- Email: [email protected]
Lower limit of weak flame in a heated channel
Available online at www.sciencedirect.com
Proceedings of the
Proceedings of the Combustion Institute 32 (2009) 3075–3081
Combustion Institute www.elsevier.com/locate/proci
Lower limit of weak flame in a heated channel Yosuke Tsuboi a, Takeshi Yokomori b, Kaoru Maruta a,* a
b
Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba, Sendai 980-8577, Japan Department of Mechanical Engineering, Keio University, 3-14-1 Hiyoshi, Kouhoku, Yokohama, Japan
Abstract Stationary and non-stationary flame responses in a heated meso-scale channel were investigated both experimentally and computationally. Special attention was paid to flame stabilities, particularly the existence of lower limit of weak flame regime. Previous microcombustion methodology with an external heat source to form stationary temperature gradient in the channel wall was employed. Normal stable flames at high and low velocity regimes, and non-stationary dynamic flames (FREI) at moderate velocity regime were confirmed experimentally. In addition to them, the lower limit of weak flame regime was experimentally identified for the first time. Measured temperature increase in such weak flame was found to be approaching to nearly zero degree. It was found that the flame temperature at the lower limit of the weak flame regime corresponds to ignition temperature of the employed mixture. One-dimensional computations with detailed chemistry and transport properties exhibited e-shaped curve which has additional lowest velocity regime with previous S-shaped curve. Computational results comprehensively support and interpret the present experimental results indicating that the lower limit of weak flame regime is induced by the weakened reaction due to less frequent molecular collisions by diffusive mass dissipation at extremely low velocity regime. Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Weak flame; Combustion with heat recirculation; Limit of mild combustion; Mass dissipation limit
1. Introduction To reduce energy consumption in combustion systems, it is important to satisfy thermodynamic requirements by realizing controlled combustion [1,2] so as to enable arbitral flame temperature and reaction speed for higher thermal efficiencies. For this, combustion with heat and/or mass recirculation processes is one of the most promising candidates. High-temperature air combustion technology (HiCOT) [3–5], which is mainly utilized for industrial furnaces, can be raised as one *
Corresponding author. Fax: +81 22 217 5311. E-mail address: [email protected] (K. Maruta).
of the successful examples of combustion with heat and mass recirculation. By this combustion mode, compact furnaces characterized by up to 30% reduction of fuel consumption, low NOx emissions and low noise have been realized. In addition to high temperature air combustion, several terminologies, such as mild combustion [6] and flameless combustion [7], can be utilized for dealing with analogous new combustion phenomena and related technologies. However, optimization of such combustion processes has not yet been fully achieved because of insufficient fundamental understanding of the limits of such combustion with the heat/mass recirculation process, as well as the chemistry in a comparatively low temperature regime.
1540-7489/$ - see front matter Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2008.06.151
3076
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
To examine general characteristics of combustion with heat recirculation process, the response of flame in a cylindrical quartz tube with a temperature gradient has been recently investigated [8,9] in the context of microcombustion. An external heat source was employed to resemble heat recirculation with a thermally thick wall in those studies. Results showed the existence of stable flames in high and low velocity regimes, as well as other dynamic flames with repetitive extinction and ignition (FREI) and cyclic oscillations in a moderate velocity regime. FREI and other dynamic flames have been covered and intensively examined by other research groups [10,11]. Stable flame in the low velocity regime has been analytically predicted [8], then observed at mean flow velocity around 2–3 cm/s experimentally and examined computationally [8,9]. Although this sort of flame is an example of mild combustion, the existence of a lower limit of the weak flame regime, namely, the weakest flame (or lowest burning velocity) is left unsolved. We have already suggested the possibility of the weakest flame [12], however, the interpretation of the phenomena was insufficient so far. The aim of the present study was to investigate flame responses in a heated channel both experimentally and computationally. Special attention was paid to flame stabilities, particularly the existence of a lower limit of the weak flame regime. Thermal management by an external heat source was employed to examine such flames. At first, flame responses to mean flow velocity were investigated experimentally. Then computations with detailed chemistry and transport properties are performed to interpret experimental results and elucidate the existence of the weakest flame. 2. Experimental setup and methods In the present investigation, we employed a methodology similar to that for our microcombustion study [9], but with improvements enabling us to focus on the lower limit of weak flame. An external heat source for modeling heat recirculation with a thermally thick wall, which can also be regarded as heat loss compensation, was employed. A schematic of the apparatus is shown in Fig. 1. A premixed methane–air mixture was supplied through mass flow controllers to the channel with inner diameter of 2.0 mm, which is slightly smaller than the conventional quenching diameter. Since the convective-diffusive zone can be expected to be large under low velocity conditions, we employed a 300 mm long channel connected with a 40 m long polyethylene tube at the upstream side, and a 3 m long extension silicon tube at the downstream side.
Fig. 1. A schematic of the experimental apparatus and temperature profile along the inner surface of the tube wall.
A flat flame burner was employed for heating the center part of the channel so that a positive temperature gradient was formed on the inner surface of the tube wall in the axial direction. For better visualization of blue flame, i.e., emissions from CH, C2, etc. in the tube, a hydrogen–air mixture was used for this burner. The temperature profile along the inner surface of the tube wall was measured by a K-type thermocouple in advance. The origin of the horizontal axis was set at the upstream edge of the flat flame burner. A maximum wall temperature of 1350 K was chosen for this study so that the mixture in the tube could be autoignited. After self-ignition, flame in the channel was stabilized at a certain location in the positive temperature gradient where local flame propagation and mixture flow were balanced. A digital still camera with a CH filter was used to detect flames in the tube. Long exposure times (1–2400 s) were employed for flames in the low velocity regime because their luminosities are extremely weak. The effects of heat release by combustion inside the tube on the wall temperature profile could be neglected in the present experiments because of the large amount of heat passing through the tube from the flat flame burner to the ambient and the large heat capacity of the tube compared with the small heat of combustion in the tube. 3. Experimental results Figure 2 shows flame images in (a) high, (b) moderate and (c) low velocity regimes (refer to Fig. 3 for each regime). Figure 3 shows flame responses including extinction and ignition points of FREI (Fig. 2b) [8,9] as a function of mean flow velocity at equivalence ratio / = 1. The dashed line in Fig. 3 denotes tube wall temperature profile. In high (100–40 cm/s) and low (5–0.2 cm/s) velocity regimes, stable normal flat flames were observed (Fig. 2a and c). In the moderate velocity
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
120
1600 φ = 1.0
80
1400
Tw
1200 1000
60 (1) Normal
800
40 Extinction
20 0
Wall Temperature, Tw (K)
Mean flow velocity (cm/s)
100
Flow
Normal flame FREI Estimated points of ignition
600
(2) FREI
Ignition
400
(3) Normal
-1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0
0.2
0.4
0.6
0.8
1.0
Location (cm)
Fig. 3. Measured flame position and extinction and ignition points of FREI for variable mean flow velocity at / = 1.0. Estimated ignition locations in upper normal flame regime are also indicated.
regime (40–5 cm/s), non-stationary flames termed FREI (Fig. 2b) and other oscillating flames were observed. The present flame responses agreed well with those of previous studies [9,13]. In the high velocity regime, the stabilized locations of normal flame and estimated points of ignition, which are based on the experimental ignition temperatures [14], are plotted in Fig. 3. After self-ignition at the downstream side, an ignition kernel moves upstream, and the flame is stabilized at an upstream location. The location of stable flame shifts toward the upstream side with the decrease of mean flow velocity, and the ignition position curve exhibits a similar tendency.
