- Email: [email protected]
Concept and combustion characteristics of ultra-micro combustors with premixed flame
Proceedings of the
Combustion Institute
Proceedings of the Combustion Institute 30 (2005) 2455–2462
www.elsevier.com/locate/proci
Concept and combustion characteristics of ultra-micro combustors with premixed flame S. Yuasa*, K. Oshimi, H. Nose, Y. Tennichi Department of Aerospace Engineering, Tokyo Metropolitan Institute of Technology (TMIT), Tokyo, Japan
Abstract Under micro-scale combustion influenced by quenching distance, high heat loss, shortened diffusion characteristic time, and flow laminarization, we clarified the most important issues for the combustor of ultra-micro gas turbines (UMGT), such as high space heating rate, low pressure loss, and premixed combustion. The stability behavior of single flames stabilized on top of micro tubes was examined using premixtures of air with hydrogen, methane, and propane to understand the basic combustion behavior of micro premixed flames. When micro tube inner diameters were smaller than 0.4 mm, all of the fuels exhibited critical equivalence ratios in fuel-rich regions, below which no flame formed, and above which the two stability limits of blow-off and extinction appeared at a certain equivalence ratio. The extinction limit for very fuel-rich premixtures was due to heat loss to the surrounding air and the tube. The extinction limit for more diluted fuel-rich premixtures was due to leakage of unburned fuel under the flame base. This clarification and the results of micro flame analysis led to a flat-flame burning method. For hydrogen, a prototype of a flat-flame ultra-micro combustor with a volume of 0.067 cm3 was made and tested. The flame stability region satisfied the optimum operation region of the UMGT with a 16 W output. The temperatures in the combustion chamber were sufficiently high, and the combustion efficiency achieved was more than 99.2%. For methane, the effects on flame stability of an upper wall in the combustion chamber were examined. The results can be explained by the heat loss and flame stretch. 2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Ultra-micro combustor; Flat-flame; Micro premixed flame; Gas turbine; Hydrogen
1. Introduction Recently, interest in ultra-micro gas turbine (UMGT) has been growing as an application of power micro-electromechanical systems (MEMS) technology. UMGT was first proposed with a 16 W output by a Massachusetts Institute of Technology (MIT) group [1]. To actualize UMGT, the combustors must be both space effi*
Corresponding author. Fax: +81 42 583 5119. E-mail address: [email protected] (S. Yuasa).
cient and feature high combustion efficiency to achieve a high space heating rate (SHR), low pressure loss, and low heat loss in the combustor. For example, the MIT UMGT originally required a combustion chamber volume of 0.04 cm3, a space heating rate of 3.3 · 103 MW/(m3 MPa), a combustor pressure loss under 5%, and a combustion efficiency over 99.5% [1]. The MIT group made two hydrogen/air micro combustors and examined their combustion characteristics. One was a fully premixed combustor with a slit jet configuration and a volume of 0.191 cm3 and the other was a dual-zone combustor with a primary and
1540-7489/$ - see front matter 2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2004.08.207
2456
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
dilution-zone configuration and a volume of 0.346 cm3. As a result, their combustors needed volumes 5–9 times as large as that of the first concept of the MIT UMGT to achieve better flame holding and complete combustion. Various ideas for micro-combustors have been proposed for power MEMS [2–4], and thermophysical issues related to micro-scale combustion have been discussed in previous papers [1,2,5]. When downsizing a combustor to micro-scale of the order of mm, fundamentally important scaling factors are the relative increase in quenching distance, higher heat losses due to the high surfaceto-volume ratio, shortened diffusion characteristic time of mass and heat, uniformization of temperature difference, and flow laminarization, which are ignored in conventional combustors [2,5]. Under these conditions, an ultra-micro combustor that adequately satisfies the performances required for UMGT has not yet been developed, even when hydrogen has been used. In this paper, we focused on an ultra-micro combustor for UMGT and studied feasibility related to the most suitable combustion methods for UMGT. Therefore, the objectives of this paper were to clarify the issues related to downsizing the combustors for UMGT, and then based on these issues, to propose a new burning concept for UMGT, to build a prototype ultra-micro combustor using hydrogen, and to examine its combustion characteristics experimentally. In addition, the burning concept was also examined using methane fuel to consider applicability to other fuels.
2. Need for ultra-micro combustors of UMGT To downsize a combustor for gas turbines, the following characteristic factors must be achieved: • • • •
high SHR, low pressure loss, low heat loss, and premixed combustion.
For the combustor of gas turbines, the ratio of chamber volume to air mass flow rate (m_ a ) is proportional to the diameter of the gas turbine. This means that miniaturization of gas turbines requires raising combustor SHR because of a relative decrease in combustor volume against m_ a . In general, SHR is determined as ðm_ f Dh=ðVol P c ÞÞ [m_ f is the fuel mass flow rate, Dh is the heat of combustion, Vol is the volume of combustion chamber, and Pc is the chamber pressure] and can be simplified as SHR u m_ a =ðVol P c Þ u S v=Vol u v=‘ [u is the equivalence ratio, S is the cross section area of combustor, v is the mean flow velocity through combustor, and ‘ is the length of combustor]. When u of combustors
is constant in any kind of gas turbines, SHR v/ ‘ @ 1/sb [sb is the residence time in the combustor]. This shows that sb depends only on SHR without any relation to the downsizing. In fact, the SHR and sb proposed by MIT, shown above, are the same order as those of a much larger gas turbine combustor with hydrogen fuel [6]. Potential advantages of UMGT are high energy density and high power to weight ratio, which is inversely proportional to characteristic length. Therefore, high SHR in a compact ultra-micro combustor is essential to actualizing these advantages. Reducing the pressure loss in a combustor is particularly important to realizing UMGT with low compressor pressure ratios. Conventional small gas turbines can only allow a pressure loss of 3–5%. Consequently, the Swiss roll burners are inadequate for the combustor because of large pressure loss due to their long paths [7,8]. On the other hand, the cycle calculation of the gas turbine that introduced the heat loss and transfer from the combustion chamber to the compressor and turbine showed that this heat transfer remarkably reduces the thermal efficiency, mainly due to a decrease in the efficiency of the compressor and turbine [9]. Reducing the heat transfer from the chamber may be the most important and difficult issue in developing an appropriate ultra-micro combustor for UMGT. Separated combustor configuration may be one of the clever ways to improve the heat performance, however, this issue must take first priority in future research. The diffusion characteristic time is represented as L2/D [L is the length and D is the diffusion coefficient]. When downsizing a combustor, the shortened diffusion characteristic time has a crucial effect on the rapid uniformity of concentration distributions near a flame in the combustor. In addition, the flow in ultra-micro combustors should be laminar. This prevents the achievement of rapid mixing between fuel and air, and high SHR due to the low transport coefficients related to mass and heat transfer in combustors. Therefore, premixed combustion instead of diffusion combustion should be selected in micro combustors. Consequently, in ultra-micro combustors, the realization of high SHR in a laminar flow may be the most challenging requirement among the prerequisites for UMGT.
