- Email: [email protected]
Experimental study of micro-scale premixed flame in quartz channels
Available online at www.sciencedirect.com
Proceedings of the
Combustion Institute
Proceedings of the Combustion Institute 32 (2009) 3083–3090
www.elsevier.com/locate/proci
Experimental study of micro-scale premixed flame in quartz channels Yong Fan *, Yuji Suzuki, Nobuhide Kasagi Department of Mechanical Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
Abstract This paper presents (1) the development of observation and optical measurement system for the investigation of flame in ultra-thin quartz channels; and (2) the observation and analysis of premixed CH4/air flame propagation and quenching in three quartz combustors with chamber depth of 0.7 mm, 1.0 mm and 1.5 mm, respectively. Flame oscillation is interpreted and discussed according to well-resolved chemiluminescence images of the flame taken by the high-speed ICCD camera. Flammability limits are obtained for different equivalence ratios and mixture velocities. It is found that the quenching distance for heated quartz wall can be smaller than 0.7 mm, and is dependant on the wall temperature. Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Gas-phase combustion; Ultra-thin combustor; Quenching; Flame oscillation; Wall temperature
1. Introduction Enjoying work and life with mobility has become the style of the modern society, and the ever-increasing power consumption of portable electric devices leads to an urgent demand of new portable power source to meet the requirement. The interaction between MEMS technology and knowledge of conventional energy generating/harvesting principles has recently evolved into an area known as Power MEMS. Various power generation concepts such as the micro gas engines and micro fuel cells have been proposed [1]. Hydrocarbon-fueled micro combustor is of great interest in portable power generation applications for higher energy density [2]. Homogeneous combustion is beneficial for high temperature applica-
*
Corresponding author. Fax: +81 3 5800 6999. E-mail address: [email protected] (Y. Fan).
tions such as the TPV system if compared with catalytic combustion [3]. However, elevated wall-effects, associated with the increase of surface-to-volume ratio, make the gas-phase combustion problematic in microscale. It has been widely accepted that flame generally quenches in chambers with small characteristic length. Quenching distance, which is defined as the minimum gap between parallel plates for flame propagation, is in the order of a few millimeters for typical hydrocarbon fuels. Two fundamental quenching mechanisms were known related to the quenching phenomena. One is thermal quenching, for which flame can be thermally deactivated by the heat loss from the flame zone to surrounding walls. The other is chemical quenching, which refers to the absorption of intermediate radicals by adjacent wall surfaces. The thermal/chemical quenching mechanisms on the flame has long been an issue in accurate prediction of flammability and flame stability.
1540-7489/$ - see front matter Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2008.06.219
3084
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
Maruta et al. [4–6] investigated the effect of external heat-input and heat recirculation on flame propagation in straight and U-shaped 2mm-diameter tubes, where the wall temperature is kept at 1120 °C. They also observed flame oscillation (termed as FREI in [4]), and pointed out that it is a periodic process of flame ignition in the hot part of the tube and extinction upstream. A 1-D nonlinear evolutionary equation of the flame front has been applied to interpret the transition from steady condition to flame oscillation [6]. A positive wall temperature gradient in the streamwise direction has been adopted in the modeling in order to represent similar wall thermal condition in the experiments, and they report that the onset of the flame oscillation is due to the inertial effect associated with the heat losses. They also conjectured that the instability might be related to the flame oscillation phenomena in a variable-section channel [7]. A different mathematical model has been proposed by Jackson et al. [8], and they report that flame oscillation can arise when Lewis number is over unit, or when heat losses present. An earlier study on flame oscillation has been conducted in a blown porous media (filtration gas combustion [FGC]) by Fateev et al. [9], which shows much in common with the flame oscillation in a narrow tube. Early investigation of the chemical quenching was made by Sloane et al. [10,11]. They measured molecular fraction of CH4, CO, H2, CO2, OH and H at low pressure for catalytic (Platinum) and non-catalytic (Gold) wall surfaces using a molecular beam mass spectrometer. They demonstrated that the surface recombination with the wall do affect the flame quenching within the region of 1 mm from the surface. However, the thermal effect is dominant. Recently, Miesse et al. [12] and Kim et al. [13] separately made experiments of quenching distance measurement in parallel plates with different wall materials/treatment. Miesse et al. [12] found inert materials such as quartz, alumina and cordierite, are effective against chemical quenching, and claim that chemical quenching instead of thermal quenching dominates for high wall temperature. Kim et al. [13] instead found even metallic material like stainless steel can have smaller quenching distance at high wall temperature after being treated into inert surface by annealing. They claim that for wall temperature ranges from 400 to 600 °C, the flame quenching becomes dependent on the radical removals at the surface, and quenching distance increases; but for wall temperature over 600 °C, the homogeneous chemical reactions overcome the wall-effect of radical removals, and quenching distance becomes smaller with the temperature increase. Previous studies of microscale flammability and stability in some extents reveal the roles of wall thermal and chemical effects. However,
detailed quenching mechanisms in micro-scale, i.e. how the thermal effect couples with the chemical effect, still remain unclear due to insufficient experimental evidence. Of great importance in due investigations are the accurate control and measurement of flame temperature and wall temperature which are related to the thermal effect, the quantitative measurement of near-wall radical concentration which is related to the chemical effect, and the flame visualization which may explain the flame/wall coupling process. The objective of the present study is to provide experimental observations and measurements of flame propagation and quenching in ultra-thin quartz tubes for the investigation of microscale combustion phenomena.
