestic gas instantaneous water heater are established step-by-step. This makes it necessary to investigate low NOx emission gas combustors. Traditional optimal ways of combustors are based on empirical models, and a large number of experiments are required. It costs not only a lot of time, but also numerous resources. With the development of computer, the numerical simulation is becoming an attractive method for its high flexibility and low cost application. The computational fluid dynamics (CFD) approach is capable of visualizing the detailed information of flow field to optimize its performance. Sharfi and Boroomand [11] validated the deviation between the numerical model and experiment data. The maximum difference between the numerical and experimental results is about 9.7% and 10.6% for the two models, respectively. Zhu et al. [12] employed the CFD technique to research the ejector geometry parameters, and showed the optimum primary nozzle position and converging angle in particular operating condition. In order to study a supersonic air ejector, Hemidi et al. [13] dealt with the comparisons between CFD and experiments. Good validation results were obtained for a wide range of operating conditions. Arghode et al. [14] numerical investigated the combustion characteristics for application to gas turbine combustors by commercial software FLUENT. It was revealed that the numerical method was able to capture the overall flow field behavior and provide insights to achieve reactions in volume distributed combustion regime to reduce NOx and CO emissions. Sukumaran and Kong [15] used the CFD modeling to study the combustion process inside the burner, and the reduced mechanism was able to predict NOx emission for different feedstock and operating conditions. Chui et al. [16] investigated burner design concepts of reducing NOx formation via improved staging by CFD. It indicated that a new burner design approach can potentially reduce the NOx compared to the existing design methods. The present developed PCB adopts a new design principle of premixed combustor. It has the advantages on simple manufacture, convenient operation, low NOx and CO emissions and highly combustion efficiency compared to the existing burner. The stainless steel radial heat exchanger and compact aluminum heat exchanger are used. The purpose of the present study is to optimize

geometry parameters of PCB by CFD. An innovative approach of random sampling method is introduced to analyze the mixing effect in mixing chamber and the experiments are conducted to study the combustion characteristics. 2. Description of the premixed cylindrical burner and experimental rig Fig. 1 is the schematic diagram of our designed PCB, which consists mainly of a mixer and a cylindrical burner. The mixer has 6 major components: gas distribution chamber, nozzles, air chamber, ejectors, insulating chamber and mixing chamber. For reducing gas static pressure before entering the ejectors, equal number nozzles are located at the outlet of gas distribution chamber. The ejectors described by Liu et al. [6] are arranged in circle. The distance between nozzle and ejector is defined as NXP (nozzle exit position). Mixing chamber, starting from the ejector outlets, is covered with full of distribution pores on the surface. Insulating chamber between distribution pores and fire holes plate is used to trap the flame’s heat. In working conditions, the natural gas going through the nozzles of gas distribution chamber is injected into ejectors. The fan supplied air is entrained by natural gas at the nozzle outlets. After further mixing in the mixing chamber, the premixed gas through the distribution pores gets to the surface of fire holes plate to realize premixed combustion. The design heat load of PCB is 27 kW. For premixed cylindrical burner (PCB), the mixing effect of natural gas plays an important role in realizing well premixed combustion. A schematic diagram and a photograph of the experimental rig are shown in Fig. 2a and b, respectively. The experimental rig consists of a stainless steel radial heat exchanger and three systems: a natural gas supply system, a water supply system and a fan air supply system. The natural gas supply system has a pressure regulator to maintain the required gas pressure and an Electromagnetic valve to adjust the gas flow rate of system. There are two thermometers to measure the inlet and outlet water temperature of heat exchanger in water supply system. The parameters of the fan (Model: FLW85-55H01), which is used in the fan air supply system, are the control voltage 12 V and the security voltage 36 V. The exhausted gas is evaluated by the emissions of CO, NOx and O2, which are measured by Testo 350 gas analyzer. The specifications of Testo 350 gas analyzer are listed in Table 1. The laboratory ambient temperature is about 20 °C with the concentration of CO and CO2 no more than 0.002% and 0.2% respectively. 3. CFD model In order to improve the accuracy and reduce time consumption of solution, several assumptions are made before numerical

Insulating chamber Fire holes plate Mix chamber Gas distribution chamber

Gas inlet

Air chamber

Mix chamber

Ejector Air inlet Nozzle Fig. 1. Schematic diagram of PCB.