In the moderate velocity regime, luminous reaction zones appeared to be broadened, as shown in Fig. 2b. A phenomenological description for FREI is as follows [8,9]. An ignition kernel starts emitting luminescence from the point of ignition at the downstream side and then the luminous zone moves toward the upstream side until it reaches to a stable point. However the flame is immediately quenched due to the large heat loss caused by the low temperature of the wall. After a time delay from quenching, re-ignition occurs at the ignition point. This cycle is repeated regularly. Note that the ignition point can be identified as the location of the visible flame kernel, and its local temperature, that is, ignition temperature, can be estimated from this location. Ignition and extinction points shift to the upstream side with the decrease of flow velocity. In the low velocity regime, stationary flames with extremely low luminosities were observed. Long exposure time was required to record those weak flame images. Figure 2c was obtained at an exposure time of 1200 s with background compensation. Although a weak flame at a mean flow velocity of 0.2 cm/s was observed, no flame was observed at a low velocity less than 0.2 cm/s, even with much longer exposure time. This indicates that the existence of a lower limit of weak flame, even with heat compensation by external heating. This finding adds to our previous results which were without a lower limit of weak flame. As shown in Fig. 3, the stabilized locations of weak flames were around x = 0, which is close to or on the extrapolated line of the ignition positions of FREI in the moderate velocity regime. This agrees very well with theoretical prediction [13] and indicates that those flames are stabilized close to ignition positions. Therefore, weak flame temperatures are considered to be close to the inner surface temperature of the tube wall in the low velocity regime. Figure 4 shows the temperature difference between the flame and the inner surface of the tube wall at the flame position as measured by a thermocouple. The temperature
30
Temperature difference (K)
Fig. 2. Flame images (/ = 1.0); (a) normal flame (Vm = 50 cm/s), (b) FREI (Vm = 20 cm/s), and (c) weak flame (Vm = 0.2 cm/s).
3077
φ φ φ φ
25
= 1.2 = 1.0 = 0.85 = 0.6
20 15 10 5 0 0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Flow velocity (cm/s)
Fig. 4. Measured temperature difference between the flame and the inner surface of the tube wall at the flame position (/ = 0.6, 0.85, 1.0 and 1.2).
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
4. Computational methods and results To elucidate experimental results, particularly for the mechanism of the weakest flame,
one-dimensional computations with detailed chemistry were conducted. A steady-state flame code [9] based on PREMIX and GRI-mech 3.0 [17] was used. The energy equation, which includes convective heat transfer between gas and wall, is as follows: K X dT _ dT 1 d kA dT þ A M qY k V k cpk dx cp dx dx cp k¼1 dx þ
K A X A 4kNu ðT w T Þ ¼ 0 x_ k hk W k cp k¼1 cp d 2
_ T, x, cp, A and k are mass In this equation, M, flow rate, gas temperature, spatial coordinate, specific heat, cross-sectional area and thermal conductivity, respectively. q, Yk, Vk, x_ k , Wk, Wk and d are density, mass fraction, diffusion velocity, production rate, specific enthalpy, molecular weight and channel diameter, respectively. The computational domain is 10 cm long. Because of the fully developed flow in a circular tube for the case between constant wall heat flux and constant wall temperature, Nusselt number of the inner wall surface was assumed to be constant (Nu = 4). Tw is the fixed wall temperature and the profile in the computation is almost the same as that of the test section shown in Fig. 1. The maximum wall temperature is 1300 K. Figure 5 shows the computed flame response to mean flow velocity at equivalence ratio / = 1. The location of the CH peak is considered to be the flame position. The dashed line with open circles in the figure shows the wall temperature profile. The origin of the horizontal axis is set at the inlet of the tube. The computational flame response showed an e-shaped curve, which has an additional lower velocity branch with the previous Sshaped curve [8,9]. Based on the detailed description of the stability analysis for the S-shaped curve in our previous reports [8,9], the present lowest velocity branch is considered to be unstable. If this assumption is true, the existence of two stable and two unstable solutions in the four regimes can
1400 100 (1) Stable 1200 10
(2) Unstable 1000
(3) Stable 800
1 (4) Unstable
600 0.1
0.01
Flame position (φ =1.0) Wall temperature
4.5
5.0
5.5
Wall temperature (K)
difference becomes smaller with the decrease of mean flow velocity and is almost zero at Vm = 0.2 cm/s, where the wall temperature is around 1225 K regardless of mixture composition. In the present experiments, we tried several types of thermocouple for the measurements. Results show that the temperature differences between the flame and the wall measured by those thermocouples were nearly equal, although the absolute values exhibited some differences. Thus, discussion based on the temperature differences between the flame and the wall surface, which are asymptote to zero in Fig. 4, is quite reasonable. It is noted that thermal quenching by intrusive thermocouple measurements is hard to occur due to the external heater in the present system since the thermocouple is inserted from the downstream side where the temperature of the leading wire of the thermocouple is maintained at high temperature and which has a small temperature gradient in it. It is also justified from the long time exposure recording in which the thermocouple was in contact with the weak flame during temperature measurements even for the weakest flame case. Based on two characteristics of weak flame, that is, (1) small temperature increase and (2) flame location close to the ignition point, the flame temperature at the lower limit of weak flame is considered to correspond to the ignition temperature of the given mixture. Although ignition and flame propagation have been considered to be separate phenomena in general, stationary propagating flame near the ignition point in weak flame was successfully established in the present experiment. Hence, ignition and flame propagation would merge with each other at this extreme condition. This fact also indicates that the general ignition temperature, which is independent of the experimental apparatus and methodology, can be expected to be identified with the position of the weakest flame by the present methodology. Ju and Xu [15] and Chao et al. [16] theoretically investigated self-sustaining flame propagation and extinction of premixed flames with heat recirculation in meso-scale channels. They also showed the weak flame regime and extinction limits of weak flame for the case of low flow velocity conditions. However, since they dealt with selfsustaining flames, heat generation by combustion was the only source of heat in their models. Thus, heat recirculation became weaker with the decrease of flame temperature. Therefore, reported extinction limits of stable weak flame were caused by thermal quenching, different from the present phenomena.