2. Micro combustor and experimental setup Figure 1 shows the quartz combustors developed in the present study. Upstream developing channel, where no flame is expected, has a height of 0.5 mm and a length of 22 mm. The downstream combustion channel has a height of 0.7 mm, 1.0 mm and 1.5 mm, which is smaller than conventional quenching distance for hydrocarbons. The channel height of 0.7 mm is even smaller than recently measured quenching distance for quartz in [12,13]. The combustion channel is 25 mm in length and 16 mm in width. The inner-wall of the combustion channel is made of polished synthetic quartz, so that optical measurements are possible. An inert material with small wall roughness like quartz is also supposed to suppress the radical adsorption and trapping at the surface. The top and bottom sides of the channel are fusion-bonded with black quartz (Nb-doped quartz) plates for absorption of IR light as heating source. The main experimental setup is shown in Fig. 2. The combustor is fixed on to an XY-h stage for fine adjustment of its position. The downstream combustion channel is heated with IR lamp heater from both sides. A quartz rod with a diameter of 20 mm is used for the introduction of IR light, so that condensed IR light can be delivered evenly to most portion of the combustion channel. The center of the rod is aligned to 10 mm upstream the exit in the centerline of the combustor. Two R-type thermocouples (/=0.5 mm) are plugged in 1-mm-diameter holes opened in the transparent quartz layer of the top and bottom walls. Wall temperature measured with the thermo couple TTC is regulated by adjusting the lamp power. CH4 fuel is supplied from a gas cylinder, and air is introduced from a compressor. The CH4 and air flow rates are separately measured and regulated by two sets of flow meters and valves.
Black 2 x φ = 1 mm Thermocouple Holes Quartz
3 mm
φ = 2 mm
3085
m m
15 mm 25 mm
8
Quartz
0.5 mm
2 mm
a
d = 0.7 mm 1.0 mm 1.5 mm
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
b
Fig. 1. Micro quartz combustor. (a) Schematic, (b) Photo. Combustion channel is 16 mm wide and 25 mm long. Combustion channel height, d = 0.7, 1.0, and 1.5 mm.
Fig. 2. Experimental setup. (a) Schematic, (b) Photo, (c) Magnified view of the combustor.
A high-speed ICCD camera (FASTCAM, Photron) is placed in the sideway, and CH*/OH* chemiluminescence images of the flame are taken for flame visualization up to 2000 frames per second. Optical band-path filter at 307 ± 10 nm is employed for OH* chemiluminescence, and 410 ± 10 nm for CH* chemiluminescence. UV laser beam from the dye laser (SCANmate, Lamda Physik) is condensed to thin laser sheet with a
thickness of 0.5 mm, and introduced from the bottom of the combustor for excitation of the OH* and CH* radicals. By utilizing the LIF technique, the present setup is capable of imaging the flame front and also flame temperature with the OH 2line method [14]. Prior to the flame observation, a radiation thermometer (IN140/5-L, IMPAC) is placed in the wall-normal direction for measurement of
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
inner wall surface temperature of the combustor as shown in Fig. 3. Four 2 mm-diameter circular windows are opened perpendicular to the top wall for passing sensing light from the inner-wall surface. The radiation thermometer has sensing wavelength of 5.14 lm, which gives specific emissivity of 0.97 for quartz. The sensing spot of the thermometer is 0.9 mm in diameter for fine spatial resolution.
Thermocouple:
,
, ,
, ,
, ,
,
, ,
; .
1000 800 600 400
3. Experimental results Figure 4 shows the inner-wall surface temperature for the wall temperature measured with the thermocouple TTC=500, 600, 700, 800, 900 and 1000 °C. The temperature profiles vary in the streamwise direction in the range of about 200 °C, and have a peak at the center of the heating quartz rod. The thermocouple temperature TTC and the inner surface temperature measured with the radiation thermometer are within 15 °C, since the thermocouple hole is only 1 mm away from the inner surface of the combustion channel. Although the wall temperature is not uniform in the streamwise direction, it is still reasonable that the thermocouple temperature is used to represent the wall temperature in the following discussions. It is also found in the preliminary experiments that, when a flame exists inside the channel, local temperature near the flame increases by about 100 °C.
Thermocouple PC
1200 Inner-wall:
Temperature (ºC)
3086
Controller
Combustor
0 5 10 15 20 25 Distance from Combustion Channel Inlet (mm)
Fig. 4. Temperature measured at the inner-wall surface.
When the wall temperature is raised beyond a critical value, flame propagates into the combustion channel. Both CH* and OH* chemiluminescence images of the flame are obtained, and the OH* images are shown in this report. Two flame patterns, steady flame (Fig. 5) and oscillating flame (Fig. 6), have been observed in all three combustion channels. Figure 7 shows the evolution of flame patterns with increasing air flow rate. For a fixed fuel flow rate, increasing air flow rate leads to the increase of mixture velocity and the decrease of equivalence ratio. When the air flow rate is small, flame is positioned at the exit, where there is enough air. With a larger air flow rate, the flame begins to propagate into the combustion channel. The flame cell keeps moving upstream from the exit periodically, and becomes oscillating. The oscillation amplitude and flame cell length gradually increases with
Lens Rod IR Lamp Heater
Radiation Thermometer XYZ Stage
2 φ1
Unit: mm
15
25 5 5 5
4 φ2 Fig. 3. Experimental setup for measurement of innerwall surface temperature.
Fig. 5. OH* chemiluminescence of steady flame in the 1.5-mm-thick combustion channel.
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
3087
t=0 s 0.001s 0.002s 0.003s 0.004s 0.005s 0.006s 0.007s 0.008s 0.009s 0.010s 0.011s 0.012s
Ignition
Propagation
0.7 mm
Recharge
*
Fig. 6. OH chemiluminescence of oscillating flame in the 0.7-mm-thick combustion channel.
increasing air flow rate, and then the flame quenches with a slightly larger air flow rate. The mechanisms of this oscillating flame will be discussed below. With an even larger air flow rate, the flame propagates into the combustion channel again, and keeps the steady position. The steady flame holds for a wide air flow rate range, and the position of steady flame moves closer to the exit with increasing the air flow rate as shown in Fig. 8. This is due to the streamwise temperature variation (Fig. 4) since laminar flame velocity can be greatly affected by the temperature. Further increasing the air flow rate, the flame is then blew out of the combustor and positioned at the exit. Finally, it is blew-off with extremely large air flow rates. Note that the quenching conditions between oscillating flame and steady flame is very narrow, and vanishes for large fuel flow rate. Figure 9 shows the schematic of the formation of steady and oscillating flames. In the steady flame, unburned gas mixture is preheated by the hot wall, which enables flame being initialized in the thin channels. On the other hand, the wall temperature is lower than the flame temperature, and thus heat is always lost from the flame to the adjacent wall. However, the flame-wall thermal-coupling should be treated somewhat differently in the analysis of steady and oscillating flames. For the steady flame, the energy conservation principle can be applied; there should be considerable local wall temperature raise due to heat loss from the flame to the wall. Thus, the wall temperature becomes less uniform. But for the oscillating flame, energy conservation is not hold in the same way because time scale of the flame passage is faster than thermal response of the combustor wall.