Hot water tube

Pore of distribution

153

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

12 2

1 gas

4

4

5

3

exhaust gas 8 11 6

water

10 7

9 8

3

water

1. gas supply system, 2. pressure regulator, 3. flow meter, 4. U shape pressure gauge, 5. Electro magnetic valve, 6. fan, 7. pressure gauge, 8. thermometer, 9. premixed burner, 10. power supply, 11. heat exchanger, 12. gas analyzer.

(a) Diagram of the experimental rig.

(b) Photograph of the experimental burner. Fig. 2. Schematic diagram of experimental rig and burner.

simulation. The air and natural gas are treated as Newtonian fluid by satisfying the law of Newton inner friction. They are assumed to be incompressible for present low Mach number flow. The natural gas is simplified as methane, and the effects of radiation are neglected in the process of mixing. The commercial software FLUENT 14.0 and Gambit 2.4 are used as the CFD solver and grid generator, respectively. The governing equations, including the equations of continuity, momentum, species transport and turbulence model [17] are solved by using the ‘‘Segregated implicit’’ solver and second order upwind discretization scheme. The SIMPLEC algorithm of pressure–velocity coupling is adopted to obtain the pressure field. Moreover, the turbulence model of SST k-x [18,19] is finally selected for its stability and ability to offer relatively high accuracy in cylindrical fields [20–22]. SST k-x model is a two layer turbulence model, which works with k-x model in the sub-layer and logarithmic part of boundary layer and with k-e in wake region [20]. The mainly difference between this form and the standard k-x model is that an additional cross-diffusion term Dx appears in the x equation [18], which is expressed the form:

Dx ¼ 2ð1  F 1 Þqrx;2

1 @k @ x x @xj @xj

ð1Þ

where q is density, k is the turbulence kinetic energy, x is the specific dissipation rate and the blending function F1 is:

  F 1 ¼ tanh /41

ð2Þ h



 i pffiffi l where /1 ¼ min max 0:09kxy ; q500 ; r 4qDkþ y2 , y is the distance from y2 x h x;2 x i 20 1 1 @k @ x (Dþ the nearest wall and Dþ x ¼ max 2q rx;2 x @x @x ; 10 x is the j

j

positive portion of the cross-diffusion term). The constant rx;2 ¼ 1:168 and l is the turbulent viscosity. The boundary conditions are set as ‘‘mass flow rate’’ condition at the inlet of natural gas and air, and ‘‘pressure outlet’’ is at the outlet of distribution pores. The outlet condition is zero static pressure because it is difficult to measure the pressure of the distribution pores before numerical simulation. Standard wall function and no slip surface are employed at the wall regions. 4. Model validation 4.1. Grid independent test To reduce the total number of grid cells, the computational region is divided into ten sub-regions. The grid independence test is conducted between two grid systems. The cell numbers of the two systems are 2.13  106 and 4.50  106, respectively. In this case, the mass flow rate of gas and air inlets is 1.89  104 kg/s and 3.96  103 kg/s, respectively. On the central axis of the mixing chamber, eight sample points are extracted from the numerical results. Fig. 3a shows the location of the sample points. Fig. 3b

154

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

Table 1 The parameters of gas analyzer.