Mean flow velocity (cm/s)
3078
400
6.0
200 6.5
Location (cm)
Fig. 5. Computed flame position with mean flow velocity (/ = 1.0).
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
be inferred by the present computation. Under this assumption, the existence of the lower limit of the experimental weak flame can be interpreted as the lower limit of the third regime in Fig. 5. The upper positive gradient (regime (1)) in Fig. 5 indicates normal stable flame. This solution corresponds to the classical upper stable solution of flame response with heat loss [18]. A flame at the highest flow velocity condition was just prior to blow-off since the maximum wall temperature was located at 6.5 cm. It can be seen that the flame position shifted toward the upstream side with the decrease of the mean flow velocity in this regime, which agrees with experimental results. The upper turning point corresponds to the conventional thermal quenching [18] caused by heat loss to the wall. Below the upper turning point, unstable solutions (regime (2)), which correspond to FREI phenomena, were obtained. In the present experiment, the thermal quenching limit did not appear due to the high temperature of the wall resulting from external heating. Instead, non-stationary flames such as FREI were observed in the moderate velocity regime, which agreed with findings of previous studies [9,12]. The lower positive gradient (regime (3)) is the low velocity regime for stable weak flame. Due to the presence of the high temperature wall, flames in this low velocity regime were realized in spite of the low temperature increase. To examine the experimental lower limit of the weak flame branch, variation of the computed temperature difference between the flame and the fixed inner wall surface temperature at the flame position was investigated as shown in Fig. 6. Flame temperature is defined as the temperature of gas at the CH peak. The computed temperature difference becomes smaller with the decrease of mean flow velocity, which agrees well with findings of the present experiment. The temperature difference becomes almost zero (Tg Tw < 1 K) at a mean flow velocity of Vm = 0.1 cm/s when wall temperature is 1230 K, which quantitatively agrees with the experimental results (1225 K).
Therefore, the lower limit of stable weak flame is considered to be at the midpoint of the lowest velocity branch in Fig. 5 (regime (4)). Since the boundary between the stable and unstable branches is unresolved, stability analysis should be conducted in a future study. It should be noted that the extremely small temperature increase does not directly correspond to flame quenching in the low velocity regime; however, the conventional reaction with intense heat release no longer occurs in regime (4), even though the heat loss from the flame zone is compensated by the external heating. If thermal quenching does not occur at the lower limit of the stable weak flame, what is the possible mechanism for the extinction of weak flame? We have reported details of the weak flame structure in a previous study [9] and have shown that weak flame has a broader reaction zone due to the effect of mass diffusion. We consider the cause of such limit to be related to the dominance of diffusive mass dissipation over the convective mass transfer in the extremely low velocity regime. To examine the effect of diffusion on the lower limit of weak flame, convective and diffusive mass fluxes of OH were compared for the case of high and low velocity regimes as shown in Figs. 7 and 8. OH was chosen because it is one of the key species for the initiation of chain branching reactions as well as its high diffusivity. Each term in Figs. 7 and 8 was estimated from qYOHVOH and qYOHVm in the computation. VOH is the diffusion velocity of OH and Vm is the mean flow velocity. In each of these two figures, the shaded area indicates the reaction zone estimated by the full-width at half maximum of the CH profile. As shown in Fig. 7, mass transport is dominated by convection in the case of the high velocity regime (Vm = 134 cm/s), which is similar to conventional flame. Therefore, total mass flux is almost the same as convective mass flux. On the other hand, diffusive mass flux is dominant in the low velocity regime (Vm = 1.82 cm/s), as shown in Fig. 8. This means that the latent effect of diffusion on total mass transport increases
Tg-Tw
-3
1.5x10
-3
1.0x10
20
Flux (g/cm 2 s)
Temperature difference (K)
30
3079
10
-4
5.0x10
0.0 OH Flux (Vm = 134 cm/s)
Diffusion Convection Total mass flux
-4
-5.0x10
0 0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Mean flow velocity (cm/s)
-3
-1.0x10
5.0
5.2
5.4
5.6
5.8
6.0
6.2
6.4
6.6
6.8
7.0
Location (cm)
Fig. 6. Computed temperature difference between the flame and the inner surface of the tube wall at the flame position (/ = 1.0).
Fig. 7. Convective mass flux, diffusive mass flux and total mass flux of OH at Vm = 134 cm/s.
3080
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081 -7
6.0x10
1.0
V = 31 cm/s
-7
4.0x10
-7
0.8
Pressure (atm)
Flux (g/cm2s)
2.0x10
0.0 -7
-2.0x10
-7
-4.0x10
-7
-6.0x10
OH Flux (Vm= 1.82 cm/s)
-6
-1.0x10
0.4 0.2
Diffusion Convection Total mass flux
-7
-8.0x10
0.6
0.0
-6
-1.2x10
5.0
5.2
5.4
5.6
5.8
6.0
6.2
6.4
6.6
6.8
4.5
7.0
5.0
due to extremely low flow velocity. Furthermore, the region of negative total mass flux in the low velocity regime is broader than that in the high velocity regime, and the absolute value of the negative total mass flux is larger than that of the positive total mass flux in the same figure. Note that the negative value means that OH moves in the upstream direction. Since the radical species represented by OH, which are vital for the chain branching reaction, diffuse out from the reaction zone in both the downstream and upstream directions, the number of collisions and the rate of production of chain carriers are both considered to be low. Thus, the flame temperature may become lower with the decrease of mean flow velocity. In order to further examine the above considerations, the low pressure condition can be given as an example of an analogous situation for the diffusion controlled condition because the reaction zone becomes broader with the decrease of pressure. Under the low pressure conditions, molecular collisions leading to chain branching reactions become less frequent due to the low density of the mixture. It is well known that the first limit of the explosion curve for the stoichiometric H2–O2 mixture [19] is caused by the lower frequency of collisions and by removal of chain carriers on the surface (wall effect) with the decrease of pressure. Although the wall surface was maintained at high temperature in the present study, there should be an essential and critical pressure condition for sustaining the chain branching reaction system [19]. Figure 9 shows computational variations of flame position with pressure at the given mean flow velocity for the case of / = 1.0. Here, the mean flow velocity is maintained at constant (31 cm/s) for different pressure levels. It is seen that the obtained flame position curve can be divided into three regimes. According to our previous experiments on combustion in a heated channel under low pressure conditions [20,21], it is considered that the upper regime indicates unstable solutions where FREI can be observed
6.0
6.5
Fig. 9. Computed flame position with pressure (/ = 1.0, Vm = 31 cm/s).
