CC BY-NC-ND
0Available online at www.sciencedirect.com
m\0 ScienceDirect
Energy Procedía 4 (2011) 892-899
Energy Procedía
www.elsevier.com/locate/procedia
GHGT-10
Lean Flammability Limit for Oxy-Fuel Fired Pulverized Coal
Combustion Systems.
Masayuki Taniguchia*, Tsuyoshi Shibataa, Kenji Yamamotoa, Christian Kuhrb,
Osamu Itoa 1*
a Energy and Environmental Systems Laboratory, Power Systems Company, Hitachi, Ltd., 7-2-1 Omika-cho, Hitach, Ibaraki, 319-1292, Japan b Hitachi Power Europe GmbH, Schifferstraße 80, 47059 Duisburg, Germany
Abstract
We developed a model based on fundamental experimental data to predict lean flammability limit, L, and flame propagation velocity, Sb, for oxy-fuel combustion conditions. The basic model system consisted of two particles. One side of the two particles burns first, then, the other particle is ignited by the heat of combustion of the one burning particle. This phenomenon was defined as flame propagation. We analyzed at what distance the first burning particle could ignite the next particle (flame propagation distance d, and related to L), and how fast the first burning particle could ignite the next particle (flame propagation time s, and related to Sb) under various conditions. The proposed model was verified with data of both fundamental and pilot-and actual-scale experiments. We also applied the model to develop burner systems for lignite-fired oxy-fuel combustion. Local Sb and L near the ignition points of the burner systems could be analyzed from the concentration and temperature profiles of the general CFD results (k-s method). Flame stability was judged by the calculated Sb and L profiles, and past results of blow-off limits obtained with actual- and pilot-scale experiments. We call this proposed technique flammability analysis. By using combination with the technique and Large Eddy Simulation, we could quickly clarify points for improvement of the burner systems. The calculated results were applied to a DS®T-burner designed by Hitachi Power Europe, installed at Schwarze Pumpe pilot plant.
© 2011 Eublished by Elsevier Ltd. "
Keywords; pulverized coal; oxy-fuel combustion; large eddy simulation; flame propagation velocity; lean flammability limit
1. Main text 1. Introduction
Recently, development of oxy-fuel combustion technology has been particularly active using pilot-scale plants [1,2]. Fuel ignition properties are fundamental combustion performance parameters for engineering design of combustion systems. Improving ignition performance leads to improved combustion performances including such items as, expanding the turn-down ratio, expanding the application fuel properties, maintaining safety of the fuel
* Corresponding author. Tel.: 81-29-276-5889; fax: 81-29-276-5622. E-mail address: masayuki.taniguchi.xc@hitachi.com.
doi:10.1016/j.egypro.2011.01.134
supplying system, improving the combustion efficiency, and reducing environmental pollutants such as NOx. Usually, for oxy-fuel combustion systems, a mixture of exhaust flue gas and oxygen is used as the combustion supporting gas [1]. Oxygen concentrations are usually variable for oxy-fuel combustion systems; therefore, ignition performances can be varied significantly.
Flame propagation velocity is one of the most important ignition performance parameters. Suda et al. [3] have studied flame propagation velocities under air and oxy-fuel combustion conditions. Lean flammability limit is also a very important ignition performance. However, the lean flammability limits of pulverized coals for oxy-fuel combustion have not been reported.
We developed a laser ignition experiment technique [4] to study both flame propagation velocity and lean flammability limit in N2/O2 atmospheres [4, 5]. In the present investigation, we extended the model to oxy-fuel combustion. We analyzed the experimental results of Suda et al. [3] and our own results, and developed a model for estimating lean flammability limit and flame propagation velocity for oxy-fuel combustion. We also applied the model to develop an engineering design for actual and/or pilot-scale burner systems for oxy-fuel combustion.