Gas chamber

Measurements

Method for measuring

Range

Resolution

O2 CO NO Gas temperature (°C)

Electrochemical sensor Electrochemical sensor Infrared Thermocouple

0–25% 0–104 ppm 0–4  103 ppm 0–650 °C

0.1% 1 ppm 1 ppm 1 °C

1

2 30mm

3

4

5 30mm

30mm

30mm

6 30mm

7

Nozzle

PCB Gas inlet

8

In

23mm

30mm

30mm

Out Central axis

Ejector Ejector

Nozzle

Air inlet

(a)

Fig. 4. Grid of the mixer: the cell number is 1.4416  105, 6.6639  104 and 2.13  106 for gas chamber, ejector and PCB, respectively.

2.13e06 cells 4.50e06 cells

16.0

380

Experimental results 12.0

Static pressure of gas inlet (Pa)

Velocity magnitude (m/s)

20.0

8.0

4.0

0.0

1

2

3

4

5

6

7

8

Position number on the central axis

Numerical results

360

340

320

300

(b) Fig. 3. Velocity magnitude of the two grid systems: (a) The interval positions of monitoring points in central axis of mix and gas distribution chamber. (b) The velocities magnitude with number of cells for 2.13  106 and 4.50  106.

280

1.1

1.2

1.3

1.4

1.5

1.6

1.7

α Fig. 5. Experimental and numerical static pressures at the gas inlet.

shows the velocity magnitudes of different sample points. It is observed that the results of the two grid systems show close agreement, which means that the influence of the grid size on the results is negligible. Therefore, the grid system with 2.13  106 cells is adopted. Fig. 4 illustrates the grid structure used for the three-dimensional mixer. 4.2. Model validation The numerical results of static pressure at the gas inlet are compared with the experimental data. a is defined as air coefficient by:

a ¼ ðactual air suppliedÞ=ðtheoretical air suppliedÞ

ð3Þ

_ air and m _ gas are the mass flow rate of air and gas, respecwhere m tively. qair and qgas are the density of air and gas, respectively. For this case the mass flow rate of gas inlet 1.8  104 kg/s, Fig. 5 shows that the discrepancies between the numerical and experimental results are small, and the maximum discrepancy is about 3.1% at a = 1.6. This means that the numerical model is feasible. The reasons of difference between the numerical and experimental results may originate from the model simplification and existence of the systematic error mainly from instruments, operation conditions of experimental environment, approximate algorithm and reading by operators, respectively, in the experimental procedure.

The calculation procedure in experiment by:

a¼

uvol;air uvol;air  uvol;emi

ð4Þ

where uvol;air is the volume fraction of oxygen in the air. And uvol;air is the volume fraction of oxygen in emission, which is measured by Testo 350 gas analyzer. The value of a is known and the mass flow rate of air is calculated by:

_ air ¼ aqair m

_ gas m

qgas

ð5Þ

5. Results and discussions Liu et al. [6] studied NXP and nozzle diameter for optimal design of household appliance burner of fuel-gas mixing in a single ejector. It was found that the pressure of the gas inlet increases with the distance between nozzle outlet and ejector inlet when nozzle diameter keeps constant. And the nozzle diameter had significant effects on the pressure of gas inlet. The results of Zhu et al. [12] indicated that the parameter NXP is importance to the ejector entrainment ratio. However, there is no fixed NXP or d (nozzle

155

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

8.0

A

No .2

No.7

14.0

6.5

10.5

No.1

.6 No

+

R2

NXP=0

-

d 4.0

14.5

23.0

.3 No

8.5

No .5

33.5

118.5

No.4

15.0

A-A

A

1,200.0

310.0

1,000.0

260.0

Static pressure (Pa)

Static pressure (Pa)

Fig. 6. The definition parameters NXP and d of PCB.