experimentally and that the middle regime corresponds to a normal flame branch. The existence of a lower regime in the present calculation is new and this implies the existence of the lower limit of stable weak flame under the low pressure conditions even though heat loss was compensated by the heated wall. Figure 10 shows variation of computed temperature difference between the flame and the inner wall surface at the flame position for different pressure levels. The temperature differences become smaller with the decrease of pressure, and it is almost zero (Tg Tw < 1 K) at the pressure levels in the lower unstable regime, which is similar to the results in Figs. 5 and 6 for the case of the lowest velocity regime. Therefore, the lower limit of stable weak flame is expected to be induced by such low pressure conditions. In the present computations, there are two different approaches. One is diffusion controlled conditions due to low flow velocity and the other is low density conditions due to low pressure. This suggests that the reaction is weakened by a lower frequency of molecular collisions even though heat loss is compensated by the heated wall. Thus, it is considered that the existence of the weakest flame observed in the present experiment is 40
Tg - Tw
Temperature difference (K)
Fig. 8. Convective mass flux, diffusive mass flux and total mass flux of OH at Vm = 1.87 cm/s.
5.5
Location (cm)
Location (cm)
30
20
10
0 0.00
0.02
0.04
0.06
0.08
Pressure (atm)
Fig. 10. Computed temperature difference between the flame and the inner surface of the tube wall at the flame position (/ = 1.0, Vm = 31 cm/s).
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
induced by less frequent molecular collisions. If this is correct, the absolute limit of mild combustion with the heat recirculation process should be finally induced by this mechanism. Such information is expected to contribute to well-controlled, near-limit operation of mild combustion. 5. Conclusions Stationary and non-stationary flame responses in an externally heated meso-scale channel were investigated both experimentally and computationally. Special attention was paid to the flame stability regimes, especially with regard to the existence of the lower limit of weak flame. Although the existence of weak flame at mean flow velocity of 0.2 cm/s was confirmed experimentally, no flame could be observed at much lower velocity conditions. This suggests the existence of a lower limit of weak flame. The measured temperature difference between the flame and the tube inner surface at the flame position became zero at the above lower limit. Thus, ignition and flame propagation are thought to merge with each other under such extreme conditions. This implies that the flame temperature at the lower limit of weak flame corresponds to the ignition temperature of the given mixture. Computational flame responses exhibited an eshaped curve, i.e., the lowest velocity regime with the previous S-shaped curve. The computed temperature difference between the flame and the inner surface of the tube at the flame position became zero in the lowest velocity regime, which quantitatively supports the experimental findings. Two kinds of computational approaches, (1) diffusion controlled conditions at low velocity, and (2) low pressure conditions, suggested that the reactions were weakened because of less frequent molecular collisions at low velocity or low pressure conditions even if the wall temperature at the reaction zone was maintained at a high temperature. The limit of mild combustion with the heat recirculation process is considered to be finally induced by the same mechanism.
Acknowledgment The authors would like to thank Mr. Susumu Hasegawa for his help with experiments and Dr.
3081
Hisashi Nakamura, assistant professor of Tohoku University for fruitful discussion.
References [1] Road map for energy saving (in Japanese), 2007, available at http://www.nedo.go.jp/informations/ other/190423_1/190423_1.html. [2] S.K. Som, A. Datta, Prog. Energy Combust. Sci. 34 (3) (2008) 351–376. [3] M. Katsuki, T. Hasegawa, Proc. Combust. Inst. 27 (1998) 3135–3146. [4] H. Tsuji, A.K. Gupta, T. Hasegawa, M. Katsuki, K. Kishimoto, M. Morita, CRC Press, Boca Raton, 2003. [5] B. Dally, E. Riesmeier, N. Peters, Combust. Flame 137 (2004) 418–431. [6] A. Cavaliere, M.D. Joannon, Prog. Energy Combust. Sci. 30 (2004) 329–366. [7] K. Maruta, K. Muso, K. Takeda, T. Niioka, Proc. Combust. Inst. 28 (2000) 2117–2123. [8] K. Maruta, J.K. Park, K.C. Oh, T. Fujimori, S. Minaev, R. Fursenko, Combust. Expl. Shock Waves 40 (5) (2004) 516–523. [9] K. Maruta, T. Kataoka, N.I. Kim, S. Minaev, R. Fursenko, Proc. Combust. Inst. 30 (2005) 2429– 2436. [10] T.L. Jackson, J. Buckmaster, Z. Lu, D.C. Kyritsis, L. Massa, Proc. Combust. Inst. 31 (2007) 955–962. [11] G. Pizza, C.E. Frouzakis, J. Mantzaras, A.G. Tomboulides, K. Boulouchos, Combust. Flame 152 (3) (2008) 433–450. [12] Y. Tsuboi, T. Yokomori, K. Maruta, ASME International Mechanical Engineering Congress and Exposition, Seattle, IMECE2007-43339, 2007. [13] S. Minaev, K. Maruta, R. Fursenko, Combust. Theory Model. 11 (2007) 187–203. [14] T. Kataoka, N.I. Kim, T. Yokomori, S. Maruyama, K. Maruta, Int. Symp. Combust. (WIPP) (2004) 93. [15] Y. Ju, B. Xu, Proc. Combust. Inst. 30 (2005) 2445– 2453. [16] C.Y.H. Chao, K.S. Hui, W. Kong, P. Cheng, J.H. Wang, Int. J. Heat Mass Trans. 50 (2007) 1302–1313. [17] Information available at: http://www.me.berkeley. edu/gri_mech. [18] F.A. Williams, Combustion Theory: The Fundamental Theory of Chemically Reacting Flow Systems, Benjamin/Cummings, 1985, 275 pp. [19] I. Glassman, Combustion, third ed., Academic Press, 1996, p. 63. [20] Y. Tsuboi, T. Yokomori, S. Hasegawa, K. Maruta, Paper 182, 21st International Colloquium on the Dynamics of Explosions and Reactive Systems, Poitiers, France, July 23–27, 2007. [21] Y. Tsuboi, T. Yokomori, K. Maruta, Combust. Sci. Technol. 180 (2008) 2029–2045.