2. Experimental equipment and the basic model
2.1 Ignition and flame propagation phenomena for pulverized coal combustion
Fig.1 shows a model for flame stabilization of pulverized coal burners [6]. Generally, pulverized coal and carrier gas (primary air) are supplied from the center of the burner. The secondary air is supplied around the primary air. A recirculation region of the high temperature burning gas is formed between the flows ofthe primary air and secondary air by the effect of the flame stabilizer. The burning gas of the recirculation region heats the pulverized coal particles in the primary air and these particles ignite and burn. When the coal concentration is high, the burning coal particles heat and ignite other nearby coal particles. The flame moves from the coal particles directly heated by the gas of the recirculation region to the surrounding coal particles. This phenomenon is considered flame propagation. The coal/oxidizer (such as air) mixture should be under flammable condition to stabilize the flame.
Fig.1 Ignition and flame propagation phenomena of pulverized coal combustion.
2.2 Laser ignition experiments
Fig.2(a) shows a schematic of the experimental equipment [4, 5]. Uniformly sized pulverized coal particles were suspended in a laminar upward flow and rapidly heated by a single-pulsed YAG laser. Velocity of the upward flow was controlled according to the particle diameter. The heated pulverized coal particles were burned in the quartz test section (50mm square cross section). The particle concentration was measured from the intensity of particle scattering by a He-Ne sheet laser.
Another continuous laser was used to determine the effect of radiant heat loss on the ignition characteristics. Its beam diameter was usually 15mm around the ignition point.
Photos of the burning particles are shown in Fig.2(b). Two kinds of phenomena were observed. The mixture of Fig.2(b-1) was under the non-flammable condition. Immediately after the pulse laser shot, light emission was observed from a small number of particles. These were burning particles that were directly ignited by the laser shot. The mixture of Fig.2(b-2) was under the flammable condition. A small number of particles ignited at first in the same way as in Fig.2(b-1). However, the number of burning particles then increased, as revealed by the laser area of light emission. That is, the flame moved from the particles directly heated by the laser to the surrounding particles. This phenomenon was considered flame propagation. Flame propagation velocity was defined as the growing rate ofthe flame radius [4].
(a) Laser ignition equipm ent (b) Flam e pictures
✓Pulsed particle feeder (H) non-flammable condition
(c) Basic Phenom ena
Single pulsed YAG laser
(for ignition) &
Continuous laser
(adjustm ent of the heat loss)
3m s after laser shot
\He-Ne sheet aser (coal concentration^ measurem ent )
Flow conditioning section Combustion supporting gas (oxidizer)
(b-2 ) flam mable condition
-^^m i on s
• * ' I
Phue propogst&B
Distance betw een particles; d Tim e of flam e propagation; s
Model; lelatLm between d end s
Obtained from expert] ents
Fig.2 Laser ignition experiments.
Fig.2(c) summarizes the basic phenomena of flame propagation. One side of two particles burns first, then, the other particle is ignited by the heat of combustion of the one burning particle. When the first particle ignites, volatile matter is pyrolized. A volatile matter flame is formed around the first particle. The flame grows due to volatilization, and the flame heats the next particle which has not ignited yet. Flame propagation is observed if the first burning particle can transfer the flame to the next particle before the volatile matter combustion of the first particle has finished. We defined the distance between particles as d and the time of flame propagation as s. Flame propagation velocity, Sb, was defined as the value of d divided by s. In this study, we analyzed experimental data to obtain the relationships between d and s under various experimental conditions.
3. Results and Discussion
3.1 Lean flammability limit and flame propagation velocity
= d/s)
Coal;bitum inous, anthracite, petroleum coke Diameter: 22^m , Oxygen: 21-100% (N2/O2) Heat flux from w al: room tem p - 3x104 W/m 2
8 « <D &
: (a) □ s
hv bitum incus (O2; 61%, N2;39%) hv bitum inous (O2; 21%, N2; 79%)
'lv bitum inous (O2: 100°%), flam e w as heated by the continuous laser
V bitum inous (O2: 100%)
-anthracite (O2; 100%) "]v bitum inous (O2; 100%,), large particle
Norm afed particle rommtratirn (-) Maxin um flam e propagation velocity: Sb-max (arb. unit)
Fig.3 Correlation between lean flammability limit and flame propagation velocity and experimental analyses based on that.