3. Stability of ultra-micro premixed jet flames Using multiple miniature premixed flames is considered to be one of the methods to burn premixtures in a micro combustor. In studying the feasibility of this method, the combustion characteristics of a single flame stabilized on top of a micro tube were basically examined. The tubes used in this experiment ranged in size from 0.03 to
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
4 mm inner diameter with a rim thickness of 0.1 mm. The fuels tested here were hydrogen, methane, and propane with a purity of over 99.9%. The premixtures were injected vertically into still air at room temperature. Figure 1(left) shows the variation of stability limits of the miniature flames for three fuels with u as a parameter of the tube inner diameter (din). In this experiment, only the blow-off and extinction limits were measured, and the flashback limits were not measured. For all the fuels, when din 6 0.4 mm, critical u was observed under which no flame formed, even in flammable regions of premixtures. The values of the critical u increased with decreasing din as shown in Fig. 1(left), and the two stability limits of the miniature flames, that is, blow-off and extinction, appeared in the fuel-rich regions at the same u. For methane, no flame was stabilized when din was less than
2457
0.1 mm, and in the case of din 6 0.05 mm, even for hydrogen and propane no flame developed within these experimental conditions. For each fuel, when din was larger than 1 mm, the critical boundary velocity gradient curves for blow-off on the tubes of different diameters coincided with each other. Where, the boundary velocity gradients were calculated with theoretical flow relationship [10]. This suggests that the stability of the flames formed on top of tubes with din larger than 1 mm is controlled by the boundary velocity gradient at the exits of tubes. On the other hand, as shown in Fig. 1(right), when din was less than 1 mm, the families of the critical boundary gradient curves corresponding to the limits in Fig. 1(left) differ much from the curve of din = 1 mm and differ from each other. This shows that the boundary velocity gradient theory does not explain these phenomena.
Fig. 1. Stability limits of micro flames of fuel/air premixtures. (Left) Mean injection velocity. (Right) Boundary velocity gradients.
2458
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
Fig. 2. Micro flame configurations just before extinction. C3H8/air, din = 0.1 mm, Uje = 1.7 m/s, ue = 20. (A) Direct photograph, (B) Schlieren photograph with combustion, (C) Schlieren photograph without combustion.
Figure 2 shows typical photographs of a miniature propane flame with din = 0.1 mm near an extinction limit at ue = 20. The image on the left is a direct photograph of the flame, and the middle image is a schlieren photo. The image on the right is a schlieren photograph of a corresponding unburned premixture jet. The flame takes a spherical shape, suggesting that the diffusion effects are mainly important in comparison with the buoyancy effects. The distance from the injector exit to the flame base was about 0.2 mm, which was much shorter than the quenching distances of propane/air premixtures [10]. It is obvious that the schlieren image region of the flame, that is, the higher temperature region, extends to the far outside the visible flame region and beyond the injector exit. This may be evidence that a large quantity of heat was transferred by conduction from the flame to its surroundings and also the injector, resulting in substantial heat loss from the flame due to the scale effects of the high surface-to-volume ratio. It should be noted from the schlieren photograph of the unburned premixture that the flow seems to rapidly spread toward the radius just after leaving the exit of the tube. This strongly suggests that some fuel can diffuse to the surroundings before reaching the tip of the flame base due to large diffusion velocity. Figure 3 shows the variation of the flame base height he with ue for a propane/air premixture near extinction limits. Although the height was kept almost constant in the fuel-rich region, it rapidly increased when ue decreased from about 12. This tendency concerning he agrees with that of the critical injection velocities Uje at extinction limits in the same u region shown in Fig. 1C(left). The relation between Uje and he for propane was essentially similar to those for hydrogen and methane. The results related to Uje, he and flame appearance obtained here suggest that when u is high enough, the amount of fuel leakage through diffusion over the distance between the tube exit and the flame base is not so large that the injected fuel cannot be diluted by air under the lower flammability limit at the flame base [11] because he is
Fig. 3. Variation of flame base height with equivalence ratio at extinction limit. C3H8/air, din = 0.1 mm.
sufficiently low. However, flame size decreases as injection velocity decreases, which, in turn, increases the heat loss to the surroundings and the tube, finally resulting in quenching. On the other hand, in the lower u regions, the flame base is so high that a relatively large amount of fuel escapes before reaching the flame base. If the concentration of fuel at the flame base decreases below the lower flammable limit, the flame cannot be sustained. Thus, we very roughly predict the extinction condition by considering the relationship of the difference between the amount of fuel supplied from the tube and the amount of fuel that diffuses from the surface of the cylindrical fuel jet assumed to be located on the tube exit. The condition is given by the relation [12] 2 d in Dhe ; U je ¼ 2 1 uuce where uc is the equivalence ratio at the lower flammability limit. This relationship suggests that as ue decreases, Uje must increase, and when din is small, Uje also accelerates. he is in proportion to Uje. Therefore, this prediction may explain that, even for a flammable premixture, when Uj decreases at a certain u in the relative low region, flame extinction occurs not due to heat loss, but to fuel leakage. The results obtained here indicate that a single miniature premixed flame is not adequate for the premixed combustor of UMGT due to the occurrence of a broad extinction region. If a cluster of miniature premixed flames are used to prevent the diffusion of fuel to the surroundings, the shape will be similar to that of a single flat flame stabilized on many micro tubes.