One whole flame oscillation process that is clearly tracked by time-resolved OH* images (1 ms interval time) is shown in Fig. 6. Flame ignites at the combustor exit at first. This gives the flame an initial heat input. When the exit flame cell gathers certain amount of heat, it grows, expands, and propagates into the combustor. However, when flame is traveling upstream in the channel, the flame continuously losses heat to the wall. The flame quenches upstream where it can not pace further. The fresh unburnt mixture then recharges to the exit for another oscillating cycle. According to the observation, one may divide one oscillation cycle into three main stages, i.e. the ignition (initialization) stage, the propagation stage, and the recharge stage. Then the oscillation frequency can be expressed as 1 ¼ ignition time frequency amplitude þ propagating velocity amplitude þ : mixture velocity
ð1Þ
Base on Eq. (1), the frequency should be lower for larger amplitude. Figure 10 shows the oscillation frequency measured for 0.7 mm-thick and 1.0 mm-thick channel combustors. Frequency in most conditions is in the range of 30–500 Hz, and in each curve decreases with increasing mixture velocity. The frequency shall decrease because the increase of mixture velocity leads to the decrease of equivalence ratio. Note that the oscillating flame is in the fuel-rich zone, flame should be able to penetrate deeper in the combustion channel for smaller equivalence ratios, which
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090 No Flame Flame outside
Oscillating flame
No Flame
25
Distance from Inlet (mm)
3088
20 15 10 5 0 400
(2)
(1)
(3)
Steady flame
(7)
(6)
(4) Flame outside
(8)
(9)
Equivalence Ratio 2
1
(5) No Flame
Flame Position Flame Length
d=0.7 mm combustor CH4: 46 sccm, TTC=800ºC 500 600 Air Flow Rate (sccm)
700
Fig. 8. Variation of position and length of steady flame with air flow rate.
(10)
Mixture Velocity (cm/s) 200
100
0 0 300 500 700 900 1100 Air Flow Rate (sccm) (d = 0.7 mm, fuel: 55sccm) Increase Air Flow Rate
Fig. 7. Evolution of flame patterns with increasing air flow rate. (1) Flame quenching at small air flow rates; (2) Flame at the exit for small air flow rates; (3) Oscillating flame with a small amplitude; (4) Oscillating flame with a large amplitude; (5) Flame quenching at moderate air flow rates; (6), (7) and (8) Steady flame at different positions with the change of air flow rate; (9) Flame at the exit for large air flow rates; (10) Flame quenching at large air flow rates.
actually coincides with the experimental observations that the amplitude becomes larger (not shown). It is also shown in Fig. 10 that the flame oscillation happens at larger mixture velocity for thinner channel combustor, and at slightly larger mixture velocity for lower wall temperature. Figure 11 shows the flammability limits in the 0.7-mm-thick channel for different wall temperatures. Compared with the steady flame, oscillating flame is in the higher equivalence ratio region. In this fuel-rich region, the CH4/air mixture can not be ignited inside the combustion channel, so that the flame is ignited first at the exit where fuel meets more air, and then propagates into the
channel. The flammability limits region becomes broader with increasing wall temperature. Compared with the data in a 2-mm-diameter tube combustor [4–6], the present combustor should have stronger wall-effect due to the higher surface-tovolume ratio and narrower channel. However, the flame in the present combustor can be stabilized at much higher velocities. It is conjectured that the recirculation zone downstream the backward-facing step in the combustion channel stabilizes the flame. It is also noted that the range of the equivalence ratio within the flammability limit is narrower than the data in the long tube combustor [15]. Maruta et al. [4–6] explained in their 1-D model that the oscillating flame is actually the transition from the normal steady flame and another ‘weak’ steady flame at extremely small mixture velocity. However, no ‘weak’ flame has been found in the present experiment. The flame oscillation occurs only in the fuel-rich conditions, which differs from the observations of flame oscillation between the normal flame and ‘weak’ flame conditions in [4–6]. Note that the ignition position for the oscillating flame is at the exit in the present study, which is also different from the ignition at a hot part of the tube in [4–6]. These can be the result of a shorter combustion channel and the bell-shape streamwise wall temperature distribution in the present study. Note that a very narrow quenching zone exists between the normal steady flame and oscillating flame zones (Fig. 11b). Since the quenching at large flow rates is obviously due to the blow-off, it is conjectured that this narrow zone with small flow rates is corresponding to the flashback phenomena. This also explains the reason why the oscillating flame propagates upstream after the ignition at the exit. Figure 12 shows the flammability limits in the 1.0-mm-thick channel. Compared with the 0.7mm-thick channel, the 1.0-mm-thick channel has a broader flammability limits region for the wall temperature of 800 °C, but the flammability
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
b
a
Initial heat input by ignition at exit
Wall
Wall
(Heat loss) Q
Q
Flame Q (Heat loss)
Q
Flame Q (Pre-heat)
Q
3089
Q (Pre-heat)
Q
Fig. 9. Schematic of the formation of (a) steady flame, and (b) oscillating flame.