800.0

gas inlet air inlet 600.0

400.0

200.0

210.0

NXP= -3.0 NXP= -1.5 NXP= 0.0 NXP= 1.5

160.0

110.0

-3

-1.5

0

60.0

1.5

1

2

diameter) meeting all kinds of mixer for the complex characteristic of the flow inside the ejector and the parameters NXP and d of premixed cylindrical burner (PCB) are optimized to improve the performance. Fig. 6 is the definition of parameters NXP and d of PCB. And the ranges of NXP and d are from 3.0 mm to 1.5 mm and 1.3 mm to 1.7 mm, respectively. The experiments of PCB is conducted based on the optimal parameters of numerical simulation, and the burning characteristic is investigated at different heat load factors (LF) and a. 5.1. Effects of NXP and d on static pressure

5

6

7

320.0

290.0

260.0

No.1 No.3 No.5 No.7

230.0

Fig. 7a depicts the variation of the inlet static pressure with respect to NXP at d = 1.6 mm. There is a slight oscillation for the static pressure of air inlet with the increase of NXP, which implies that the impact of NXP on air inlet is negligible. However, the static pressure decreases noticeable with the increase of NXP for gas inlet. It is more than 1000 Pa when NXP < 1.5 mm. The surpass pressure is used to conquer the resistance of nozzle and pipe to provide flow velocity at fire hole. Fig. 7b shows the static pressure of all nozzle outlets at different NXP at d = 1.6 mm. It is found that the static pressure of nozzle outlets is more uniform at NXP = 1.5 mm, and where the average value is relatively smaller at NXP = 3.0 mm compared to other cases. In this case, the static pressure of nozzle outlets is lower and it is also good for controlling the entertainment ratio defined _ air =m _ gas (where m _ air and m _ gas is the mass flow rate of air and as m gas, respectively) at ejector inlets. The pressure difference of air (and gas) inlet and nozzle outlet is the lowest when

4

Fig. 7b. Effect of NXP on the static pressure of nozzle outlets at d = 1.6 mm.

Static pressure (Pa)

Fig. 7a. Effect of NXP on the static pressure of gas and air inlets at d = 1.6 mm.

3

The number of nozzle

NXP

200.0 1.2

1.3

No.2 No.4 No.6

1.4

1.5

1.6

1.7

1.8

d (mm) Fig. 8a. Effect of d on static pressure of nozzle outlets at NXP = 3.0 mm.

NXP = 1.5 mm, however, it goes against suitable entertainment ratio of air when NXP P 0.0 mm and it is not recommended in the following study. Figs. 8a and 8b show the effect of different d on the static pressure of nozzle outlets, gas and air inlets at NXP = 3.0 mm. Fig. 8a shows the uniformity of static pressure and mean average value at nozzle outlets. When d = 1.7 mm, the uniformity is the best compared to other cases. There is a minimum average static pressure

156

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

at d = 1.6 mm, which will help to improve the entertainment ratio. The gas flow resistance and velocities at nozzle outlets reduce with the increase of d. It results in the decrease of total pressure at gas inlet. As a result, with identical gas flow rate, the static pressure is greatly reduced at nozzle outlets of d = 1.6 mm compared with d less than 1.5 mm. However, when d = 1.7 mm, the velocities of gas at nozzle outlets are further reduced. This results in that a lot of air is blocked at the ejector inlets and the static pressure of nozzle outlets is increasing. In Fig. 8b, it is found that the effect of d on the static pressure of air inlet is negligible. However, the static pressure of gas inlet decreases largely with the increase of d. The local resistance of gas flow decreases with the increase of nozzle diameter, and that causes the increase of entertainment ratio. Based on the requirements of GB50028-2006 (town gas design specifications of China), the normal pressure of natural gas is about 2000 Pa. After 2,200.0

gas inlet air inlet

Static pressure (Pa)

1,800.0

1,400.0

1,000.0

600.0

200.0

1.3

1.4

1.5

1.6

1.7

d (mm) Fig. 8b. Effect of d on static pressure of gas inlets at NXP = 3.0 mm.