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
4. Concept and combustion characteristics of flatflame ultra-micro combustors 4.1. Concept and apparatus The Reynolds number for the ultra-micro combustor proposed by the MIT group was 100–200 at a maximum cross section of the combustor, which is 3–4 orders of magnitude lower than those for conventional combustors. Thus, the flows in ultra-micro combustors would be laminar. In addition, for UMGT, current MEMS technology is limited to a two-dimensional centrifugal turbo design, causing the ultra-micro combustors to take on a flat disk shape. Under this condition, one of the burning methods to form a laminar flame in a disk-shaped combustor is a flat-flame burner with a porous plate. It is well known that, for premixtures with higher burning velocities than the incoming flow velocity through the porous plate, heat losses by conduction to the porous plate lower the burning velocity of the flame until balance is obtained between the new burning velocity with heat loss and the incoming flow velocity [13]. If the height of the flat-flame micro combustor is equal to the flame zone thickness, SHR would be remarkably high, thus meeting the level required by UMGT. Based on this concept, we made a flat-flame ultra-micro combustor using hydrogen, as shown in Fig. 4A. The basic dimensions of the combustor were essentially the same as those in the UMGT specifications proposed by the MIT group [1]. The combustor had a porous plate made of stainless steel with an average filtering accuracy of 5 lm and a nozzle made of BN (boron nitride). The height of the combustion chamber with an annular region (5 · 10.5 mm) was 1 mm and the volume was 0.067 cm3. After combustion, the burned gas exited radially outward to the atmosphere through a slit with a height of 0.3 mm between the nozzle and the quartz tube. The total pressure loss of this combustor was about 5% at maximum, which satisfied the pressure loss requirement of UMGT. The experiment was performed at atmospheric pressure and room temper-
2459
ature. The air mass flow rates m_ a were varied from 0.004 to 0.08 g/s, where a typical m_ a value was determined so as to agree with the air volume flow rate between the present experimental condition and the practical UMGT condition, working at 0.4 MPa and m_ a ¼ 0:15 g=s. For hydrogen, temperature distributions in the combustion chamber and nozzle were measured by a K-thermocouple with a wire diameter of 0.05 mm. The wires of the thermocouple were inserted in the combustion chamber through two slits on the combustion chamber wall and a junction was placed in the middle at a constant radius position of 3.35 mm. The uncertainty of temperature caused by radiation and conduction was estimated to be within 10–20 K even at 1300 K. Therefore, the correction for these losses was not considered. The combustion efficiencies were calculated from the concentrations of the unburned hydrogen in the exhausted gas collected via a quartz micro probe with an inner diameter of 0.2 mm at the exit slit and were analyzed using a hotwire-type semiconductor hydrogen detector. Water was trapped before entering the detector. Ignition was achieved with a pilot flame from the outside of the chamber. On the other hand, for UMGT using hydrocarbon fuels, the combustor height of 1 mm seems to be rather narrow because the quenching distances of hydrocarbons are wider than that of hydrogen. To determine adequate combustor dimensions and to evaluate the concept of the flat-flame ultra-micro combustor, flame stability behavior for methane/air premixed flat flames was basically examined using a similar burner to that used for hydrogen. The new burner shown in Fig. 4B had a porous plate of 20 mm in diameter made of stainless steel and no center shaft. A stagnation wall with variable height was placed over the porous plate. 4.2. Experimental results and discussion 4.2.1. Hydrogen 4.2.1.1. Flame appearance. Figure 5A shows a typical direct photograph of a hydrogen/air flat-flame
Fig. 4. Schematic of flat-flame ultra-micro combustor for (A) H2 and test combustor for (B) CH4.
2460
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
Fig. 5. H2/air flat-flame appearance in ultra-micro combustor at m_ a ¼ 0:037 g=s and u = 0.4. (A) Direct photograph with an exposure time of 180 s and (B) image intensifier photograph with an exposure time of 1/ 30 s.
formed in the ultra-micro combustor at m_ a ¼ 0:037 g=s and u = 0.4. The same flame was photographed using an image intensifier, and CCD camera system is shown in Fig. 5B. This flame was very stable, occupied the whole volume of the combustion chamber of 0.067 cm3, and burned well within this small space. It is obvious that the flame zone was almost equal to the combustion chamber volume itself, so that the flame achieved extremely high SHR of 7100 MW/(m3 MPa) at u = 0.4 in a laminar flow. Thus, this flat-flame satisfied the flat-flame burning method requirement. With increasing u, the flame remained stable and did not show any change in appearance up to u = 1.0. 4.2.1.2. Flame stability limits. Figure 6 presents the flame stability limit of the present combustor. Extinction occurred at low u conditions, but the flash-back did not occur even at u = 1.0. The stable flame region sufficiently satisfied the design operation region of UMGT combustors (assuming a 16 W power output). The minimum occurred at a m_ a of around 0.02 g/s, then rose sharply as m_ a
Fig. 6. Flame stability limits of ultra-micro combustor for H2.
decreased, and then slightly increased as m_ a increased as well. Heat losses are considered to have a critical effect on the flame stability lower limit. In general, heat loss is proportional to surface area, and the heat generation is proportional to the fuel mass flow rate ðm_ f ¼ m_ a uÞ. Thus, the ratio of heat loss to heat generation is proportional to L2 =ðm_ a uÞ; that is, a decline in m_ a increases this ratio, making the effects of the heat loss greater, and results in a severe temperature drop near the combustion chamber walls where the flame quenches. This can explain the sharp rise of the lower limit for m_ a smaller than 0.02 g/s. On the contrary, as m_ a continues to increase, blow-off in the combustion chamber could be inferred due to the limitation of the chemical reaction time in an excessively stretched flame zone, or the increased incoming velocity that exceeds the burning velocity of the premixture. 4.2.1.3. Temperature distribution and heat losses. Typical temperature distributions in the combustion chamber for u = 0.35 and 0.45 at m_ a ¼ 0:037 g=s are presented in Fig. 7. Temperatures on the surfaces of the porous plate and the nozzle showed the averages of those of the actual surfaces and those of the gas phase close to the surfaces. The porous plate was substantially heated by heat conduction from the flame. Of note, the maximum temperature positions moved slightly toward the porous plate surface at higher u. This is explained by the fact that as u increased, the flame zone becomes thinner because the burning velocity becomes higher. At u = 0.45, the maximum temperature was almost the same as the adiabatic flame temperature of hydrogen/air Tad, / = 0.45. These results confirmed the occurrence of complete combustion in the combustion chamber at u = 0.45, and the existence of substantial heat loss to both the nozzle and to the porous plate.
Fig. 7. Temperature distributions in ultra-micro combustor for H2 at m_ a ¼ 0:037 g=s.