Frequency (Hz)
400
73sccm 64sccm
300 200
73sccm
46sccm 27sccm 100 37sccm 0 20
55sccm 30
Steady Flame
d=0.7 mm, TTC=800ºC d=1.0 mm, TTC=800ºC d=1.0 mm, TTC=1000ºC
CH4: 55sccm 46sccm
40 50 60 Mixture Velocity (cm/s)
70
80
Fig. 10. Oscillation frequency measured with high-speed ICCD camera.
150 100
Mixture Velocity (cm/s)
500
Oscillatiing Flame
150 65 sccm 56 sccm
a TTC = 800ºC CH4: 74 sccm
100
46 sccm 37 sccm 28 sccm
50
Mixture Velocity (cm/s)
No Flame
0 200 CH4: 68 sccm 150 55 sccm 46 sccm 36 sccm 100 28 sccm
b TTC = 1000ºC
50 0 0.0
0.5
1.0
1.5
Equivalence Ratio Fig. 11. Flammability limits in the 0.7-mm-thick combustion channel. (a) TTC = 800 °C, (b) TTC = 1000 °C.
a TTC = 800ºC CH4: 92 sccm 83 sccm
50 0 200 CH4: 92 sccm
150
b TTC = 1000ºC
73 sccm 54 sccm
100
Steady Flame
73 sccm 64 sccm 46 sccm 37 sccm
Oscillatiing Flame
37 sccm
50 28 sccm
0 0.0
0.5 1.0 Equivalence Ratio
1.5
Fig. 12. Flammability limits in the 1.0-mm-thick channel. (a) TTC = 800 °C, (b) TTC = 1000 °C.
region is only slightly enlarged for the higher wall temperature of 1000 °C. This is probably because the wall-effect on quenching for the thinner combustion channels becomes more severe at low wall temperature. The maximum CH4 flow rate for steady flame is 92 sccm for the 1.0-mm-thick channel, while it is only 74 sccm for the 0.7-mmthick channel. This is due to the blow-off, because the maximum flow rates correspond to almost the same bulk gas velocities. As shown in Fig. 13, critical wall temperature for flame to exist becomes higher for thinner combustion channel, which means that more thermal input is necessary for thinner channels due to stronger wall effect. Miesse et al. [12] report that the quenching distance for high wall temperature (near 1000 °C) strongly depends on the surface
Critical Wall Temperature (ºC)
3090
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
region becomes narrower for thinner channel, and broader for higher wall temperature. (3) Quenching distance in the quartz channel is dependant on the wall temperature, and can be smaller than 0.7 mm for the wall temperature over 800 °C.
1200 1000 800 600 400
Acknowledgment 200 0 0.0
0.5 1.0 Channel Height (mm)
1.5
Fig. 13. Effect of the combustion channel height on the critical wall temperature.
material and chemical quenching prevails, whereas the thermal quenching plays a dominant role at lower wall temperature. However, in the present result at 800 °C, flame can still propagate in a 0.7-mm-thick quartz channel, which is smaller than the wall gap of 1 mm previously reported [12]. Since the quenching distance is still dependant on the wall temperature, one can expect thermal quenching mechanism play an important role even at high wall temperature. 4. Conclusions Micro quartz combustor that enables optical access has been developed for the investigation of quenching mechanisms in micro channels. Microscale steady and oscillating flames are confirmed through CH*/OH* chemiluminescence. Premixed CH4/air combustion is examined in the quartz combustion channels with the height of 0.7, 1.0 and 1.5 mm. Discussions are made on the thermal-coupling between the flame and wall in microscale flames. The following conclusions are derived: (1) Flame oscillation in ultra-thin channels can be interpreted as a periodic process of three stages, i.e. the initialization and expansion of flame cell at ignition stage, the flame propagation and quenching at the propagation stage, and the recharge of gaseous mixture to the exit at the recharge stage. (2) Flammability limits
The present work is supported by Grant-inAid for Scientific Research (B) (No. 19360094) by JSPS, Japan. YF was partially supported by the 21st Century COE Program, ‘‘Mechanical Systems Innovation,” by MEXT, Japan.
References [1] S.A. Jacobson, A.H. Epstein, An Informal Survey of Power MEMS, Paper ISMME2003-K18, Tsukuba, 2003. [2] A.C. Fernandez-Pello, Proc. Combust. Inst 29 (2002) 883–899. [3] T. Okamasa, G.-G. Lee, Y. Suzuki, N. Kasagi, S. Matsuda, J. Micromech. Microeng. 16 (2006) S198– S205. [4] K. Maruta, J.K. Parc, K.C. Oh, T. Fujimori, S.S. Minaev, R.V. Fursenko, Combust. Expl. Shock Waves 40 (5) (2004) 516–523. [5] K. Maruta, T. Kataoka, N. Kim, S. Minaev, R. Fursenko, Proc. Combust. Inst. 30 (2005) 2429– 2436. [6] S. Minaev, K. Maruta, R. Fursenko, Combust. Theor. Model. 11 (2) (2007) 187–203. [7] S.S. Minaev, V.S. Babkin, Combust. Expl. Shock Waves 37 (1) (2001) 13–20. [8] T.L. Jackson, J. Buckmaster, Z. Lu, D.C. Kyritsis, L. Massa, Proc. Combust. Inst. 31 (2007) 955–962. [9] G.A. Fateev, O.S. Rabinovich, M.A. Silenkov, Proc. Combust. Inst. 27 (1998) 3147–3153. [10] T.M. Sloane, J.W. Ratcliffe, Combust. Flame 47 (1982) 83–92. [11] T.M. Sloane, A.Y. Schoene, Combust. Flame 49 (1983) 109–122. [12] C.M. Miesse, R.I. Masel, C.D. Jensen, M.A. Shannon, M. Short, AIChE J. 50 (2004) 3205–3214. [13] K.T. Kim, D.H. Lee, S. Kwon, Combust. Flame 146 (2006) 19–28. [14] R. Cattolica, Appl. Opt. 20 (1981) 1156–1166. [15] G.F. Wang, B.J. Zhong, J. Tsinghua Univ. (Sci. Tech.) 46 (2006) 276–279.