Proceedings of the
Proceedings of the Combustion Institute 32 (2009) 3075–3081
Combustion Institute www.elsevier.com/locate/proci
Lower limit of weak flame in a heated channel Yosuke Tsuboi a, Takeshi Yokomori b, Kaoru Maruta a,* a
b
Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba, Sendai 980-8577, Japan Department of Mechanical Engineering, Keio University, 3-14-1 Hiyoshi, Kouhoku, Yokohama, Japan
Abstract Stationary and non-stationary flame responses in a heated meso-scale channel were investigated both experimentally and computationally. Special attention was paid to flame stabilities, particularly the existence of lower limit of weak flame regime. Previous microcombustion methodology with an external heat source to form stationary temperature gradient in the channel wall was employed. Normal stable flames at high and low velocity regimes, and non-stationary dynamic flames (FREI) at moderate velocity regime were confirmed experimentally. In addition to them, the lower limit of weak flame regime was experimentally identified for the first time. Measured temperature increase in such weak flame was found to be approaching to nearly zero degree. It was found that the flame temperature at the lower limit of the weak flame regime corresponds to ignition temperature of the employed mixture. One-dimensional computations with detailed chemistry and transport properties exhibited e-shaped curve which has additional lowest velocity regime with previous S-shaped curve. Computational results comprehensively support and interpret the present experimental results indicating that the lower limit of weak flame regime is induced by the weakened reaction due to less frequent molecular collisions by diffusive mass dissipation at extremely low velocity regime. Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Weak flame; Combustion with heat recirculation; Limit of mild combustion; Mass dissipation limit
1. Introduction To reduce energy consumption in combustion systems, it is important to satisfy thermodynamic requirements by realizing controlled combustion [1,2] so as to enable arbitral flame temperature and reaction speed for higher thermal efficiencies. For this, combustion with heat and/or mass recirculation processes is one of the most promising candidates. High-temperature air combustion technology (HiCOT) [3–5], which is mainly utilized for industrial furnaces, can be raised as one *
Corresponding author. Fax: +81 22 217 5311. E-mail address: [email protected] (K. Maruta).
of the successful examples of combustion with heat and mass recirculation. By this combustion mode, compact furnaces characterized by up to 30% reduction of fuel consumption, low NOx emissions and low noise have been realized. In addition to high temperature air combustion, several terminologies, such as mild combustion [6] and flameless combustion [7], can be utilized for dealing with analogous new combustion phenomena and related technologies. However, optimization of such combustion processes has not yet been fully achieved because of insufficient fundamental understanding of the limits of such combustion with the heat/mass recirculation process, as well as the chemistry in a comparatively low temperature regime.
1540-7489/$ - see front matter Ó 2009 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2008.06.151
3076
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
To examine general characteristics of combustion with heat recirculation process, the response of flame in a cylindrical quartz tube with a temperature gradient has been recently investigated [8,9] in the context of microcombustion. An external heat source was employed to resemble heat recirculation with a thermally thick wall in those studies. Results showed the existence of stable flames in high and low velocity regimes, as well as other dynamic flames with repetitive extinction and ignition (FREI) and cyclic oscillations in a moderate velocity regime. FREI and other dynamic flames have been covered and intensively examined by other research groups [10,11]. Stable flame in the low velocity regime has been analytically predicted [8], then observed at mean flow velocity around 2–3 cm/s experimentally and examined computationally [8,9]. Although this sort of flame is an example of mild combustion, the existence of a lower limit of the weak flame regime, namely, the weakest flame (or lowest burning velocity) is left unsolved. We have already suggested the possibility of the weakest flame [12], however, the interpretation of the phenomena was insufficient so far. The aim of the present study was to investigate flame responses in a heated channel both experimentally and computationally. Special attention was paid to flame stabilities, particularly the existence of a lower limit of the weak flame regime. Thermal management by an external heat source was employed to examine such flames. At first, flame responses to mean flow velocity were investigated experimentally. Then computations with detailed chemistry and transport properties are performed to interpret experimental results and elucidate the existence of the weakest flame. 2. Experimental setup and methods In the present investigation, we employed a methodology similar to that for our microcombustion study [9], but with improvements enabling us to focus on the lower limit of weak flame. An external heat source for modeling heat recirculation with a thermally thick wall, which can also be regarded as heat loss compensation, was employed. A schematic of the apparatus is shown in Fig. 1. A premixed methane–air mixture was supplied through mass flow controllers to the channel with inner diameter of 2.0 mm, which is slightly smaller than the conventional quenching diameter. Since the convective-diffusive zone can be expected to be large under low velocity conditions, we employed a 300 mm long channel connected with a 40 m long polyethylene tube at the upstream side, and a 3 m long extension silicon tube at the downstream side.
Fig. 1. A schematic of the experimental apparatus and temperature profile along the inner surface of the tube wall.
A flat flame burner was employed for heating the center part of the channel so that a positive temperature gradient was formed on the inner surface of the tube wall in the axial direction. For better visualization of blue flame, i.e., emissions from CH, C2, etc. in the tube, a hydrogen–air mixture was used for this burner. The temperature profile along the inner surface of the tube wall was measured by a K-type thermocouple in advance. The origin of the horizontal axis was set at the upstream edge of the flat flame burner. A maximum wall temperature of 1350 K was chosen for this study so that the mixture in the tube could be autoignited. After self-ignition, flame in the channel was stabilized at a certain location in the positive temperature gradient where local flame propagation and mixture flow were balanced. A digital still camera with a CH filter was used to detect flames in the tube. Long exposure times (1–2400 s) were employed for flames in the low velocity regime because their luminosities are extremely weak. The effects of heat release by combustion inside the tube on the wall temperature profile could be neglected in the present experiments because of the large amount of heat passing through the tube from the flat flame burner to the ambient and the large heat capacity of the tube compared with the small heat of combustion in the tube. 3. Experimental results Figure 2 shows flame images in (a) high, (b) moderate and (c) low velocity regimes (refer to Fig. 3 for each regime). Figure 3 shows flame responses including extinction and ignition points of FREI (Fig. 2b) [8,9] as a function of mean flow velocity at equivalence ratio / = 1. The dashed line in Fig. 3 denotes tube wall temperature profile. In high (100–40 cm/s) and low (5–0.2 cm/s) velocity regimes, stable normal flat flames were observed (Fig. 2a and c). In the moderate velocity
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
120
1600 φ = 1.0
80
1400
Tw
1200 1000
60 (1) Normal
800
40 Extinction
20 0
Wall Temperature, Tw (K)
Mean flow velocity (cm/s)
100
Flow
Normal flame FREI Estimated points of ignition
600
(2) FREI
Ignition
400
(3) Normal
-1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0
0.2
0.4
0.6
0.8
1.0
Location (cm)
Fig. 3. Measured flame position and extinction and ignition points of FREI for variable mean flow velocity at / = 1.0. Estimated ignition locations in upper normal flame regime are also indicated.
regime (40–5 cm/s), non-stationary flames termed FREI (Fig. 2b) and other oscillating flames were observed. The present flame responses agreed well with those of previous studies [9,13]. In the high velocity regime, the stabilized locations of normal flame and estimated points of ignition, which are based on the experimental ignition temperatures [14], are plotted in Fig. 3. After self-ignition at the downstream side, an ignition kernel moves upstream, and the flame is stabilized at an upstream location. The location of stable flame shifts toward the upstream side with the decrease of mean flow velocity, and the ignition position curve exhibits a similar tendency.