Examples of relationships between coal concentration and flame propagation velocity and between coal concentration and flame propagation probability are shown in Fig.3. Coal concentrations and flame propagation velocities are shown as normalized value. The coal concentration is in inverse proportion to 3 power of the distance:
d. Fig.3(a) shows normalized flame propagation velocity. When coal concentration increased, flame propagation velocity increased. But there was an upper limit value to the flame propagation velocity. The flame propagation probability is shown in Fig.3(b). The flame propagation probability was calculated as the ratio of the number of experiments in which flame propagation was observed and the number of experiments in which ignition was observed. [4]. We defined the lean flammability limit as the coal concentration when the flame propagation probability was zero.
The results of Figs.3(a) and (b) meant that lean flammability limit correlated with flame propagation velocity. Flame propagation velocities were almost zero at the lean flammability limit concentration. Fig.3(c) shows analyses of experimental results obtained by the relationships shown in Figs.3(a) and (b). Lean flammability limit L was inversely proportion to the maximum flame propagation velocity, Sb-max, obtained by varying coal concentration.
3.2 Effects of experimental conditions on flame propagation performances
Fig.4(a) shows the difference of the flame propagation phenomena with coal with different volatile content. For high volatile content coal, the growth rate of the volatile flame was large, because pyrolysis rate was large. The fl ame could be transmitted in a short time from one burning particles to another one, even though the distance between the particles was large. Flame propagation velocities became large for high volatile content coals. The time of flame propagation s became long. Fig.4(b) shows the relationship between pyrolysis rates of coals and flame propagation velocities. The pyrolysis rates were calculated results for the same heating rate condition as that ofactual boilers (20000K/s). Each pyrolysis rate constant was obtained by laboratory-scale experiments [5]. Flame propagation velocities strongly depended on pyrolysis rates. Fig.4(c) relates coal properties and pyrolysis rates. The influence of the coal properties on flame propagation velocity and lean flammability limit could be analyzed by getting the database which expressed relationships between coal properties, pyrolysis rates and flame propagation velocities for various kinds of coal and other solid fuels.
(a) Model
high volatile coal
CD or y P
Bitum inous ^
fib0^ Wood
^^ Lignite
t"l Anthracite
ignition
Pyrolysis rate (arb. un it) Fig.4 Effects of coal properties.
0 50 100
Volatile m atter (wt%, dry ash-free)
The flame propagation velocities in the N2/O2 and CO2/O2 atmospheres were experimentally obtained by Suda et al [3]. Their results are shown in Fig.5(a). The flame propagation velocities strongly depended on composition of combustion supporting gas (oxidizer). The flame propagation velocity increased significantly with O2 concentration. The flame propagation velocities in the CO2/O2 atmosphere were smaller than those in the N2/O2 atmosphere when oxygen concentrations were the same. Fig.5(b) shows temperatures of flames formed around the particles; Tvf shows the flame temperature, and Tvf* is the standard value. The experimental results [4] were measured with two color pyrometers. The calculated flame temperature was defined as adiabatic flame temperature when volatile matter was burnt under the stoichiometric condition. The flame temperature also strongly depended on composition of combustion supporting gas. The difference of this flame temperature produced the difference of the flame propagation velocity. Fig.5(c) shows the relationship between the flame temperatures and the flame propagation velocities. Both sets of results in the N2/O2 and CO2/O2 atmospheres are shown. The flame propagation velocity was strongly influenced by the flame temperature. When the flame temperature was fixed, the flame propagation velocity had hardly any influence from the gas composition.
We developed a model to predict both flame propagation velocity and lean flammability limit, based on these results. The model can analyze effects of coal properties, coal particle diameter, coal concentration, oxygen
concentration, radiant heat loss from flame to surroundings, and composition of combustion supporting gas. Relationships between coal concentrations and flame propagation velocities were calculated under various compositions of combustion supporting gas. The results are compared in Figs.5(d) and (e) with the experiment results [3,4]. Fig.5(d) shows the effects of oxygen concentration in the N2/O2 atmosphere . Fig.5(e) compares N2/O2 and CO2/O2 combustion when the oxygen concentration was 40%. The calculated results agreed with the experimental ones.