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
At u = 0.35, however, maximum temperature was considerably lower than the adiabatic temperatures Tad, u = 0.35. Speculatively, the reason for this may be that the flame zone becomes thicker as u decreases, and then temperature drops due to heat losses to the nozzle before it reaches the maximum. This means that heat loss effects increase as u decreases and would prevent complete combustion. Regarding the heat transfer from the flame, the heat loss to the porous plate was assumed not to have fatal effects on the flame behavior, since the porous plate would exchange sufficient heat with the unburned premixture passing through. However, the heat loss to the nozzle clearly caused a temperature drop. The temperature distributions near the nozzle showed that the heat losses to the nozzle were approximately 15% of the heat generation of the hydrogen/air premixture of m_ a ¼ 0:037 g=s at u = 0.4. This assumption was quite close to the heat loss estimation from the temperature distribution in the nozzle measured by an additional experiment. These heat losses would not only worsen the combustion efficiency by decreasing the reaction rate, but would also impede self-sustaining operation of gas turbine if about 10% of heat generated was lost through conduction to the compressor and turbine [9]. 4.2.1.4. Combustion efficiency. It was found that at a constant m_ a ¼ 0:037 g=s, combustion efficiencies for u > 0.4 reached more than 99.2%. In this region, the hydrogen burned completely, despite the heat loss transferred to the nozzle, which occurred after the temperature reached the maximum. This suggests that the heat loss to the nozzle had little effect on the flat-flame combustion on the porous plate itself and only caused a temperature decrease in the exhausted gas after complete combustion. However, measurements indicated that the combustion efficiency decreased drastically for u < 0.4. This suggests that the flame zone becomes thicker as u decreases and combustion reactions would be inhibited by heat loss to the nozzle. Finally, the flame quenched due to severe heat loss, precluding a self-sustaining reaction. On the other hand, the combustion efficiency versus m_ a also showed that complete combustion was achieved at a wide range of m_ a , except near the flame stability lower limit. It should be noted that for hydrogen, the excellence of the flat-flame burning method in a micro combustor for UMGT was confirmed experimentally. 4.2.2. Methane Flame stability limits of methane/air premixture without nozzle showed the same tendency as that for the hydrogen/air premixture shown in Fig. 6, that is, at m_ a 6 0:02 g=s they increased sharply. When m_ a > 0:02 g=s, flame stability limits
2461
remained almost constant at u = 0.53. However, above m_ a ¼ 0:14 g=s, flames that formed on the porous plate surface were no longer flat-flames because the incoming flow velocity exceeded the burning velocity of the methane/air premixture at u = 1.0. To examine the nozzle effect, flame stability limits at m_ a ¼ 0:037 g=s with a stagnation wall were measured. After forming a stable flat-flame, a wall made of several materials (solid Al2O3, solid Al2O3 with Pt foil (thickness t = 0.01 mm), Pt foil and Fe foil (t = 0.03 mm)) of 20 mm in diameter was lowered toward the porous plate surface, that is, decreasing wall height hw, until the flame quenched. The intention of setting catalytic wall on the burned side was to burn unburned fuel contained in burned gas to obtain overall complete combustion. Figure 8 shows these quenching limits. When hw P 7 mm, flame stability limits with all wall materials were almost the same as those measured without a wall (approximately u = 0.53), which means that the stagnation wall has no effect on flame stabilities over this hw. On the other hand, when hw < 7 mm, the quenching limits shifted toward higher u as hw decreased, for the lean region (u < 1.0). Also hw showed a minimum value at around u = 0.9. With regard to the stagnation wall material effects, solid Al2O3 and solid Al2O3 with Pt foil showed no remarkable difference and no red-heating of the wall was observed for either material. Also, Pt and Fe foils exhibited almost the same curves, and both foils were heated until bright by the flame. With regard to catalytic effects, the Pt foil, that is, the catalytic effects on the stagnation wall surface, showed no improvement in the flame stability limits, despite expectations that the reaction of the premixture on the stagnation wall surface would allow lower wall height hw for methane. This result means that catalysis on burned side does not contribute to flame stability
Fig. 8. Variation of combustor heights at extinction limits of flat-flame for CH4/air premixtures.
2462
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
improvement where complete combustion occurs or the wall surface temperatures are low. The catalysis effect must be significant on the unburned side with high temperature porous plate surface to work. When comparing solid and foil materials, foils showed wider flame stability limits for all u range in this experiment. In addition, smaller hw with Fe foil than those with Pt foil were obtained. Interestingly, hw was almost the same as the quenching distance of the methane/air premixture at around u = 1.0 [10]. These mean that better insulation effect of Fe foil than that of Pt foil due to smaller heat conductivity (Pt = 78, Fe = 34 W/m K at 1000 K) governed the results here, rather than Pt foil catalysis effect within the present experiments. Here, the average flame strain rate determined as 2Ujp/hw s1 (Ujp is incoming flow velocity) is inversely proportional to hw at constant Ujp. In Fig. 8, strain rates for all wall materials have almost the same curves of hw with right-hand side scale. For the methane/air premixture, the flame strain rate has a maximum value at slightly leaner flame than u = 1.0. This result shows good agreement with results using a cylindrical counterflow flame configuration where stretch is caused [14], which is due to selected diffusion effect at u < 1 because the Lewis number of the methane/air premixture is less than unity. Therefore, the stagnation wall exerts two effects on flame stability: one is heat loss and the other is flame stretch. However, an order of strain rates obtained here was much smaller than the data without significant heat loss, meaning that heat loss effect of the stagnation wall may be more critical than flame stretch. For hydrogen, the flat-flame burning method is suitable and is able to achieve high SHR without any catalytic components. For a hydrocarbon fuel such as methane, when hw = 4 mm, dilution by air after burning at u = 0.9 could be effective, and SHR could be 650 MW/(m3 MPa). Therefore, this burning method without any catalysis is effective for relatively lower SHR. However, to increase SHR with such fuel, catalytic components are necessary to activate surface reaction or to improve flame stability. 5. Conclusions 1. For micro-scale combustion, the realization of high SHR in a laminar flow with low pressure loss may be the most challenging requirement among the prerequisites for UMGT. 2. Single miniature premixed flames of hydrogen, methane, and propane with air exhibited critical equivalence ratios in fuel-rich regions, below which no flame formed, and above which the two stability limits of blow-off and extinc-
tion appeared. The extinction occurred by heat loss to the surrounding air and the tube, and leakage of unburned fuel under the flame base. 3. The clarification and the results of micro flame analysis led to the concept of a flat-flame burning method. A prototype of a hydrogen-fueled flat-flame ultra-micro combustor with a volume of 0.067 cm3 for UMGT was developed. Outstanding combustion characteristics related to flame stability, space heating rate, and combustion efficiency were attained. 4. For methane, the wall height of a flat-flame combustor played a crucial role in the flame stability due to heat loss and flame stretch effects.
Acknowledgments This work was supported in part by the New Energy and Industrial Technology Development Organization (NEDO) of Japan for the Proposal-based Energy/Environment International Joint Research Program, ‘‘Development of Buttonsized Gas Generator Technology,’’ in 2001 and ‘‘Leading Research and Development of Ultra Micro Gas Turbines’’ in 2002. The authors express thanks to Mr. Wong for his cooperation in conducting the experiment.
References [1] I.A. Waitz, G. Gauba, Y.-S. Tzeng, J. Fluids Eng. 120 (1998) 109–117. [2] A.C.F. Pello, Proc. Combust. Inst. 29 (2002) 883–899. [3] W.A. Sirignano, T.K. Pham, D.D. -Rankin, Proc. Combust. Inst. 29 (2002) 925–931. [4] M. Kirtas, M. Disseau, D. Scarborough, J. Jagoda, S. Menon, Proc. Combust. Inst. 29 (2002) 917–923. [5] S. Yuasa, J. GTSJ 29 (4) (2001) 247–254 (in Japanese). [6] S. Yuasa, M. Shigeta, K. Minakawa, K. Nishida, Proc. IGTC 27 (1995) 347–352. [7] F.J. Weinberg, Proc. Combust. Inst. 15 (1975) 1–17. [8] L. Sitzki et al., Proc. ASPACC 3 (2001) 473–476. [9] K. Oshimi, S. Yuasa, J. GTSJ 31 (5) (2003) 52–57 (in Japanese). [10] B. Lewis, G. Elbe, Combustion, Flames and Explosions of Gases. Academic Press, London, UK, 1987. [11] L.M. Matta, Y. Neumeier, B. Lemon, B.T. Zinn, Proc. Combust. Inst. 29 (2002) 933–939. [12] Y. Tennichi, MasterÕs Thesis, TMIT, Japan 2002 (in Japanese). [13] R.M. Fristorm, A. Westenberg, Flame Structure. McGraw-Hill, NewYork, 1965, pp. 104–106. [14] H. Kobayashi, M. Kitano, Combust. Flame 76 (1989) 285–295.