In the moderate velocity regime, luminous reaction zones appeared to be broadened, as shown in Fig. 2b. A phenomenological description for FREI is as follows [8,9]. An ignition kernel starts emitting luminescence from the point of ignition at the downstream side and then the luminous zone moves toward the upstream side until it reaches to a stable point. However the flame is immediately quenched due to the large heat loss caused by the low temperature of the wall. After a time delay from quenching, re-ignition occurs at the ignition point. This cycle is repeated regularly. Note that the ignition point can be identified as the location of the visible flame kernel, and its local temperature, that is, ignition temperature, can be estimated from this location. Ignition and extinction points shift to the upstream side with the decrease of flow velocity. In the low velocity regime, stationary flames with extremely low luminosities were observed. Long exposure time was required to record those weak flame images. Figure 2c was obtained at an exposure time of 1200 s with background compensation. Although a weak flame at a mean flow velocity of 0.2 cm/s was observed, no flame was observed at a low velocity less than 0.2 cm/s, even with much longer exposure time. This indicates that the existence of a lower limit of weak flame, even with heat compensation by external heating. This finding adds to our previous results which were without a lower limit of weak flame. As shown in Fig. 3, the stabilized locations of weak flames were around x = 0, which is close to or on the extrapolated line of the ignition positions of FREI in the moderate velocity regime. This agrees very well with theoretical prediction [13] and indicates that those flames are stabilized close to ignition positions. Therefore, weak flame temperatures are considered to be close to the inner surface temperature of the tube wall in the low velocity regime. Figure 4 shows the temperature difference between the flame and the inner surface of the tube wall at the flame position as measured by a thermocouple. The temperature
30
Temperature difference (K)
Fig. 2. Flame images (/ = 1.0); (a) normal flame (Vm = 50 cm/s), (b) FREI (Vm = 20 cm/s), and (c) weak flame (Vm = 0.2 cm/s).
3077
φ φ φ φ
25
= 1.2 = 1.0 = 0.85 = 0.6
20 15 10 5 0 0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Flow velocity (cm/s)
Fig. 4. Measured temperature difference between the flame and the inner surface of the tube wall at the flame position (/ = 0.6, 0.85, 1.0 and 1.2).
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
4. Computational methods and results To elucidate experimental results, particularly for the mechanism of the weakest flame,
one-dimensional computations with detailed chemistry were conducted. A steady-state flame code [9] based on PREMIX and GRI-mech 3.0 [17] was used. The energy equation, which includes convective heat transfer between gas and wall, is as follows: K X dT _ dT 1 d kA dT þ A M qY k V k cpk dx cp dx dx cp k¼1 dx þ
K A X A 4kNu ðT w T Þ ¼ 0 x_ k hk W k cp k¼1 cp d 2
_ T, x, cp, A and k are mass In this equation, M, flow rate, gas temperature, spatial coordinate, specific heat, cross-sectional area and thermal conductivity, respectively. q, Yk, Vk, x_ k , Wk, Wk and d are density, mass fraction, diffusion velocity, production rate, specific enthalpy, molecular weight and channel diameter, respectively. The computational domain is 10 cm long. Because of the fully developed flow in a circular tube for the case between constant wall heat flux and constant wall temperature, Nusselt number of the inner wall surface was assumed to be constant (Nu = 4). Tw is the fixed wall temperature and the profile in the computation is almost the same as that of the test section shown in Fig. 1. The maximum wall temperature is 1300 K. Figure 5 shows the computed flame response to mean flow velocity at equivalence ratio / = 1. The location of the CH peak is considered to be the flame position. The dashed line with open circles in the figure shows the wall temperature profile. The origin of the horizontal axis is set at the inlet of the tube. The computational flame response showed an e-shaped curve, which has an additional lower velocity branch with the previous Sshaped curve [8,9]. Based on the detailed description of the stability analysis for the S-shaped curve in our previous reports [8,9], the present lowest velocity branch is considered to be unstable. If this assumption is true, the existence of two stable and two unstable solutions in the four regimes can
1400 100 (1) Stable 1200 10
(2) Unstable 1000
(3) Stable 800
1 (4) Unstable
600 0.1
0.01
Flame position (φ =1.0) Wall temperature
4.5
5.0
5.5
Wall temperature (K)
difference becomes smaller with the decrease of mean flow velocity and is almost zero at Vm = 0.2 cm/s, where the wall temperature is around 1225 K regardless of mixture composition. In the present experiments, we tried several types of thermocouple for the measurements. Results show that the temperature differences between the flame and the wall measured by those thermocouples were nearly equal, although the absolute values exhibited some differences. Thus, discussion based on the temperature differences between the flame and the wall surface, which are asymptote to zero in Fig. 4, is quite reasonable. It is noted that thermal quenching by intrusive thermocouple measurements is hard to occur due to the external heater in the present system since the thermocouple is inserted from the downstream side where the temperature of the leading wire of the thermocouple is maintained at high temperature and which has a small temperature gradient in it. It is also justified from the long time exposure recording in which the thermocouple was in contact with the weak flame during temperature measurements even for the weakest flame case. Based on two characteristics of weak flame, that is, (1) small temperature increase and (2) flame location close to the ignition point, the flame temperature at the lower limit of weak flame is considered to correspond to the ignition temperature of the given mixture. Although ignition and flame propagation have been considered to be separate phenomena in general, stationary propagating flame near the ignition point in weak flame was successfully established in the present experiment. Hence, ignition and flame propagation would merge with each other at this extreme condition. This fact also indicates that the general ignition temperature, which is independent of the experimental apparatus and methodology, can be expected to be identified with the position of the weakest flame by the present methodology. Ju and Xu [15] and Chao et al. [16] theoretically investigated self-sustaining flame propagation and extinction of premixed flames with heat recirculation in meso-scale channels. They also showed the weak flame regime and extinction limits of weak flame for the case of low flow velocity conditions. However, since they dealt with selfsustaining flames, heat generation by combustion was the only source of heat in their models. Thus, heat recirculation became weaker with the decrease of flame temperature. Therefore, reported extinction limits of stable weak flame were caused by thermal quenching, different from the present phenomena.