Coalhv bituminous, Diameter: 58 ^m for (a), (c) and (e), 2 2^ m for (b) and (d) Symbols; exp. Lines; calc. Experimental data of (a), (c) and (e): Suda et al [3], (b) and (d): Taniguchi et al [4]
0.8 0.4 0
. (d) ¿J--A- -O2 100%
' a/A \7
N2/O2 : O2 61%
0 0.5 1
Coal concentration (arb. unit)
Oxy-fue
O2 (vol%)
-500 0 500 Tvf-Tvf* ( (K) Fig.5 Effects of composition of combustion supporting gas.
0 0.5 1.0
Coal concentration (arb. unit)
(a) Laser ignition (b) P ibt scale
coal feed line W ater w all Caster w all Experi ents : K iyam a et al [7] (walltemp. (walltemp. (walltemp. 300-400K) 600-900K) 1000-1300K)
.Caster
Oxy-fuel (02 21%) hv-bitum inous
Flammable
Oxy-fuell
(O2 20%)
Lignite^
A ir, hv bitum inous
— Q,.
Pure oxygen,
lv-bitum inous non-Flammabh
Pilot scale „ W ater w a ll
burner
(c) Pilot ^ A ctual scale Experin ents: Kjga et al [8] Pilot scale A ctual
(coal feed rate = 3t/h) boiler
Operating W Caster burner wa
C u Fine coal
.rb a( 2 A Stable
£ ih coarse
it! 1 t.
CQ Unstable In term ediate
l Stable
Unstable
0 10 20 30 Volatile m atter content (w t% as fired)
0 0.2 0.4 0.6 0.8 Norm a lized heat loss rate -Standard value (-)
Heat flux from w aH $ /m2)
Fig.6 Effects of heat loss rate from the flame to the furnace wall. Symbols, experimental; lines, calculated.
For engineering design of actual boilers, it is important to consider the effect of heat loss from the flame to the furnace wall. Lean flammability limit and flame propagation velocity were varied with the difference of the heat loss rate by the furnace conditions, such as whether the furnace was covered with a water wall or caster. The basic expression of the radiant heat loss was obtained from the laser ignition experiments [6]:
1/L =a Qwall (1L)
where, L is lean flammability limit, a is a constant, Qwall is radiant heat flux from a wall or from the continuous laser, and L0 is standard lean flammability limit for standard heat flux condition (usually, wall temperature = room temperature). Example effects of heat flux from a wall on lean flammability limit are shown in Fig.6(a). Experimental data of Fig.6(a) were obtained by the laser ignition experiment. When the wall temperature was low, the lean flammability limit concentration was high. As wall temperature became higher, lean flammability limit decreased, and ignition became easy. The difference between the air and oxy-fuel combustion is also shown in the figure. Lean flammability limit concentration became higher for oxy-fuel combustion, when oxygen concentration was the same.
The model was verified by experiments in pilot- and actual-scale furnaces (Figs.6(b) and (c)). The effects of coal properties on lean blow-off limit concentration are shown in Fig.6(b). The lean blow-off limits were obtained experimentally for six kinds of coals by using the same burner and furnace conditions [7]. The effects of heat loss are shown in Fig.6.(c) The lean blow-off limit was obtained experimentally by varying the area ratio of the caster and water wall that covered the furnace wall [8]. We assumed that the flame became unstable when Sb was lower than the constant value that was decided for each burner structure. We defined the constant value as minimum-Sb to form a stable flame by the burners. The minimum-Sb value was around 0.05m/s for the present experimental conditions. The lines in Figs.6(b) and (c) mean the coal concentrations when the calculated flame propagation velocities were equal to the minimum-Sb. Calculated results of Figs.6(b) and (c) agreed with the experimental ones. A more detailed explanation is offered in reference [6].
3.3. Application of the model: proposal of flammability analysis
We applied the model to develop the engineering design of actual and/or pilot-scale burner systems. We call our proposed technique flammability analysis. This technique is a post-processing analysis of CFD calculations. A general CFD method, such as the k-e method can be used for this analysis. Calculated results of coal concentration, gas composition and temperature profiles were read at first. Next, experimental conditions, such as coal properties, particle diameter distribution and furnace wall temperature, were read. Flame propagation velocity profiles were calculated by using the information. Flame stability was judged by the calculated flame propagation velocities and past results of actual- and pilot-scale experimental results. If the flame propagation velocity Sb was larger than the minimum-Sb, the condition of the mixture of coal and oxidizer was judged as flammable. Flame propagation velocity near the recirculation region was very important for considering the flame stability.