the solenoid valve, the pressure of natural gas is decreased to 1000 Pa. This requires the resistance of PCB is less than 1000 Pa, and the nozzle diameter of d P 1.6 mm is recommended. 5.2. The mixing of gas and air The total number of distribution pore is 360. To further illustrate the concentration distribution of gas after mixing at NXP = 3.0 mm and d = 1.6 mm, the random sampling method is adopted to estimate the population through the samples, which is randomly extracted from the overall database. Its process is as follows: Step 1: Number the overall data, direction and order. The number of distribution pore is divided into two directions, which consist of axis and perimeter. As shown in Fig. 9, the two directions are named as 1–18 and 1–40, respectively. Step 2: Select the starting point. The rows and columns are arbitrarily selected at the beginning of the sample points in Random Number Table. In axial direction, the digit of second row and fifth column is extracted to act as the first sample point. In perimeter direction, the digit of seventh row and third column is extracted to act as the first simple point. Step 3: Form samples. The sample dates are extracted one by one from left to right. If the values of date points out of the labeled range, the points are skipped in Random Number Table. In axial direction, the result of randomly sampling is shown in Table 2. Fig. 10a shows the mass fraction of CH4 at distribution pores of randomly sampling in axial direction. It is found that the uniformity of CH4 fraction is poor and the mean concentration is relatively lower at serial number 1–10, compared to that of others (No. 11–18). It increases with the serial number increase of distribution pore. When the serial number is over 10, the component of CH4 rises to more than 4% and the uniformity attaches significant improvement. It reveals that the distance of axial direction is one of the mainly factors with respect to the uniformity of CH4 concentration.

Axial section

20

20

20

20

20

Perimeter

a

20

20

20

g

f

e

Axial direction

d

c

b

Cut-off rule

h

3 2 1 5 3 1 4 2

Pore of distribution Perimeter direction Fig. 9. The distribution and number of pore and section.

Table 2 The randomly sampling of axial direction. Axis

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

No. (pore)

15

36

21

16

15

26

33

32

7

20

5

26

15

34

23

20

31

28

157

0.050

0.050

0.040

0.040

Mass fraction of CH4

Mass fraction of CH4

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

0.030

0.020

0.020

0.010

0.030

0.010 0

2

4

6

8

10

12

14

16

18

0

4

8

12

16

20

24

28

32

36

40

Perimeter number

Axial number

Fig. 10b. The concentration distribution of CH4 at the pores in perimeter direction.

Fig. 10a. The concentration distribution of CH4 at the pores in axial direction.

position of thirteenth row and sixth column. This may be due to the fact that the numerical dissipation is modified in the process of solution.

In perimeter direction, the process is completely coincident with axial direction on randomly sampling and the result of randomly sampling is shown in Table 3. The result of CH4 mass fraction is depicted in Fig. 10b. It is observed that the concentrations of CH4 at position number h ðh ¼ 4n þ 1; ð0 6 n 6 9; n 2 NÞÞ are relatively lower than that of others. This indicates that the entertainment ratio is small for larger pore diameter and larger pore diameter may cause larger kinetic energy loss during turbulent mixing in the distribution pores. It has a negative effect on burning performance if the concentration of CH4 is too large or too small. To find the optimum array of distribution pores, a further study is required. The concentration of CH4 at different cross-sections in the mixing chamber is shown in Fig. 11. The specific position of the axial section is illustrated in Fig. 9. Results show that the maximum mass fraction is found near the central axis and it increases with the increase of the distance starting from the ejector outlets. It reaches the maximum at the cross-section d. Then, it decreases until the end cross-section of mixing chamber. It is found that the position between the inlet and outlet of the ejector periphery reduces further diffusion of methane. At the same time, the long spreading distance contributes to the mixing of methane and air. Since high concentration of CH4 causes incomplete combustion, and low concentration causes a massive calories loss by excess air, the quantity of distribution pores near the position of ejectors should be less than those near the end of the mixing chamber for homogeneous combustion. Fig. 12 shows the concentration distribution of CH4 at the distribution pores. It is found that the mixed effect at the position far away from ejectors is better than that near it, where a large amount of methane is collected near the central axis. With the spreading of mixed gas, it gradually diffuses from the region of central axis to distribution pores. The number of pores whose concentrations is closed to the average value is about two-third of pores. It reveals that the uniformity of CH4 is able to achieve by further mixing. However, there is a minimum concentration near the