Proceedings of the
Combustion Institute
Proceedings of the Combustion Institute 32 (2009) 3083–3090
www.elsevier.com/locate/proci
Experimental study of micro-scale premixed flame in quartz channels Yong Fan *, Yuji Suzuki, Nobuhide Kasagi Department of Mechanical Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan
Abstract This paper presents (1) the development of observation and optical measurement system for the investigation of flame in ultra-thin quartz channels; and (2) the observation and analysis of premixed CH4/air flame propagation and quenching in three quartz combustors with chamber depth of 0.7 mm, 1.0 mm and 1.5 mm, respectively. Flame oscillation is interpreted and discussed according to well-resolved chemiluminescence images of the flame taken by the high-speed ICCD camera. Flammability limits are obtained for different equivalence ratios and mixture velocities. It is found that the quenching distance for heated quartz wall can be smaller than 0.7 mm, and is dependant on the wall temperature. Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Gas-phase combustion; Ultra-thin combustor; Quenching; Flame oscillation; Wall temperature
1. Introduction Enjoying work and life with mobility has become the style of the modern society, and the ever-increasing power consumption of portable electric devices leads to an urgent demand of new portable power source to meet the requirement. The interaction between MEMS technology and knowledge of conventional energy generating/harvesting principles has recently evolved into an area known as Power MEMS. Various power generation concepts such as the micro gas engines and micro fuel cells have been proposed [1]. Hydrocarbon-fueled micro combustor is of great interest in portable power generation applications for higher energy density [2]. Homogeneous combustion is beneficial for high temperature applica-
*
Corresponding author. Fax: +81 3 5800 6999. E-mail address: [email protected] (Y. Fan).
tions such as the TPV system if compared with catalytic combustion [3]. However, elevated wall-effects, associated with the increase of surface-to-volume ratio, make the gas-phase combustion problematic in microscale. It has been widely accepted that flame generally quenches in chambers with small characteristic length. Quenching distance, which is defined as the minimum gap between parallel plates for flame propagation, is in the order of a few millimeters for typical hydrocarbon fuels. Two fundamental quenching mechanisms were known related to the quenching phenomena. One is thermal quenching, for which flame can be thermally deactivated by the heat loss from the flame zone to surrounding walls. The other is chemical quenching, which refers to the absorption of intermediate radicals by adjacent wall surfaces. The thermal/chemical quenching mechanisms on the flame has long been an issue in accurate prediction of flammability and flame stability.
1540-7489/$ - see front matter Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2008.06.219
3084
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
Maruta et al. [4–6] investigated the effect of external heat-input and heat recirculation on flame propagation in straight and U-shaped 2mm-diameter tubes, where the wall temperature is kept at 1120 °C. They also observed flame oscillation (termed as FREI in [4]), and pointed out that it is a periodic process of flame ignition in the hot part of the tube and extinction upstream. A 1-D nonlinear evolutionary equation of the flame front has been applied to interpret the transition from steady condition to flame oscillation [6]. A positive wall temperature gradient in the streamwise direction has been adopted in the modeling in order to represent similar wall thermal condition in the experiments, and they report that the onset of the flame oscillation is due to the inertial effect associated with the heat losses. They also conjectured that the instability might be related to the flame oscillation phenomena in a variable-section channel [7]. A different mathematical model has been proposed by Jackson et al. [8], and they report that flame oscillation can arise when Lewis number is over unit, or when heat losses present. An earlier study on flame oscillation has been conducted in a blown porous media (filtration gas combustion [FGC]) by Fateev et al. [9], which shows much in common with the flame oscillation in a narrow tube. Early investigation of the chemical quenching was made by Sloane et al. [10,11]. They measured molecular fraction of CH4, CO, H2, CO2, OH and H at low pressure for catalytic (Platinum) and non-catalytic (Gold) wall surfaces using a molecular beam mass spectrometer. They demonstrated that the surface recombination with the wall do affect the flame quenching within the region of 1 mm from the surface. However, the thermal effect is dominant. Recently, Miesse et al. [12] and Kim et al. [13] separately made experiments of quenching distance measurement in parallel plates with different wall materials/treatment. Miesse et al. [12] found inert materials such as quartz, alumina and cordierite, are effective against chemical quenching, and claim that chemical quenching instead of thermal quenching dominates for high wall temperature. Kim et al. [13] instead found even metallic material like stainless steel can have smaller quenching distance at high wall temperature after being treated into inert surface by annealing. They claim that for wall temperature ranges from 400 to 600 °C, the flame quenching becomes dependent on the radical removals at the surface, and quenching distance increases; but for wall temperature over 600 °C, the homogeneous chemical reactions overcome the wall-effect of radical removals, and quenching distance becomes smaller with the temperature increase. Previous studies of microscale flammability and stability in some extents reveal the roles of wall thermal and chemical effects. However,
detailed quenching mechanisms in micro-scale, i.e. how the thermal effect couples with the chemical effect, still remain unclear due to insufficient experimental evidence. Of great importance in due investigations are the accurate control and measurement of flame temperature and wall temperature which are related to the thermal effect, the quantitative measurement of near-wall radical concentration which is related to the chemical effect, and the flame visualization which may explain the flame/wall coupling process. The objective of the present study is to provide experimental observations and measurements of flame propagation and quenching in ultra-thin quartz tubes for the investigation of microscale combustion phenomena.
