Identification of Cool and Blue Flames in Compression Ignition

Transcription

1 Y. OHTA and M. FURUTANI archivum combustionis Vol. 11 (1991) No. 1~2, pp. 43~52, Polish Academy of Sciences Identification of Cool and Blue Flames in Compression Ignition Nagoya Institute of Technology, Nagoya 466, Japan Low-temperature ignition of hydrocarbon fuels is characterized by weak, pale blue light emission. It has been difficult to distinguish the low-temperature flame; cool flames, the first onset of the lowtemperature flames, and blue flame, the second, from the other and final hot flame. The present paper shows how to identify the compression-ignition flames spectroscopically. Cool, blue, or hot flame is identified by the HCO band emission, blue and red colorations nm wave lebngh emission in the HCO Vaidya band is almost the only tool for identifying the blue flame. Visible red spectrum near 6 nm is most effective to perceive the hot-flame appearance. Autoignited hot flames turn yellow or red. The H2O emission at nm and soot formation would be plausible cause of red coloration. Carbon formation starts in compression-ignition low-temperature flame just after the cool flame has degenerated. Introduction End-gas autoignition in spark-ignition engines and ignition of the liquid-fuel jet in compression-ignition engines are the ignition processes governed by low-temperature oxidation reactions rather than by hightemperature thermal ignition reactions. It is well known that cool flame, the first low-temperature flame, and / or blue flame, the second one, appear as the precursors of the hot true ignition during its induction period. Weak blue-light emissions radiated from these low-temperature flames have been used to detect their onsets, and strong red-light emission for the detection of final hot flame. Cool flame shows a very weak pale blue color and often shows degenerate emission intensity profile pattern, compared with distinct deep blue color and straightforward pattern of the blue flame. The identification of each of them has been done leaning on the above empirical knowledge, which sometimes leads to uncertainties on the judgement. It is not easy to identify separately the two radiations because of their similar emission spectra; emission at the same wave length as the cool-flame Emeleus band is always observed in blue flame and even in hot flame. No other precise method is yet found to distinguish the blue flame from others, notwithstanding that the onset of blue flame is strongly related to engine knock. Engine knock is due to an autoignition of the low-temperature charge ahead of the normally propagating flame front. However, cool flame has not yet been observed in real engines, though blue flame can be. The existence and contribution of cool flame is still a matter of argument for the engine knock. In this paper it is tried to clarify how to identify the compression-ignition flames using their radiant emissions spectroscopically. The final hot-flame ignition preceded by low-temperature flames in the cylinder is not colored in blue but is always yellow or red, obtained even from a lean mixture with an equivalence ratio.5 which is well beyond the smoke limit and normal flammability limit. The reason of the red coloration and the possibility of soot formation in hot-flame appearance are examined here. Experimental Technique The appearance of low-temperature flames and final hot flame produced by piston compression is usually heterogeneous in the geometric sense [1, 2] and each of them overlap in time. That behavior prevents clear separation of the flames and causes uncertainties in identification. A lean n-heptane/air mixture, equivalence ratio.8, was compressed up to the temperature and pressure conditions ranging to 64 K and 1.1 to 1.4 MPa using a rapid-compression machine, and preflame light emissions were observed up to the autoignition. n-heptane was used as fuel to obtain the typical lowtemperature flames with a distinct pressure rise according to each of the flames as well as light emissions. This combination of the fuel and conditions allowed to minimize the uncertainties and to provide the least doubtful sequence of flame appearances, one after another in an order of cool, blue, and hot flame. Still more, by choosing the temperature and pressure at the end of compression strictly, it was tried to stop the ignited flames at the cool and blue flame level; without development to the hot flame stage for minimizing uncertainties. Details about the rapid-compression machine used (cylinder bore: 65 mm, stroke:14 mm) are given elsewhere [3]. Quartz windows and a strain-gauge type pressure transducer (Kyowa PE3KF) were mounted at the sidewalls and head of the cylinder respectively. Photomultipliers (RCA 1P28) were used to detect radiant 1