5.3. Experimental investigation 5.3.1. Effect of NXP and a on the emission of NOx and CO Figs. 13a and 13b shows the effect of NXP on the emission concentration of NOx and CO. It is found that the emission concentration of NOx shows obviously decreases with the increases of a when NXP keeps constant. The emission concentration of NOx at a = 1.18 is approximately 16 times larger than that at a = 1.63 when NXP = 1.5 mm, which indicates that a has great effects on NOx emission. But this influence disappears gradually; especially, when a P 1.45. The variation of CO concentration with NXP is almost identical to that of NOx in Fig. 13b. When the air coefficient a is more than 1.25, the concentration of CO emission of NXP = 1.5 mm is about 1.2 times that of NXP = 3.0 mm. Combined with Figs. 7a and 7b (NXP < 0.0 mm), it is concluded that NXP = 3.0 mm is an optimal position for low emission of NOx and CO. 5.3.2. Effect of nozzle diameter Fig. 14a shows that the emission concentration of NOx at d = 1.6 mm is lower than d = 1.7 mm when a 6 1.27. This is caused by the velocity and static pressure of natural gas on the nozzle outlets. Fig. 8a shows that the average static pressure at d = 1.6 mm is less than that at d = 1.7 mm, which forms relatively large entertainment ratio. In the same condition of flow rate, the average velocity at d = 1.6 mm is faster than that at d = 1.7 mm, which leads to turbulence intensity enhanced and blend well. So it is found that the NOx emission is reduced as a result of complete mixing. However, this effect is just opposite when a > 1.27. This is due to the fact that the excess air entertained into air chamber improves the total pressure of air. Gas resistance is increased when it pass through the ejectors at d = 1.7 mm. In this case, the mixed effect at d = 1.7 mm is better than that at d = 1.6 mm.

Table 3 The randomly sampling of perimeter direction. Perimeter position

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

No. (pore)

3

18

17

2

7

14

9

12

3

14

3

18

11

14

7

2

7

18

13

4

Perimeter position

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

No. (pore)

1

12

15

8

3

16

5

6

15

18

7

12

11

8

3

16

13

8

7

4

158

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

Fig. 14b shows the characteristics of the CO emission concentration at NXP = 3.0 mm. When a 6 1.26, it can be observed that the trend lines are consistent with that of NOx concentration in Fig. 14a. But when a > 1.26, the effect of different d on the CO emission concentration is almost disappeared. 5.3.3. Effect of load factor The load factor is defined as LF = (actual heat load/design heat load)  100%. In Fig. 15a, the NOx content in the exhaust gas increases with the increase of LF when a keeps constant. This suggests that the combustion performance of PCB is becoming bad as actual heat load approached the design condition. NOx emissions is dramatically decline at low LF, when a 6 1.4. This

effect is waning with the increase of a. When a > 1.4, the NOx emission concentration is less than 20 ppm. This is because TNOx generation is mainly affected by temperature and O2 concentration in PCB. The increase of O2 concentration helps the generation of T-NOx, and it will decrease the flame temperature too, which goes against T-NOx generation. The gas velocity in fire holes increases with the increase of a, which makes the residence time of combustion products shortened in high-temperature region. And the heat loss is enhanced as the flame cone grows long in a premixed combustion, which forms low-temperature environment in PCB and reduces NOx emission. Therefore, it is confirmed that appropriate air coefficient helps to reduce NOx emissions.

(a) Axial position a.

(b) Axial position b.

(c) Axial position c.

(d) Axial position d.

(e) Axial position e.

(f) Axial position f.

(g) Axial position g.

(h) Axial position h.

Fig. 11. The concentration of CH4 at different axial cross-sections in mixing chamber.

159

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

Axial direction

(a) Three-dimensional view.