Combustion Institute
Proceedings of the Combustion Institute 30 (2005) 2455–2462
www.elsevier.com/locate/proci
Concept and combustion characteristics of ultra-micro combustors with premixed flame S. Yuasa*, K. Oshimi, H. Nose, Y. Tennichi Department of Aerospace Engineering, Tokyo Metropolitan Institute of Technology (TMIT), Tokyo, Japan
Abstract Under micro-scale combustion influenced by quenching distance, high heat loss, shortened diffusion characteristic time, and flow laminarization, we clarified the most important issues for the combustor of ultra-micro gas turbines (UMGT), such as high space heating rate, low pressure loss, and premixed combustion. The stability behavior of single flames stabilized on top of micro tubes was examined using premixtures of air with hydrogen, methane, and propane to understand the basic combustion behavior of micro premixed flames. When micro tube inner diameters were smaller than 0.4 mm, all of the fuels exhibited critical equivalence ratios in fuel-rich regions, below which no flame formed, and above which the two stability limits of blow-off and extinction appeared at a certain equivalence ratio. The extinction limit for very fuel-rich premixtures was due to heat loss to the surrounding air and the tube. The extinction limit for more diluted fuel-rich premixtures was due to leakage of unburned fuel under the flame base. This clarification and the results of micro flame analysis led to a flat-flame burning method. For hydrogen, a prototype of a flat-flame ultra-micro combustor with a volume of 0.067 cm3 was made and tested. The flame stability region satisfied the optimum operation region of the UMGT with a 16 W output. The temperatures in the combustion chamber were sufficiently high, and the combustion efficiency achieved was more than 99.2%. For methane, the effects on flame stability of an upper wall in the combustion chamber were examined. The results can be explained by the heat loss and flame stretch. 2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Ultra-micro combustor; Flat-flame; Micro premixed flame; Gas turbine; Hydrogen
1. Introduction Recently, interest in ultra-micro gas turbine (UMGT) has been growing as an application of power micro-electromechanical systems (MEMS) technology. UMGT was first proposed with a 16 W output by a Massachusetts Institute of Technology (MIT) group [1]. To actualize UMGT, the combustors must be both space effi*
Corresponding author. Fax: +81 42 583 5119. E-mail address: [email protected] (S. Yuasa).
cient and feature high combustion efficiency to achieve a high space heating rate (SHR), low pressure loss, and low heat loss in the combustor. For example, the MIT UMGT originally required a combustion chamber volume of 0.04 cm3, a space heating rate of 3.3 · 103 MW/(m3 MPa), a combustor pressure loss under 5%, and a combustion efficiency over 99.5% [1]. The MIT group made two hydrogen/air micro combustors and examined their combustion characteristics. One was a fully premixed combustor with a slit jet configuration and a volume of 0.191 cm3 and the other was a dual-zone combustor with a primary and
1540-7489/$ - see front matter 2004 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.proci.2004.08.207
2456
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
dilution-zone configuration and a volume of 0.346 cm3. As a result, their combustors needed volumes 5–9 times as large as that of the first concept of the MIT UMGT to achieve better flame holding and complete combustion. Various ideas for micro-combustors have been proposed for power MEMS [2–4], and thermophysical issues related to micro-scale combustion have been discussed in previous papers [1,2,5]. When downsizing a combustor to micro-scale of the order of mm, fundamentally important scaling factors are the relative increase in quenching distance, higher heat losses due to the high surfaceto-volume ratio, shortened diffusion characteristic time of mass and heat, uniformization of temperature difference, and flow laminarization, which are ignored in conventional combustors [2,5]. Under these conditions, an ultra-micro combustor that adequately satisfies the performances required for UMGT has not yet been developed, even when hydrogen has been used. In this paper, we focused on an ultra-micro combustor for UMGT and studied feasibility related to the most suitable combustion methods for UMGT. Therefore, the objectives of this paper were to clarify the issues related to downsizing the combustors for UMGT, and then based on these issues, to propose a new burning concept for UMGT, to build a prototype ultra-micro combustor using hydrogen, and to examine its combustion characteristics experimentally. In addition, the burning concept was also examined using methane fuel to consider applicability to other fuels.
2. Need for ultra-micro combustors of UMGT To downsize a combustor for gas turbines, the following characteristic factors must be achieved: • • • •
high SHR, low pressure loss, low heat loss, and premixed combustion.
For the combustor of gas turbines, the ratio of chamber volume to air mass flow rate (m_ a ) is proportional to the diameter of the gas turbine. This means that miniaturization of gas turbines requires raising combustor SHR because of a relative decrease in combustor volume against m_ a . In general, SHR is determined as ðm_ f Dh=ðVol P c ÞÞ [m_ f is the fuel mass flow rate, Dh is the heat of combustion, Vol is the volume of combustion chamber, and Pc is the chamber pressure] and can be simplified as SHR u m_ a =ðVol P c Þ u S v=Vol u v=‘ [u is the equivalence ratio, S is the cross section area of combustor, v is the mean flow velocity through combustor, and ‘ is the length of combustor]. When u of combustors
is constant in any kind of gas turbines, SHR v/ ‘ @ 1/sb [sb is the residence time in the combustor]. This shows that sb depends only on SHR without any relation to the downsizing. In fact, the SHR and sb proposed by MIT, shown above, are the same order as those of a much larger gas turbine combustor with hydrogen fuel [6]. Potential advantages of UMGT are high energy density and high power to weight ratio, which is inversely proportional to characteristic length. Therefore, high SHR in a compact ultra-micro combustor is essential to actualizing these advantages. Reducing the pressure loss in a combustor is particularly important to realizing UMGT with low compressor pressure ratios. Conventional small gas turbines can only allow a pressure loss of 3–5%. Consequently, the Swiss roll burners are inadequate for the combustor because of large pressure loss due to their long paths [7,8]. On the other hand, the cycle calculation of the gas turbine that introduced the heat loss and transfer from the combustion chamber to the compressor and turbine showed that this heat transfer remarkably reduces the thermal efficiency, mainly due to a decrease in the efficiency of the compressor and turbine [9]. Reducing the heat transfer from the chamber may be the most important and difficult issue in developing an appropriate ultra-micro combustor for UMGT. Separated combustor configuration may be one of the clever ways to improve the heat performance, however, this issue must take first priority in future research. The diffusion characteristic time is represented as L2/D [L is the length and D is the diffusion coefficient]. When downsizing a combustor, the shortened diffusion characteristic time has a crucial effect on the rapid uniformity of concentration distributions near a flame in the combustor. In addition, the flow in ultra-micro combustors should be laminar. This prevents the achievement of rapid mixing between fuel and air, and high SHR due to the low transport coefficients related to mass and heat transfer in combustors. Therefore, premixed combustion instead of diffusion combustion should be selected in micro combustors. Consequently, in ultra-micro combustors, the realization of high SHR in a laminar flow may be the most challenging requirement among the prerequisites for UMGT.