Mean flow velocity (cm/s)
3078
400
6.0
200 6.5
Location (cm)
Fig. 5. Computed flame position with mean flow velocity (/ = 1.0).
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
be inferred by the present computation. Under this assumption, the existence of the lower limit of the experimental weak flame can be interpreted as the lower limit of the third regime in Fig. 5. The upper positive gradient (regime (1)) in Fig. 5 indicates normal stable flame. This solution corresponds to the classical upper stable solution of flame response with heat loss [18]. A flame at the highest flow velocity condition was just prior to blow-off since the maximum wall temperature was located at 6.5 cm. It can be seen that the flame position shifted toward the upstream side with the decrease of the mean flow velocity in this regime, which agrees with experimental results. The upper turning point corresponds to the conventional thermal quenching [18] caused by heat loss to the wall. Below the upper turning point, unstable solutions (regime (2)), which correspond to FREI phenomena, were obtained. In the present experiment, the thermal quenching limit did not appear due to the high temperature of the wall resulting from external heating. Instead, non-stationary flames such as FREI were observed in the moderate velocity regime, which agreed with findings of previous studies [9,12]. The lower positive gradient (regime (3)) is the low velocity regime for stable weak flame. Due to the presence of the high temperature wall, flames in this low velocity regime were realized in spite of the low temperature increase. To examine the experimental lower limit of the weak flame branch, variation of the computed temperature difference between the flame and the fixed inner wall surface temperature at the flame position was investigated as shown in Fig. 6. Flame temperature is defined as the temperature of gas at the CH peak. The computed temperature difference becomes smaller with the decrease of mean flow velocity, which agrees well with findings of the present experiment. The temperature difference becomes almost zero (Tg Tw < 1 K) at a mean flow velocity of Vm = 0.1 cm/s when wall temperature is 1230 K, which quantitatively agrees with the experimental results (1225 K).
Therefore, the lower limit of stable weak flame is considered to be at the midpoint of the lowest velocity branch in Fig. 5 (regime (4)). Since the boundary between the stable and unstable branches is unresolved, stability analysis should be conducted in a future study. It should be noted that the extremely small temperature increase does not directly correspond to flame quenching in the low velocity regime; however, the conventional reaction with intense heat release no longer occurs in regime (4), even though the heat loss from the flame zone is compensated by the external heating. If thermal quenching does not occur at the lower limit of the stable weak flame, what is the possible mechanism for the extinction of weak flame? We have reported details of the weak flame structure in a previous study [9] and have shown that weak flame has a broader reaction zone due to the effect of mass diffusion. We consider the cause of such limit to be related to the dominance of diffusive mass dissipation over the convective mass transfer in the extremely low velocity regime. To examine the effect of diffusion on the lower limit of weak flame, convective and diffusive mass fluxes of OH were compared for the case of high and low velocity regimes as shown in Figs. 7 and 8. OH was chosen because it is one of the key species for the initiation of chain branching reactions as well as its high diffusivity. Each term in Figs. 7 and 8 was estimated from qYOHVOH and qYOHVm in the computation. VOH is the diffusion velocity of OH and Vm is the mean flow velocity. In each of these two figures, the shaded area indicates the reaction zone estimated by the full-width at half maximum of the CH profile. As shown in Fig. 7, mass transport is dominated by convection in the case of the high velocity regime (Vm = 134 cm/s), which is similar to conventional flame. Therefore, total mass flux is almost the same as convective mass flux. On the other hand, diffusive mass flux is dominant in the low velocity regime (Vm = 1.82 cm/s), as shown in Fig. 8. This means that the latent effect of diffusion on total mass transport increases
Tg-Tw
-3
1.5x10
-3
1.0x10
20
Flux (g/cm 2 s)
Temperature difference (K)
30
3079
10
-4
5.0x10
0.0 OH Flux (Vm = 134 cm/s)
Diffusion Convection Total mass flux
-4
-5.0x10
0 0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
Mean flow velocity (cm/s)
-3
-1.0x10
5.0
5.2
5.4
5.6
5.8
6.0
6.2
6.4
6.6
6.8
7.0
Location (cm)
Fig. 6. Computed temperature difference between the flame and the inner surface of the tube wall at the flame position (/ = 1.0).
Fig. 7. Convective mass flux, diffusive mass flux and total mass flux of OH at Vm = 134 cm/s.
3080
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081 -7
6.0x10
1.0
V = 31 cm/s
-7
4.0x10
-7
0.8
Pressure (atm)
Flux (g/cm2s)
2.0x10
0.0 -7
-2.0x10
-7
-4.0x10
-7
-6.0x10
OH Flux (Vm= 1.82 cm/s)
-6
-1.0x10
0.4 0.2
Diffusion Convection Total mass flux
-7
-8.0x10
0.6
0.0
-6
-1.2x10
5.0
5.2
5.4
5.6
5.8
6.0
6.2
6.4
6.6
6.8
4.5
7.0
5.0
due to extremely low flow velocity. Furthermore, the region of negative total mass flux in the low velocity regime is broader than that in the high velocity regime, and the absolute value of the negative total mass flux is larger than that of the positive total mass flux in the same figure. Note that the negative value means that OH moves in the upstream direction. Since the radical species represented by OH, which are vital for the chain branching reaction, diffuse out from the reaction zone in both the downstream and upstream directions, the number of collisions and the rate of production of chain carriers are both considered to be low. Thus, the flame temperature may become lower with the decrease of mean flow velocity. In order to further examine the above considerations, the low pressure condition can be given as an example of an analogous situation for the diffusion controlled condition because the reaction zone becomes broader with the decrease of pressure. Under the low pressure conditions, molecular collisions leading to chain branching reactions become less frequent due to the low density of the mixture. It is well known that the first limit of the explosion curve for the stoichiometric H2–O2 mixture [19] is caused by the lower frequency of collisions and by removal of chain carriers on the surface (wall effect) with the decrease of pressure. Although the wall surface was maintained at high temperature in the present study, there should be an essential and critical pressure condition for sustaining the chain branching reaction system [19]. Figure 9 shows computational variations of flame position with pressure at the given mean flow velocity for the case of / = 1.0. Here, the mean flow velocity is maintained at constant (31 cm/s) for different pressure levels. It is seen that the obtained flame position curve can be divided into three regimes. According to our previous experiments on combustion in a heated channel under low pressure conditions [20,21], it is considered that the upper regime indicates unstable solutions where FREI can be observed
6.0
6.5
Fig. 9. Computed flame position with pressure (/ = 1.0, Vm = 31 cm/s).