Flam e picture
CASE I A r com bustion
Detailed C FD com bustion calculation (LES)
few weeks
General C FD com bustion calculation (k-£ )
CASE UE a DS^T-burner designed by Hitachi Power Europe, installed at Schwarze Pumpe pilot plant.
Flam m ab iity analysis few hours
Flam inability; r Good
Difficult /V,
Good J \
positron
Fig.7 Typical results obtained by using the flammability analysis.
Typical application examples are shown in Fig.7. Fig.7 shows the following, from the left:
(1) a flame picture which was provided by the experiment;
(2) calculated temperature profiles obtained by using the LES (Large Eddy Simulation) method;
(3) calculated temperature profiles obtained by using the k-e method; and
(4) results of flammability analysis
Calculations by k-e and LES methods were described in the literature [9, 10].
CASE I is the results of air combustion. The flame picture showed that flame stability was good. The results by the LES method also showed that a stable flame could be obtained. LES was a good analytical technique to evaluate the stability of pulverized coal flames [9]; however, the calculation load was large. CASE II was an example of oxy-fuel combustion. According to the analysis results by the LES method, stability of the flame was inferior to that of CASE I. The difference between CASES I and II was clear. Revision of burner structure would be needed for CASE II. Calculated results of the k-e method were also compared. The difference between CASES I and II was not clear. The calculation load of k-e was small; however, it was hard to evaluate the flame stability. The results of flammability analysis are also shown in the figure. A clear difference was obtained for the calculated flame propagation velocities of CASES I and II. Flame propagation velocity of CASE II was inferior to that of CASE I. This conclusion accorded with that of the LES method. The flammability analysis also used the k-e method, but evaluation of the flame stability was enabled.
CASE III is another example of oxy-fuel combustion. The burner structure was revised based on the results of the flammability analysis. Clear difference was observed in flame propagation velocity of CASES II and III. Flame stability of CASE III was improved. CASE III was a DS®T-burner designed by Hitachi Power Europe, installed at Schwarze Pumpe pilot plant.
3.4 More case studies
(a) Burner structure
Recirculatim region
Flam e stabilizer (FSR)
Exhaust gas + O
(b) Initial design
(c) A fter revision
â» 20
\ ^FSR
NH:]:!^. mhinum -Sb ..^k.......
Recirculation region 1
enlarged
Exhaust gas + coal
Exhaust gas
1 2 Norm alized r (-)
by controlling local oxygen% distribution
m iim um -Sb
Exhaust gas + O; Exhaust gas + coal
Exhaust gas + O
Normalized r (-)
Exhaust gas + O 2 Exhaust gas + coal
Fig.8 Examples of CASE studies.
Figs.8 show examples of CASE studies that applied this technique to a burner design for lignite-fired oxy-fuel combustion. Fig.8(a) shows examples of the burner design. Pulverized coal was supplied by combustion flue gas, and injected from a burner exit. Usually, oxygen concentration of the flue gas was less than 10 vol%. Oxygen for combustion was mixed with the flue gas, and supplied by roughly two systems. Some gas was supplied from the central part of the burner. The remaining gas (secondary gas) was supplied from the circumference side of the coal flow. In many cases, the secondary gas was injected as swirl flow. A flame stabilizer was arranged between the coal flow and the secondary gas flow. Recirculation regions were formed downstream from the flame stabilizer. Coal concentration, gas composition and temperature profiles were obtained by the CFD calculation (k-e method).
Flame propagation velocity profile was obtained from this information in the burner neighborhood shown as the dotted line. The position and size of the recirculation region were judged from calculated flow velocity profiles. Flame stability became good when flame propagation velocity near the recirculation region increased.