3. Stability of ultra-micro premixed jet flames Using multiple miniature premixed flames is considered to be one of the methods to burn premixtures in a micro combustor. In studying the feasibility of this method, the combustion characteristics of a single flame stabilized on top of a micro tube were basically examined. The tubes used in this experiment ranged in size from 0.03 to
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
4 mm inner diameter with a rim thickness of 0.1 mm. The fuels tested here were hydrogen, methane, and propane with a purity of over 99.9%. The premixtures were injected vertically into still air at room temperature. Figure 1(left) shows the variation of stability limits of the miniature flames for three fuels with u as a parameter of the tube inner diameter (din). In this experiment, only the blow-off and extinction limits were measured, and the flashback limits were not measured. For all the fuels, when din 6 0.4 mm, critical u was observed under which no flame formed, even in flammable regions of premixtures. The values of the critical u increased with decreasing din as shown in Fig. 1(left), and the two stability limits of the miniature flames, that is, blow-off and extinction, appeared in the fuel-rich regions at the same u. For methane, no flame was stabilized when din was less than
2457
0.1 mm, and in the case of din 6 0.05 mm, even for hydrogen and propane no flame developed within these experimental conditions. For each fuel, when din was larger than 1 mm, the critical boundary velocity gradient curves for blow-off on the tubes of different diameters coincided with each other. Where, the boundary velocity gradients were calculated with theoretical flow relationship [10]. This suggests that the stability of the flames formed on top of tubes with din larger than 1 mm is controlled by the boundary velocity gradient at the exits of tubes. On the other hand, as shown in Fig. 1(right), when din was less than 1 mm, the families of the critical boundary gradient curves corresponding to the limits in Fig. 1(left) differ much from the curve of din = 1 mm and differ from each other. This shows that the boundary velocity gradient theory does not explain these phenomena.
Fig. 1. Stability limits of micro flames of fuel/air premixtures. (Left) Mean injection velocity. (Right) Boundary velocity gradients.
2458
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
Fig. 2. Micro flame configurations just before extinction. C3H8/air, din = 0.1 mm, Uje = 1.7 m/s, ue = 20. (A) Direct photograph, (B) Schlieren photograph with combustion, (C) Schlieren photograph without combustion.
Figure 2 shows typical photographs of a miniature propane flame with din = 0.1 mm near an extinction limit at ue = 20. The image on the left is a direct photograph of the flame, and the middle image is a schlieren photo. The image on the right is a schlieren photograph of a corresponding unburned premixture jet. The flame takes a spherical shape, suggesting that the diffusion effects are mainly important in comparison with the buoyancy effects. The distance from the injector exit to the flame base was about 0.2 mm, which was much shorter than the quenching distances of propane/air premixtures [10]. It is obvious that the schlieren image region of the flame, that is, the higher temperature region, extends to the far outside the visible flame region and beyond the injector exit. This may be evidence that a large quantity of heat was transferred by conduction from the flame to its surroundings and also the injector, resulting in substantial heat loss from the flame due to the scale effects of the high surface-to-volume ratio. It should be noted from the schlieren photograph of the unburned premixture that the flow seems to rapidly spread toward the radius just after leaving the exit of the tube. This strongly suggests that some fuel can diffuse to the surroundings before reaching the tip of the flame base due to large diffusion velocity. Figure 3 shows the variation of the flame base height he with ue for a propane/air premixture near extinction limits. Although the height was kept almost constant in the fuel-rich region, it rapidly increased when ue decreased from about 12. This tendency concerning he agrees with that of the critical injection velocities Uje at extinction limits in the same u region shown in Fig. 1C(left). The relation between Uje and he for propane was essentially similar to those for hydrogen and methane. The results related to Uje, he and flame appearance obtained here suggest that when u is high enough, the amount of fuel leakage through diffusion over the distance between the tube exit and the flame base is not so large that the injected fuel cannot be diluted by air under the lower flammability limit at the flame base [11] because he is
Fig. 3. Variation of flame base height with equivalence ratio at extinction limit. C3H8/air, din = 0.1 mm.
sufficiently low. However, flame size decreases as injection velocity decreases, which, in turn, increases the heat loss to the surroundings and the tube, finally resulting in quenching. On the other hand, in the lower u regions, the flame base is so high that a relatively large amount of fuel escapes before reaching the flame base. If the concentration of fuel at the flame base decreases below the lower flammable limit, the flame cannot be sustained. Thus, we very roughly predict the extinction condition by considering the relationship of the difference between the amount of fuel supplied from the tube and the amount of fuel that diffuses from the surface of the cylindrical fuel jet assumed to be located on the tube exit. The condition is given by the relation [12] 2 d in Dhe ; U je ¼ 2 1 uuce where uc is the equivalence ratio at the lower flammability limit. This relationship suggests that as ue decreases, Uje must increase, and when din is small, Uje also accelerates. he is in proportion to Uje. Therefore, this prediction may explain that, even for a flammable premixture, when Uj decreases at a certain u in the relative low region, flame extinction occurs not due to heat loss, but to fuel leakage. The results obtained here indicate that a single miniature premixed flame is not adequate for the premixed combustor of UMGT due to the occurrence of a broad extinction region. If a cluster of miniature premixed flames are used to prevent the diffusion of fuel to the surroundings, the shape will be similar to that of a single flat flame stabilized on many micro tubes.