2. Micro combustor and experimental setup Figure 1 shows the quartz combustors developed in the present study. Upstream developing channel, where no flame is expected, has a height of 0.5 mm and a length of 22 mm. The downstream combustion channel has a height of 0.7 mm, 1.0 mm and 1.5 mm, which is smaller than conventional quenching distance for hydrocarbons. The channel height of 0.7 mm is even smaller than recently measured quenching distance for quartz in [12,13]. The combustion channel is 25 mm in length and 16 mm in width. The inner-wall of the combustion channel is made of polished synthetic quartz, so that optical measurements are possible. An inert material with small wall roughness like quartz is also supposed to suppress the radical adsorption and trapping at the surface. The top and bottom sides of the channel are fusion-bonded with black quartz (Nb-doped quartz) plates for absorption of IR light as heating source. The main experimental setup is shown in Fig. 2. The combustor is fixed on to an XY-h stage for fine adjustment of its position. The downstream combustion channel is heated with IR lamp heater from both sides. A quartz rod with a diameter of 20 mm is used for the introduction of IR light, so that condensed IR light can be delivered evenly to most portion of the combustion channel. The center of the rod is aligned to 10 mm upstream the exit in the centerline of the combustor. Two R-type thermocouples (/=0.5 mm) are plugged in 1-mm-diameter holes opened in the transparent quartz layer of the top and bottom walls. Wall temperature measured with the thermo couple TTC is regulated by adjusting the lamp power. CH4 fuel is supplied from a gas cylinder, and air is introduced from a compressor. The CH4 and air flow rates are separately measured and regulated by two sets of flow meters and valves.
Black 2 x φ = 1 mm Thermocouple Holes Quartz
3 mm
φ = 2 mm
3085
m m
15 mm 25 mm
8
Quartz
0.5 mm
2 mm
a
d = 0.7 mm 1.0 mm 1.5 mm
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
b
Fig. 1. Micro quartz combustor. (a) Schematic, (b) Photo. Combustion channel is 16 mm wide and 25 mm long. Combustion channel height, d = 0.7, 1.0, and 1.5 mm.
Fig. 2. Experimental setup. (a) Schematic, (b) Photo, (c) Magnified view of the combustor.
A high-speed ICCD camera (FASTCAM, Photron) is placed in the sideway, and CH*/OH* chemiluminescence images of the flame are taken for flame visualization up to 2000 frames per second. Optical band-path filter at 307 ± 10 nm is employed for OH* chemiluminescence, and 410 ± 10 nm for CH* chemiluminescence. UV laser beam from the dye laser (SCANmate, Lamda Physik) is condensed to thin laser sheet with a
thickness of 0.5 mm, and introduced from the bottom of the combustor for excitation of the OH* and CH* radicals. By utilizing the LIF technique, the present setup is capable of imaging the flame front and also flame temperature with the OH 2line method [14]. Prior to the flame observation, a radiation thermometer (IN140/5-L, IMPAC) is placed in the wall-normal direction for measurement of
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
inner wall surface temperature of the combustor as shown in Fig. 3. Four 2 mm-diameter circular windows are opened perpendicular to the top wall for passing sensing light from the inner-wall surface. The radiation thermometer has sensing wavelength of 5.14 lm, which gives specific emissivity of 0.97 for quartz. The sensing spot of the thermometer is 0.9 mm in diameter for fine spatial resolution.
Thermocouple:
,
, ,
, ,
, ,
,
, ,
; .
1000 800 600 400
3. Experimental results Figure 4 shows the inner-wall surface temperature for the wall temperature measured with the thermocouple TTC=500, 600, 700, 800, 900 and 1000 °C. The temperature profiles vary in the streamwise direction in the range of about 200 °C, and have a peak at the center of the heating quartz rod. The thermocouple temperature TTC and the inner surface temperature measured with the radiation thermometer are within 15 °C, since the thermocouple hole is only 1 mm away from the inner surface of the combustion channel. Although the wall temperature is not uniform in the streamwise direction, it is still reasonable that the thermocouple temperature is used to represent the wall temperature in the following discussions. It is also found in the preliminary experiments that, when a flame exists inside the channel, local temperature near the flame increases by about 100 °C.
Thermocouple PC
1200 Inner-wall:
Temperature (ºC)
3086
Controller
Combustor
0 5 10 15 20 25 Distance from Combustion Channel Inlet (mm)
Fig. 4. Temperature measured at the inner-wall surface.
When the wall temperature is raised beyond a critical value, flame propagates into the combustion channel. Both CH* and OH* chemiluminescence images of the flame are obtained, and the OH* images are shown in this report. Two flame patterns, steady flame (Fig. 5) and oscillating flame (Fig. 6), have been observed in all three combustion channels. Figure 7 shows the evolution of flame patterns with increasing air flow rate. For a fixed fuel flow rate, increasing air flow rate leads to the increase of mixture velocity and the decrease of equivalence ratio. When the air flow rate is small, flame is positioned at the exit, where there is enough air. With a larger air flow rate, the flame begins to propagate into the combustion channel. The flame cell keeps moving upstream from the exit periodically, and becomes oscillating. The oscillation amplitude and flame cell length gradually increases with
Lens Rod IR Lamp Heater
Radiation Thermometer XYZ Stage
2 φ1
Unit: mm
15
25 5 5 5
4 φ2 Fig. 3. Experimental setup for measurement of innerwall surface temperature.
Fig. 5. OH* chemiluminescence of steady flame in the 1.5-mm-thick combustion channel.
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
3087
t=0 s 0.001s 0.002s 0.003s 0.004s 0.005s 0.006s 0.007s 0.008s 0.009s 0.010s 0.011s 0.012s
Ignition
Propagation
0.7 mm
Recharge
*
Fig. 6. OH chemiluminescence of oscillating flame in the 0.7-mm-thick combustion channel.
increasing air flow rate, and then the flame quenches with a slightly larger air flow rate. The mechanisms of this oscillating flame will be discussed below. With an even larger air flow rate, the flame propagates into the combustion channel again, and keeps the steady position. The steady flame holds for a wide air flow rate range, and the position of steady flame moves closer to the exit with increasing the air flow rate as shown in Fig. 8. This is due to the streamwise temperature variation (Fig. 4) since laminar flame velocity can be greatly affected by the temperature. Further increasing the air flow rate, the flame is then blew out of the combustor and positioned at the exit. Finally, it is blew-off with extremely large air flow rates. Note that the quenching conditions between oscillating flame and steady flame is very narrow, and vanishes for large fuel flow rate. Figure 9 shows the schematic of the formation of steady and oscillating flames. In the steady flame, unburned gas mixture is preheated by the hot wall, which enables flame being initialized in the thin channels. On the other hand, the wall temperature is lower than the flame temperature, and thus heat is always lost from the flame to the adjacent wall. However, the flame-wall thermal-coupling should be treated somewhat differently in the analysis of steady and oscillating flames. For the steady flame, the energy conservation principle can be applied; there should be considerable local wall temperature raise due to heat loss from the flame to the wall. Thus, the wall temperature becomes less uniform. But for the oscillating flame, energy conservation is not hold in the same way because time scale of the flame passage is faster than thermal response of the combustor wall.