(b) View from mass fraction axis.

Fig. 12. Predicted distribution of CH4 mass fraction at distribution pores.

90.0

90.0

NXP= 0.0

d d=1.6mm

NXP= -1.5

d=1.7mm d

NXP= -3.0 60.0

NOx (ppm)

NOx (ppm)

60.0

30.0

30.0

0.0 1.1

1.2

1.3

1.4

1.5

1.6

0.0 1.1

1.7

1.2

a

1.3

1.4

1.5

a

Fig. 13a. Effect of NXP on NOx emission concentration at d = 1.6 mm.

Fig. 14a. Effect of d on NOx emission concentration at NXP = 3.0 mm.

330.0

310.0

NXP= 0.0

dd=1.6mm

NXP= -1.5

d=1.7mm d

NXP= -3.0 230.0

CO (ppm)

CO (ppm)

210.0

130.0

110.0

10.0 1.1

1.2

1.3

1.4

1.5

1.6

1.7

a

30.0 1.1

1.2

1.3

1.4

1.5

a

Fig. 13b. Effect of NXP on CO emission concentration at d = 1.6 mm.

Fig. 14b. Effect of d on CO emission concentration at NXP = 3.0 mm.

In Fig. 15b, it is found that the fall range of CO emission concentration is very noticeable during a = 1.2 to a = 1.4 when LF keeps constant. When a > 1.4, the effect of LF on CO emission concentration is becoming weak. The low gradient curve

at LF = 37.5% illustrates that the decrease of CO emission is not significant by increasing a in the case of low LF. The reason is that the CO emission is small, and it is not sensitive to the change of a.

160

D.-F. Zhao et al. / Energy Conversion and Management 99 (2015) 151–160

(2) The LF has a great influence on NOx and CO emissions, but this influence is becoming weak gradually when a > 1.4. When LF < 37.5%, the degree of influence is weak even with the decrease of a. (3) When a P 1.4, NOx and CO emissions are less than 20 ppm and 50 ppm at NXP = 3.0 mm and d P 1.6 mm, respectively. (4) In mix chamber, the uniformity of methane concentration increases with the increase of distance from ejector outlet.

90.0

LF=100% LF=75.0% LF=50.0% LF=37.5%

NO x (ppm)

60.0

Acknowledgements

30.0

0.0 1.1

1.2

1.3

1.4

1.5

1.6

1.7

a Fig. 15a. Load factor effect on the NOx emission concentration at d = 1.6 mm and NXP = 3.0 mm.

The authors gratefully acknowledge the support of scientific and technological cooperation project of Wuqing District and Tianjin University; project (NR2013K04) supported by Beijing Key Lab of Heating, Gas Supply, Ventilating and Air Conditioning Engineering, China; project (20130909) supported by the Higher School Science and Technology Development Fund of Tianjin, China. References

300.0

LF=100% LF=75.0% LF=50.0% LF=37.5%

CO (ppm)

200.0

100.0

0.0 1.1

1.2

1.3

1.4

1.5

1.6

1.7

a Fig. 15b. Load factor effect on the CO emission concentration at d = 1.6 mm and NXP = 3.0 mm.

6. Conclusions The numerical and experimental study is used to investigate the mixing and combustion performance of PCB. The effects of mixing on burning characteristic are discussed and the optimal geometric parameters and the operation conditions of PCB are obtained. The main results are summarized as follows: (1) The effect of NXP and d on the static pressure of air inlet is negligible. However, with respect to the natural gas inlet, the ejector performance is improved when nozzle outlet moves away from the mixing section and it reaches the best at NXP = 3.0 mm. The optimum diameter of nozzle is d P 1.6 mm for reducing the resistance of gas flow. In addition, when a 6 1.27, the emission concentration of NOx and CO of d = 1.6 mm is lower than that of d = 1.7 mm. But, when a > 1.27, the emission concentration of NOx of d = 1.6 mm is higher than that of d = 1.7 mm.