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
4. Concept and combustion characteristics of flatflame ultra-micro combustors 4.1. Concept and apparatus The Reynolds number for the ultra-micro combustor proposed by the MIT group was 100–200 at a maximum cross section of the combustor, which is 3–4 orders of magnitude lower than those for conventional combustors. Thus, the flows in ultra-micro combustors would be laminar. In addition, for UMGT, current MEMS technology is limited to a two-dimensional centrifugal turbo design, causing the ultra-micro combustors to take on a flat disk shape. Under this condition, one of the burning methods to form a laminar flame in a disk-shaped combustor is a flat-flame burner with a porous plate. It is well known that, for premixtures with higher burning velocities than the incoming flow velocity through the porous plate, heat losses by conduction to the porous plate lower the burning velocity of the flame until balance is obtained between the new burning velocity with heat loss and the incoming flow velocity [13]. If the height of the flat-flame micro combustor is equal to the flame zone thickness, SHR would be remarkably high, thus meeting the level required by UMGT. Based on this concept, we made a flat-flame ultra-micro combustor using hydrogen, as shown in Fig. 4A. The basic dimensions of the combustor were essentially the same as those in the UMGT specifications proposed by the MIT group [1]. The combustor had a porous plate made of stainless steel with an average filtering accuracy of 5 lm and a nozzle made of BN (boron nitride). The height of the combustion chamber with an annular region (5 · 10.5 mm) was 1 mm and the volume was 0.067 cm3. After combustion, the burned gas exited radially outward to the atmosphere through a slit with a height of 0.3 mm between the nozzle and the quartz tube. The total pressure loss of this combustor was about 5% at maximum, which satisfied the pressure loss requirement of UMGT. The experiment was performed at atmospheric pressure and room temper-
2459
ature. The air mass flow rates m_ a were varied from 0.004 to 0.08 g/s, where a typical m_ a value was determined so as to agree with the air volume flow rate between the present experimental condition and the practical UMGT condition, working at 0.4 MPa and m_ a ¼ 0:15 g=s. For hydrogen, temperature distributions in the combustion chamber and nozzle were measured by a K-thermocouple with a wire diameter of 0.05 mm. The wires of the thermocouple were inserted in the combustion chamber through two slits on the combustion chamber wall and a junction was placed in the middle at a constant radius position of 3.35 mm. The uncertainty of temperature caused by radiation and conduction was estimated to be within 10–20 K even at 1300 K. Therefore, the correction for these losses was not considered. The combustion efficiencies were calculated from the concentrations of the unburned hydrogen in the exhausted gas collected via a quartz micro probe with an inner diameter of 0.2 mm at the exit slit and were analyzed using a hotwire-type semiconductor hydrogen detector. Water was trapped before entering the detector. Ignition was achieved with a pilot flame from the outside of the chamber. On the other hand, for UMGT using hydrocarbon fuels, the combustor height of 1 mm seems to be rather narrow because the quenching distances of hydrocarbons are wider than that of hydrogen. To determine adequate combustor dimensions and to evaluate the concept of the flat-flame ultra-micro combustor, flame stability behavior for methane/air premixed flat flames was basically examined using a similar burner to that used for hydrogen. The new burner shown in Fig. 4B had a porous plate of 20 mm in diameter made of stainless steel and no center shaft. A stagnation wall with variable height was placed over the porous plate. 4.2. Experimental results and discussion 4.2.1. Hydrogen 4.2.1.1. Flame appearance. Figure 5A shows a typical direct photograph of a hydrogen/air flat-flame
Fig. 4. Schematic of flat-flame ultra-micro combustor for (A) H2 and test combustor for (B) CH4.
2460
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
Fig. 5. H2/air flat-flame appearance in ultra-micro combustor at m_ a ¼ 0:037 g=s and u = 0.4. (A) Direct photograph with an exposure time of 180 s and (B) image intensifier photograph with an exposure time of 1/ 30 s.
formed in the ultra-micro combustor at m_ a ¼ 0:037 g=s and u = 0.4. The same flame was photographed using an image intensifier, and CCD camera system is shown in Fig. 5B. This flame was very stable, occupied the whole volume of the combustion chamber of 0.067 cm3, and burned well within this small space. It is obvious that the flame zone was almost equal to the combustion chamber volume itself, so that the flame achieved extremely high SHR of 7100 MW/(m3 MPa) at u = 0.4 in a laminar flow. Thus, this flat-flame satisfied the flat-flame burning method requirement. With increasing u, the flame remained stable and did not show any change in appearance up to u = 1.0. 4.2.1.2. Flame stability limits. Figure 6 presents the flame stability limit of the present combustor. Extinction occurred at low u conditions, but the flash-back did not occur even at u = 1.0. The stable flame region sufficiently satisfied the design operation region of UMGT combustors (assuming a 16 W power output). The minimum occurred at a m_ a of around 0.02 g/s, then rose sharply as m_ a
Fig. 6. Flame stability limits of ultra-micro combustor for H2.
decreased, and then slightly increased as m_ a increased as well. Heat losses are considered to have a critical effect on the flame stability lower limit. In general, heat loss is proportional to surface area, and the heat generation is proportional to the fuel mass flow rate ðm_ f ¼ m_ a uÞ. Thus, the ratio of heat loss to heat generation is proportional to L2 =ðm_ a uÞ; that is, a decline in m_ a increases this ratio, making the effects of the heat loss greater, and results in a severe temperature drop near the combustion chamber walls where the flame quenches. This can explain the sharp rise of the lower limit for m_ a smaller than 0.02 g/s. On the contrary, as m_ a continues to increase, blow-off in the combustion chamber could be inferred due to the limitation of the chemical reaction time in an excessively stretched flame zone, or the increased incoming velocity that exceeds the burning velocity of the premixture. 4.2.1.3. Temperature distribution and heat losses. Typical temperature distributions in the combustion chamber for u = 0.35 and 0.45 at m_ a ¼ 0:037 g=s are presented in Fig. 7. Temperatures on the surfaces of the porous plate and the nozzle showed the averages of those of the actual surfaces and those of the gas phase close to the surfaces. The porous plate was substantially heated by heat conduction from the flame. Of note, the maximum temperature positions moved slightly toward the porous plate surface at higher u. This is explained by the fact that as u increased, the flame zone becomes thinner because the burning velocity becomes higher. At u = 0.45, the maximum temperature was almost the same as the adiabatic flame temperature of hydrogen/air Tad, / = 0.45. These results confirmed the occurrence of complete combustion in the combustion chamber at u = 0.45, and the existence of substantial heat loss to both the nozzle and to the porous plate.
Fig. 7. Temperature distributions in ultra-micro combustor for H2 at m_ a ¼ 0:037 g=s.