One whole flame oscillation process that is clearly tracked by time-resolved OH* images (1 ms interval time) is shown in Fig. 6. Flame ignites at the combustor exit at first. This gives the flame an initial heat input. When the exit flame cell gathers certain amount of heat, it grows, expands, and propagates into the combustor. However, when flame is traveling upstream in the channel, the flame continuously losses heat to the wall. The flame quenches upstream where it can not pace further. The fresh unburnt mixture then recharges to the exit for another oscillating cycle. According to the observation, one may divide one oscillation cycle into three main stages, i.e. the ignition (initialization) stage, the propagation stage, and the recharge stage. Then the oscillation frequency can be expressed as 1 ¼ ignition time frequency amplitude þ propagating velocity amplitude þ : mixture velocity
ð1Þ
Base on Eq. (1), the frequency should be lower for larger amplitude. Figure 10 shows the oscillation frequency measured for 0.7 mm-thick and 1.0 mm-thick channel combustors. Frequency in most conditions is in the range of 30–500 Hz, and in each curve decreases with increasing mixture velocity. The frequency shall decrease because the increase of mixture velocity leads to the decrease of equivalence ratio. Note that the oscillating flame is in the fuel-rich zone, flame should be able to penetrate deeper in the combustion channel for smaller equivalence ratios, which
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090 No Flame Flame outside
Oscillating flame
No Flame
25
Distance from Inlet (mm)
3088
20 15 10 5 0 400
(2)
(1)
(3)
Steady flame
(7)
(6)
(4) Flame outside
(8)
(9)
Equivalence Ratio 2
1
(5) No Flame
Flame Position Flame Length
d=0.7 mm combustor CH4: 46 sccm, TTC=800ºC 500 600 Air Flow Rate (sccm)
700
Fig. 8. Variation of position and length of steady flame with air flow rate.
(10)
Mixture Velocity (cm/s) 200
100
0 0 300 500 700 900 1100 Air Flow Rate (sccm) (d = 0.7 mm, fuel: 55sccm) Increase Air Flow Rate
Fig. 7. Evolution of flame patterns with increasing air flow rate. (1) Flame quenching at small air flow rates; (2) Flame at the exit for small air flow rates; (3) Oscillating flame with a small amplitude; (4) Oscillating flame with a large amplitude; (5) Flame quenching at moderate air flow rates; (6), (7) and (8) Steady flame at different positions with the change of air flow rate; (9) Flame at the exit for large air flow rates; (10) Flame quenching at large air flow rates.
actually coincides with the experimental observations that the amplitude becomes larger (not shown). It is also shown in Fig. 10 that the flame oscillation happens at larger mixture velocity for thinner channel combustor, and at slightly larger mixture velocity for lower wall temperature. Figure 11 shows the flammability limits in the 0.7-mm-thick channel for different wall temperatures. Compared with the steady flame, oscillating flame is in the higher equivalence ratio region. In this fuel-rich region, the CH4/air mixture can not be ignited inside the combustion channel, so that the flame is ignited first at the exit where fuel meets more air, and then propagates into the
channel. The flammability limits region becomes broader with increasing wall temperature. Compared with the data in a 2-mm-diameter tube combustor [4–6], the present combustor should have stronger wall-effect due to the higher surface-tovolume ratio and narrower channel. However, the flame in the present combustor can be stabilized at much higher velocities. It is conjectured that the recirculation zone downstream the backward-facing step in the combustion channel stabilizes the flame. It is also noted that the range of the equivalence ratio within the flammability limit is narrower than the data in the long tube combustor [15]. Maruta et al. [4–6] explained in their 1-D model that the oscillating flame is actually the transition from the normal steady flame and another ‘weak’ steady flame at extremely small mixture velocity. However, no ‘weak’ flame has been found in the present experiment. The flame oscillation occurs only in the fuel-rich conditions, which differs from the observations of flame oscillation between the normal flame and ‘weak’ flame conditions in [4–6]. Note that the ignition position for the oscillating flame is at the exit in the present study, which is also different from the ignition at a hot part of the tube in [4–6]. These can be the result of a shorter combustion channel and the bell-shape streamwise wall temperature distribution in the present study. Note that a very narrow quenching zone exists between the normal steady flame and oscillating flame zones (Fig. 11b). Since the quenching at large flow rates is obviously due to the blow-off, it is conjectured that this narrow zone with small flow rates is corresponding to the flashback phenomena. This also explains the reason why the oscillating flame propagates upstream after the ignition at the exit. Figure 12 shows the flammability limits in the 1.0-mm-thick channel. Compared with the 0.7mm-thick channel, the 1.0-mm-thick channel has a broader flammability limits region for the wall temperature of 800 °C, but the flammability
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
b
a
Initial heat input by ignition at exit
Wall
Wall
(Heat loss) Q
Q
Flame Q (Heat loss)
Q
Flame Q (Pre-heat)
Q
3089
Q (Pre-heat)
Q
Fig. 9. Schematic of the formation of (a) steady flame, and (b) oscillating flame.
Frequency (Hz)
400
73sccm 64sccm
300 200
73sccm
46sccm 27sccm 100 37sccm 0 20
55sccm 30
Steady Flame
d=0.7 mm, TTC=800ºC d=1.0 mm, TTC=800ºC d=1.0 mm, TTC=1000ºC
CH4: 55sccm 46sccm
40 50 60 Mixture Velocity (cm/s)
70
80
Fig. 10. Oscillation frequency measured with high-speed ICCD camera.