experimentally and that the middle regime corresponds to a normal flame branch. The existence of a lower regime in the present calculation is new and this implies the existence of the lower limit of stable weak flame under the low pressure conditions even though heat loss was compensated by the heated wall. Figure 10 shows variation of computed temperature difference between the flame and the inner wall surface at the flame position for different pressure levels. The temperature differences become smaller with the decrease of pressure, and it is almost zero (Tg Tw < 1 K) at the pressure levels in the lower unstable regime, which is similar to the results in Figs. 5 and 6 for the case of the lowest velocity regime. Therefore, the lower limit of stable weak flame is expected to be induced by such low pressure conditions. In the present computations, there are two different approaches. One is diffusion controlled conditions due to low flow velocity and the other is low density conditions due to low pressure. This suggests that the reaction is weakened by a lower frequency of molecular collisions even though heat loss is compensated by the heated wall. Thus, it is considered that the existence of the weakest flame observed in the present experiment is 40
Tg - Tw
Temperature difference (K)
Fig. 8. Convective mass flux, diffusive mass flux and total mass flux of OH at Vm = 1.87 cm/s.
5.5
Location (cm)
Location (cm)
30
20
10
0 0.00
0.02
0.04
0.06
0.08
Pressure (atm)
Fig. 10. Computed temperature difference between the flame and the inner surface of the tube wall at the flame position (/ = 1.0, Vm = 31 cm/s).
Y. Tsuboi et al. / Proceedings of the Combustion Institute 32 (2009) 3075–3081
induced by less frequent molecular collisions. If this is correct, the absolute limit of mild combustion with the heat recirculation process should be finally induced by this mechanism. Such information is expected to contribute to well-controlled, near-limit operation of mild combustion. 5. Conclusions Stationary and non-stationary flame responses in an externally heated meso-scale channel were investigated both experimentally and computationally. Special attention was paid to the flame stability regimes, especially with regard to the existence of the lower limit of weak flame. Although the existence of weak flame at mean flow velocity of 0.2 cm/s was confirmed experimentally, no flame could be observed at much lower velocity conditions. This suggests the existence of a lower limit of weak flame. The measured temperature difference between the flame and the tube inner surface at the flame position became zero at the above lower limit. Thus, ignition and flame propagation are thought to merge with each other under such extreme conditions. This implies that the flame temperature at the lower limit of weak flame corresponds to the ignition temperature of the given mixture. Computational flame responses exhibited an eshaped curve, i.e., the lowest velocity regime with the previous S-shaped curve. The computed temperature difference between the flame and the inner surface of the tube at the flame position became zero in the lowest velocity regime, which quantitatively supports the experimental findings. Two kinds of computational approaches, (1) diffusion controlled conditions at low velocity, and (2) low pressure conditions, suggested that the reactions were weakened because of less frequent molecular collisions at low velocity or low pressure conditions even if the wall temperature at the reaction zone was maintained at a high temperature. The limit of mild combustion with the heat recirculation process is considered to be finally induced by the same mechanism.
Acknowledgment The authors would like to thank Mr. Susumu Hasegawa for his help with experiments and Dr.
3081
Hisashi Nakamura, assistant professor of Tohoku University for fruitful discussion.
References [1] Road map for energy saving (in Japanese), 2007, available at http://www.nedo.go.jp/informations/ other/190423_1/190423_1.html. [2] S.K. Som, A. Datta, Prog. Energy Combust. Sci. 34 (3) (2008) 351–376. [3] M. Katsuki, T. Hasegawa, Proc. Combust. Inst. 27 (1998) 3135–3146. [4] H. Tsuji, A.K. Gupta, T. Hasegawa, M. Katsuki, K. Kishimoto, M. Morita, CRC Press, Boca Raton, 2003. [5] B. Dally, E. Riesmeier, N. Peters, Combust. Flame 137 (2004) 418–431. [6] A. Cavaliere, M.D. Joannon, Prog. Energy Combust. Sci. 30 (2004) 329–366. [7] K. Maruta, K. Muso, K. Takeda, T. Niioka, Proc. Combust. Inst. 28 (2000) 2117–2123. [8] K. Maruta, J.K. Park, K.C. Oh, T. Fujimori, S. Minaev, R. Fursenko, Combust. Expl. Shock Waves 40 (5) (2004) 516–523. [9] K. Maruta, T. Kataoka, N.I. Kim, S. Minaev, R. Fursenko, Proc. Combust. Inst. 30 (2005) 2429– 2436. [10] T.L. Jackson, J. Buckmaster, Z. Lu, D.C. Kyritsis, L. Massa, Proc. Combust. Inst. 31 (2007) 955–962. [11] G. Pizza, C.E. Frouzakis, J. Mantzaras, A.G. Tomboulides, K. Boulouchos, Combust. Flame 152 (3) (2008) 433–450. [12] Y. Tsuboi, T. Yokomori, K. Maruta, ASME International Mechanical Engineering Congress and Exposition, Seattle, IMECE2007-43339, 2007. [13] S. Minaev, K. Maruta, R. Fursenko, Combust. Theory Model. 11 (2007) 187–203. [14] T. Kataoka, N.I. Kim, T. Yokomori, S. Maruyama, K. Maruta, Int. Symp. Combust. (WIPP) (2004) 93. [15] Y. Ju, B. Xu, Proc. Combust. Inst. 30 (2005) 2445– 2453. [16] C.Y.H. Chao, K.S. Hui, W. Kong, P. Cheng, J.H. Wang, Int. J. Heat Mass Trans. 50 (2007) 1302–1313. [17] Information available at: http://www.me.berkeley. edu/gri_mech. [18] F.A. Williams, Combustion Theory: The Fundamental Theory of Chemically Reacting Flow Systems, Benjamin/Cummings, 1985, 275 pp. [19] I. Glassman, Combustion, third ed., Academic Press, 1996, p. 63. [20] Y. Tsuboi, T. Yokomori, S. Hasegawa, K. Maruta, Paper 182, 21st International Colloquium on the Dynamics of Explosions and Reactive Systems, Poitiers, France, July 23–27, 2007. [21] Y. Tsuboi, T. Yokomori, K. Maruta, Combust. Sci. Technol. 180 (2008) 2029–2045.