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
At u = 0.35, however, maximum temperature was considerably lower than the adiabatic temperatures Tad, u = 0.35. Speculatively, the reason for this may be that the flame zone becomes thicker as u decreases, and then temperature drops due to heat losses to the nozzle before it reaches the maximum. This means that heat loss effects increase as u decreases and would prevent complete combustion. Regarding the heat transfer from the flame, the heat loss to the porous plate was assumed not to have fatal effects on the flame behavior, since the porous plate would exchange sufficient heat with the unburned premixture passing through. However, the heat loss to the nozzle clearly caused a temperature drop. The temperature distributions near the nozzle showed that the heat losses to the nozzle were approximately 15% of the heat generation of the hydrogen/air premixture of m_ a ¼ 0:037 g=s at u = 0.4. This assumption was quite close to the heat loss estimation from the temperature distribution in the nozzle measured by an additional experiment. These heat losses would not only worsen the combustion efficiency by decreasing the reaction rate, but would also impede self-sustaining operation of gas turbine if about 10% of heat generated was lost through conduction to the compressor and turbine [9]. 4.2.1.4. Combustion efficiency. It was found that at a constant m_ a ¼ 0:037 g=s, combustion efficiencies for u > 0.4 reached more than 99.2%. In this region, the hydrogen burned completely, despite the heat loss transferred to the nozzle, which occurred after the temperature reached the maximum. This suggests that the heat loss to the nozzle had little effect on the flat-flame combustion on the porous plate itself and only caused a temperature decrease in the exhausted gas after complete combustion. However, measurements indicated that the combustion efficiency decreased drastically for u < 0.4. This suggests that the flame zone becomes thicker as u decreases and combustion reactions would be inhibited by heat loss to the nozzle. Finally, the flame quenched due to severe heat loss, precluding a self-sustaining reaction. On the other hand, the combustion efficiency versus m_ a also showed that complete combustion was achieved at a wide range of m_ a , except near the flame stability lower limit. It should be noted that for hydrogen, the excellence of the flat-flame burning method in a micro combustor for UMGT was confirmed experimentally. 4.2.2. Methane Flame stability limits of methane/air premixture without nozzle showed the same tendency as that for the hydrogen/air premixture shown in Fig. 6, that is, at m_ a 6 0:02 g=s they increased sharply. When m_ a > 0:02 g=s, flame stability limits
2461
remained almost constant at u = 0.53. However, above m_ a ¼ 0:14 g=s, flames that formed on the porous plate surface were no longer flat-flames because the incoming flow velocity exceeded the burning velocity of the methane/air premixture at u = 1.0. To examine the nozzle effect, flame stability limits at m_ a ¼ 0:037 g=s with a stagnation wall were measured. After forming a stable flat-flame, a wall made of several materials (solid Al2O3, solid Al2O3 with Pt foil (thickness t = 0.01 mm), Pt foil and Fe foil (t = 0.03 mm)) of 20 mm in diameter was lowered toward the porous plate surface, that is, decreasing wall height hw, until the flame quenched. The intention of setting catalytic wall on the burned side was to burn unburned fuel contained in burned gas to obtain overall complete combustion. Figure 8 shows these quenching limits. When hw P 7 mm, flame stability limits with all wall materials were almost the same as those measured without a wall (approximately u = 0.53), which means that the stagnation wall has no effect on flame stabilities over this hw. On the other hand, when hw < 7 mm, the quenching limits shifted toward higher u as hw decreased, for the lean region (u < 1.0). Also hw showed a minimum value at around u = 0.9. With regard to the stagnation wall material effects, solid Al2O3 and solid Al2O3 with Pt foil showed no remarkable difference and no red-heating of the wall was observed for either material. Also, Pt and Fe foils exhibited almost the same curves, and both foils were heated until bright by the flame. With regard to catalytic effects, the Pt foil, that is, the catalytic effects on the stagnation wall surface, showed no improvement in the flame stability limits, despite expectations that the reaction of the premixture on the stagnation wall surface would allow lower wall height hw for methane. This result means that catalysis on burned side does not contribute to flame stability
Fig. 8. Variation of combustor heights at extinction limits of flat-flame for CH4/air premixtures.
2462
S. Yuasa et al. / Proceedings of the Combustion Institute 30 (2005) 2455–2462
improvement where complete combustion occurs or the wall surface temperatures are low. The catalysis effect must be significant on the unburned side with high temperature porous plate surface to work. When comparing solid and foil materials, foils showed wider flame stability limits for all u range in this experiment. In addition, smaller hw with Fe foil than those with Pt foil were obtained. Interestingly, hw was almost the same as the quenching distance of the methane/air premixture at around u = 1.0 [10]. These mean that better insulation effect of Fe foil than that of Pt foil due to smaller heat conductivity (Pt = 78, Fe = 34 W/m K at 1000 K) governed the results here, rather than Pt foil catalysis effect within the present experiments. Here, the average flame strain rate determined as 2Ujp/hw s1 (Ujp is incoming flow velocity) is inversely proportional to hw at constant Ujp. In Fig. 8, strain rates for all wall materials have almost the same curves of hw with right-hand side scale. For the methane/air premixture, the flame strain rate has a maximum value at slightly leaner flame than u = 1.0. This result shows good agreement with results using a cylindrical counterflow flame configuration where stretch is caused [14], which is due to selected diffusion effect at u < 1 because the Lewis number of the methane/air premixture is less than unity. Therefore, the stagnation wall exerts two effects on flame stability: one is heat loss and the other is flame stretch. However, an order of strain rates obtained here was much smaller than the data without significant heat loss, meaning that heat loss effect of the stagnation wall may be more critical than flame stretch. For hydrogen, the flat-flame burning method is suitable and is able to achieve high SHR without any catalytic components. For a hydrocarbon fuel such as methane, when hw = 4 mm, dilution by air after burning at u = 0.9 could be effective, and SHR could be 650 MW/(m3 MPa). Therefore, this burning method without any catalysis is effective for relatively lower SHR. However, to increase SHR with such fuel, catalytic components are necessary to activate surface reaction or to improve flame stability. 5. Conclusions 1. For micro-scale combustion, the realization of high SHR in a laminar flow with low pressure loss may be the most challenging requirement among the prerequisites for UMGT. 2. Single miniature premixed flames of hydrogen, methane, and propane with air exhibited critical equivalence ratios in fuel-rich regions, below which no flame formed, and above which the two stability limits of blow-off and extinc-
tion appeared. The extinction occurred by heat loss to the surrounding air and the tube, and leakage of unburned fuel under the flame base. 3. The clarification and the results of micro flame analysis led to the concept of a flat-flame burning method. A prototype of a hydrogen-fueled flat-flame ultra-micro combustor with a volume of 0.067 cm3 for UMGT was developed. Outstanding combustion characteristics related to flame stability, space heating rate, and combustion efficiency were attained. 4. For methane, the wall height of a flat-flame combustor played a crucial role in the flame stability due to heat loss and flame stretch effects.
Acknowledgments This work was supported in part by the New Energy and Industrial Technology Development Organization (NEDO) of Japan for the Proposal-based Energy/Environment International Joint Research Program, ‘‘Development of Buttonsized Gas Generator Technology,’’ in 2001 and ‘‘Leading Research and Development of Ultra Micro Gas Turbines’’ in 2002. The authors express thanks to Mr. Wong for his cooperation in conducting the experiment.
References [1] I.A. Waitz, G. Gauba, Y.-S. Tzeng, J. Fluids Eng. 120 (1998) 109–117. [2] A.C.F. Pello, Proc. Combust. Inst. 29 (2002) 883–899. [3] W.A. Sirignano, T.K. Pham, D.D. -Rankin, Proc. Combust. Inst. 29 (2002) 925–931. [4] M. Kirtas, M. Disseau, D. Scarborough, J. Jagoda, S. Menon, Proc. Combust. Inst. 29 (2002) 917–923. [5] S. Yuasa, J. GTSJ 29 (4) (2001) 247–254 (in Japanese). [6] S. Yuasa, M. Shigeta, K. Minakawa, K. Nishida, Proc. IGTC 27 (1995) 347–352. [7] F.J. Weinberg, Proc. Combust. Inst. 15 (1975) 1–17. [8] L. Sitzki et al., Proc. ASPACC 3 (2001) 473–476. [9] K. Oshimi, S. Yuasa, J. GTSJ 31 (5) (2003) 52–57 (in Japanese). [10] B. Lewis, G. Elbe, Combustion, Flames and Explosions of Gases. Academic Press, London, UK, 1987. [11] L.M. Matta, Y. Neumeier, B. Lemon, B.T. Zinn, Proc. Combust. Inst. 29 (2002) 933–939. [12] Y. Tennichi, MasterÕs Thesis, TMIT, Japan 2002 (in Japanese). [13] R.M. Fristorm, A. Westenberg, Flame Structure. McGraw-Hill, NewYork, 1965, pp. 104–106. [14] H. Kobayashi, M. Kitano, Combust. Flame 76 (1989) 285–295.