150 100
Mixture Velocity (cm/s)
500
Oscillatiing Flame
150 65 sccm 56 sccm
a TTC = 800ºC CH4: 74 sccm
100
46 sccm 37 sccm 28 sccm
50
Mixture Velocity (cm/s)
No Flame
0 200 CH4: 68 sccm 150 55 sccm 46 sccm 36 sccm 100 28 sccm
b TTC = 1000ºC
50 0 0.0
0.5
1.0
1.5
Equivalence Ratio Fig. 11. Flammability limits in the 0.7-mm-thick combustion channel. (a) TTC = 800 °C, (b) TTC = 1000 °C.
a TTC = 800ºC CH4: 92 sccm 83 sccm
50 0 200 CH4: 92 sccm
150
b TTC = 1000ºC
73 sccm 54 sccm
100
Steady Flame
73 sccm 64 sccm 46 sccm 37 sccm
Oscillatiing Flame
37 sccm
50 28 sccm
0 0.0
0.5 1.0 Equivalence Ratio
1.5
Fig. 12. Flammability limits in the 1.0-mm-thick channel. (a) TTC = 800 °C, (b) TTC = 1000 °C.
region is only slightly enlarged for the higher wall temperature of 1000 °C. This is probably because the wall-effect on quenching for the thinner combustion channels becomes more severe at low wall temperature. The maximum CH4 flow rate for steady flame is 92 sccm for the 1.0-mm-thick channel, while it is only 74 sccm for the 0.7-mmthick channel. This is due to the blow-off, because the maximum flow rates correspond to almost the same bulk gas velocities. As shown in Fig. 13, critical wall temperature for flame to exist becomes higher for thinner combustion channel, which means that more thermal input is necessary for thinner channels due to stronger wall effect. Miesse et al. [12] report that the quenching distance for high wall temperature (near 1000 °C) strongly depends on the surface
Critical Wall Temperature (ºC)
3090
Y. Fan et al. / Proceedings of the Combustion Institute 32 (2009) 3083–3090
region becomes narrower for thinner channel, and broader for higher wall temperature. (3) Quenching distance in the quartz channel is dependant on the wall temperature, and can be smaller than 0.7 mm for the wall temperature over 800 °C.
1200 1000 800 600 400
Acknowledgment 200 0 0.0
0.5 1.0 Channel Height (mm)
1.5
Fig. 13. Effect of the combustion channel height on the critical wall temperature.
material and chemical quenching prevails, whereas the thermal quenching plays a dominant role at lower wall temperature. However, in the present result at 800 °C, flame can still propagate in a 0.7-mm-thick quartz channel, which is smaller than the wall gap of 1 mm previously reported [12]. Since the quenching distance is still dependant on the wall temperature, one can expect thermal quenching mechanism play an important role even at high wall temperature. 4. Conclusions Micro quartz combustor that enables optical access has been developed for the investigation of quenching mechanisms in micro channels. Microscale steady and oscillating flames are confirmed through CH*/OH* chemiluminescence. Premixed CH4/air combustion is examined in the quartz combustion channels with the height of 0.7, 1.0 and 1.5 mm. Discussions are made on the thermal-coupling between the flame and wall in microscale flames. The following conclusions are derived: (1) Flame oscillation in ultra-thin channels can be interpreted as a periodic process of three stages, i.e. the initialization and expansion of flame cell at ignition stage, the flame propagation and quenching at the propagation stage, and the recharge of gaseous mixture to the exit at the recharge stage. (2) Flammability limits
The present work is supported by Grant-inAid for Scientific Research (B) (No. 19360094) by JSPS, Japan. YF was partially supported by the 21st Century COE Program, ‘‘Mechanical Systems Innovation,” by MEXT, Japan.
References [1] S.A. Jacobson, A.H. Epstein, An Informal Survey of Power MEMS, Paper ISMME2003-K18, Tsukuba, 2003. [2] A.C. Fernandez-Pello, Proc. Combust. Inst 29 (2002) 883–899. [3] T. Okamasa, G.-G. Lee, Y. Suzuki, N. Kasagi, S. Matsuda, J. Micromech. Microeng. 16 (2006) S198– S205. [4] K. Maruta, J.K. Parc, K.C. Oh, T. Fujimori, S.S. Minaev, R.V. Fursenko, Combust. Expl. Shock Waves 40 (5) (2004) 516–523. [5] K. Maruta, T. Kataoka, N. Kim, S. Minaev, R. Fursenko, Proc. Combust. Inst. 30 (2005) 2429– 2436. [6] S. Minaev, K. Maruta, R. Fursenko, Combust. Theor. Model. 11 (2) (2007) 187–203. [7] S.S. Minaev, V.S. Babkin, Combust. Expl. Shock Waves 37 (1) (2001) 13–20. [8] T.L. Jackson, J. Buckmaster, Z. Lu, D.C. Kyritsis, L. Massa, Proc. Combust. Inst. 31 (2007) 955–962. [9] G.A. Fateev, O.S. Rabinovich, M.A. Silenkov, Proc. Combust. Inst. 27 (1998) 3147–3153. [10] T.M. Sloane, J.W. Ratcliffe, Combust. Flame 47 (1982) 83–92. [11] T.M. Sloane, A.Y. Schoene, Combust. Flame 49 (1983) 109–122. [12] C.M. Miesse, R.I. Masel, C.D. Jensen, M.A. Shannon, M. Short, AIChE J. 50 (2004) 3205–3214. [13] K.T. Kim, D.H. Lee, S. Kwon, Combust. Flame 146 (2006) 19–28. [14] R. Cattolica, Appl. Opt. 20 (1981) 1156–1166. [15] G.F. Wang, B.J. Zhong, J. Tsinghua Univ. (Sci. Tech.) 46 (2006) 276–279.