The flame propagation velocity profile obtained with the initial design is shown in Fig.8(b). Flame propagation velocity was lower than the minimum-Sb value in most of the area. It was hard to obtain a stable flame. Coal and oxygen concentration profiles are also shown in Fig.8(b). Oxygen concentration was low where coal concentration was high and vice versa. In this system, coal was supplied by combustion flue gas. It was easy to lower the oxygen concentration where many coal particles were flowing. Coal concentration should be increased where oxygen concentration was high.
Fig.8(c) shows flame propagation velocity and coal concentration profiles obtained after revision (detailed results of CASE III in Fig.7). A part of the coal particle flow was separated from the flue gas flow (lower oxygen concentration). These coal particles could flow to the recirculation region (higher oxygen concentration) by modification of burner structure. The recirculation region was enlarged by modification of secondary flow. Flame propagation velocity became considerably larger than the minimum-Sb value because the coal concentration increased where the oxygen concentration was large. By using the flammability analysis, we could quickly clarify improvement points of the burner.
4. Conclusion
We developed a model to predict lean flammability limit L and flame propagation velocity Sb for oxy-fuel combustion conditions based on fundamental experimental data. The proposed model was verified with data of both fundamental and pilot- and actual-scale experiments. The model could predict both flame propagation velocities and lean flammability limits for N2/O2 and CO2/O2 combustion systems.
We also applied the model to develop burner systems for lignite-fired oxy-fuel combustion. Local Sb and L near the ignition points of the burner systems could be obtained from concentration and temperature profiles of the general CFD results (k-e method). Flame stability was judged by the calculated Sb and L profiles, and past results of blow-off limits obtained from actual- and pilot-scale experiments. We proposed this technique as flammability analysis. By using the technique, we could quickly clarify improvement points of the burner systems. The calculated results were applied to a DS®r -burner designed by Hitachi Power Europe, installed at Schwarze Pumpe pilot plant.
References
1) Tigges KD, Klauke F, Bergins, Busekrus CK, Niesbach, Ehmann JM, et al. Conversion of existing coal-fired power plants to oxyfuel combustion: case study with experimental results and CFD-simulations. Energy Procedia; 2009; 1: 549-56.
2) Stromberg L, Lindgren G, Jacoby J, Giering R, Anheden M, Burchhardt U, et al. Update on Vattenfall's 30MWth oxyfuel pilot plant in Schwarze Pumpe. Energy Procedia; 2009; 1: 581-9.
3) Suda T, Masuko K, Sato J, Yamamoto A, Okazaki K. Effect of Carbon dioxide on flame propagation of pulverized coal clouds in CO2/O2 combustion. Fuel; 2007; 86: 2008-15.
4) Taniguchi M, Kobayashi H, Azuhata S. Laser ignition and flame propagation of pulverized coal dust clouds. In: 26th Symp. (Int.) on Combustion. Pittsburgh: The Combustion Institute; 1996, pp. 3189-95.
5) Taniguchi M, Kobayashi H, Kiyama K, Shimogori Y. Comparison of flame propagation properties of petroleum coke and coals of different rank. Fuel; 2009; 88: 1478-84.
6) Taniguchi M, Yamamoto K. Fundamental research on oxy-fuel combustion: The NOx and coal ignition reactions. In: Grace CT, editor. Coal Combustion Research, New York: Nova Science Publishers Inc; 2010, in press.
7) Kiyama K, Tsumura T, Sei N, Taniguchi M, Nomura S. Combustion technology for high fuel ratio coal. Proceedings of International Conference on Power Engineering-97 2, Tokyo: 1997, pp. 477-81.
8) Kiga T, Makino K, Okushima K, Abe T, Hirano T, Takagi T. Development of IHI wide range pulverized coal burner. Ishikawajima Harima Giho; 1987; 27: 333-8.
9) Yamamoto K, Taniguchi M, Kobayashi H, Sakata T, Kudo K. Validation of coal combustion model by using experimental data of utility boilers. JSME International Journal Series B; 2005; 48: 571-8.
10) Yamamoto K, Murota T, Okazaki T, Taniguchi M. Large eddy simulation of a pulverized coal jet flame ignited by a preheated gas flow. In: 33rd Symp. (Int.) on Combustion. Pittsburgh: The Combustion Institute; 2010, in press.