[1] Wei Z, Li X, Xu L, Tan C. Optimization of operating parameters for low NOx emission in hightemperature air combustion. Energy Fuels 2012;26(5):2821–9. [2] Gonca G. Investigation of the effects of steam injection on performance and NO emissions of a diesel engine running with ethanol–diesel bend. Energy Convers Manage 2014;2014(77):450–7. [3] Kesgin U. Study on prediction of the effects of design and operating parameters on NOx emissions form a leanburn natural gas engine. Energy Convers Manage 2003;44(4):907–21. [4] Coelho PJ, Peters N. Numerical simulation of a mild combustion burner. Combus Flame 2001;124(3):503–18. [5] Palash SM, Kalam MA, Masjuki HH, Arbab MI, Masum BM, Sanjid A. Impacts of NOx reducing antioxidant additive on performance and emissions of a multicylinder diesel engine fueled with Jatropha biodiesel blends. Energy Convers Manage 2014;2014(77):557–85. [6] Liu F, You X, Wang Q, Zhang R. On optimal design and experimental validation of household appliance burner of low pollutant emission. Energy Convers Manage 2013;2013(76):837–45. [7] Docquier N, Candel S. Combustion control and sensors: a review. Prog Energy Combust 2002;28(2):107–50. [8] Zhang T, Liu F, You X. Optimization of gas mixing system of premixed burner based on CFD analysis. Energy Convers Manage 2014;2014(85):131–9. [9] Lee S, Kum SM, Lee CE. An experimental study of a cylindrical multi-hole premixed burner for the development of a condensing gas boiler. Energy 2011;36(7):4150–7. [10] Panwar NL, Salvi BL, Siva Reddy V. Performance evaluation of producer gas burner for industrial application. Biomass Bioenergy 2011;35(3):1373–7. [11] Sharfi N, Boroomand M. An investigation of thermo-compressor design by analysis and experiment: Part 1. Validation of the numerical method. Energy Convers Manage 2013;2013(69):217–27. [12] Zhu Y, Cai W, Wen C, Li Y. Numerical investigation of geometry parameters for design of high performance ejectors. Appl Therm Eng 2009;29(5–6):898–905. [13] Hemidi A, Henry F, Leclaire S, Seynhaeve JM, Bartosiewicz Y. CFD analysis of a supersonic air ejector. Part I: experimental validation of single-phase and twophase operation. Appl Therm Eng 2009;29(8–9):1523–31. [14] Arghode VK, Gupta AK, Bryden KM. High intensity colorless distributed combustion for ultra low emissions and enhanced performance. Appl Energy 2012;2012(92):822–30. [15] Sukumaran S, Kong SC. Modeling fuel NOx formation from combustion of biomass-derived producer gas in a large-scale burner. Combust Flame 2013;160(10):2159–68. [16] Chui EH, Majeski AJ, Douglas MA, Tan Y, Thambimuthu KV. Numerical investigation of oxy-coal combustion to evaluate burner and combustor design concepts. Energy 2004;29:1285–96. [17] Shih Y, Hou H, Chiang H. On similitude of the cross flow fan in a split-type airconditioner. Appl Therm Eng 2008;28(14–15):1853–64. [18] FLUENT Inc., Fluent user’s guide. Version 6.3, Lebanon, NH, USA; 2006. [19] Menter FR. Two equation eddy-viscosity turbulence models for engineering applications. AIAA J 1994;32(8):1598–605. [20] Agrawal L, Mandal JC, Marathe AG. Computations of laminar and turbulent mixed convection in a driven cavity using pseudo-compressibility approach. Comput Fluids 2001;30(5):607–20. [21] Veynante D, Vervisch L. Turbulent combustion modeling. Prog Energy Combust 2002;28:193–266. [22] Šarlej M, Petr P, Hájek J, Stehlík P. Computational support in experimental burner design optimisation. Appl Therm Eng 2007;27(16):2727–31.