2 emissions. A blue and a red glass filter (Toshiba V-42; 32 ~ 51 nm, R-62; 62 nm ~) and interference filters with designated wave length listed below were placed ahead of the photomultipliers to detect emission monochromatically. 1) HCHO band: nm. A spectrum called Emeleus cool-flame band lies between 34.5 and nm, and is known to be generated by a transfer A 1 A" - X 1 A1 2) OH band: 36.4 nm, A 2 - X 2 system. 3) HCO band: nm. The well-known Vaidya hydrocarbon band spreads over 251 ~ 49 nm. 4) CH band: nm, A 2 - X 2 system. 5) C2 band: nm, (, ) group in the A X 3 system. or Sensitivity [%] Transmissivity 1 OH (36.4) HCO (329.8) Color Filter V-42 "" C (,) 2 (516.5) Photomultiplier R928 1P28 Color Filter R-62 Young Soot (68) HCHO CH C 2(,1) C 2(,2) (395.2) (431.4) (563.6) (619.1) Wave Length [nm] "" Fig. 1. Spectral sensitivities of photomultipliers and transmissivities of color glass and interference filters. Figure 1 shows the spectral sensitivities of photomultipliers and transmissivities of color glass and interference filters. The "blue color" used here is the light passed through the color filter V-42. The "red color" is a light from a very limited range overlapped between the red color filter R-62 and photomultiplier 1P28 near the wave length 6 nm. As nm: the CH band and nm: C2 band lie in the Emeleus cool-flame emission band in visible blue region, it has to be noted that the light intensities in these band do not guarantee the existence of the corresponding species. To examine whether or not carbon is formed in the compression-ignition flames, a light beam of a solid-state laser (Sharp LT22MC, 78 nm, 2 mw) was admitted into the chamber through a window and the transmitted laser light arriving at the other window was measured by a photodiode (Hamamatsu Photonics S1226-8BQ) through an interference filter which is transparent only for the laser beam. It is expected that the laser beam is attenuated by solid carbon. 68 nm wave length emission has been used to detect the soot precursor [4]. A combination of an interference filter of 68 nm and a photomultiplier R928 (Hamamatsu Photonics) was also applied for examining the soot formation (cf. Fig. 1). Results and Discussion 1) Identification of Cool and Blue Flames Figure 2 is an example of the emission trace of the nm: HCHO band, shown with those passing through blue- and red-color filters in the hot-flame eliminated case. Figure 3 is the one of nm: CH band and figure 4 is of nm: C2 band. Time scale is marked from the time the piston is arrested. Light emission records swing downwards. Red-light emission does not appear in whole process, i.e. the hot-flame onset is successfully eliminated. Without any doubt, the first substantial pressure rise after the compression is due to the cool-flame. The corresponding blue-light emission once developed at that time ceases to appear in a moment; the retardation shows the typical degenerating feature of cool-flames. Emissions of nm: CH band and nm: C2 band show almost the same behavior as the nm: HCHO band. Relation of excited formaldehyde to the cool-flame spectrum has well been recognized but not of CH *, nor C2 *. As the Emeleus continuous band covers nm and nm, there is every likelihood of its being observed through these band pass filters. 2

3 [MPa] nm (HCHO) τ 1 τ 2 Cool Flame Appears Here No Hot Flame Appears Blue Flame Appears P: 1.36 MPa T: 555 K Fig nm: HCHO band emission and blue coloration in compression-ignition cool and blue flames. n-heptane/air φ=.8. Hot flame is eliminated artificially, proved by no red coloration. [MPa] nm (CH) τ 1 τ 2.5 P: 1.35 MPa T: 561 K Fig nm (CH) band emission and blue coloration in compression-ignition cool and blue flames. n-heptane/air φ=.8. Hot flame is eliminated. The second gently-sloping pressure rise is easily acknowledged as a blue flame by emission intensities when compared with the cool-flame case. Emission from the blue flame is also found in the nm corresponding to the HCHO band. And nm (CH band) and nm (C2 band) show similar responses. It is indicated that intense CH* band is associated with blue flame but C2 * is absent from the blue-flame spectrum [5]. These features of CH or C2 band are not available for blue-flame identification. The emission from the wave length corresponding to the HCHO is found during every flame appearance. It is quite difficult to accept that HCHO exists during hot-flame occurrence. A continuum emission other than HCHO is often indicated as characteristic for low-temperature explosions [6]. The pre-blue induction time, the so-called τ2 period, is characterized by carbon-monoxide generation and accumulation. The CO spectrum is known to be continuous. It will be considered that the background spectrum is due to the CO, and the cool- and blue-flame emissions appear in the similar wave length of CO continuum emission in visible blue region. In this way the CO spectrum makes it difficult to distinguish each flame from others. 3

4 [MPa] nm (C ) 2 τ τ 1 2 P: 1.44 MPa T: 64 K Fig nm (C 2 ) band emission and blue coloration in compression-ignition cool and blue flames. n-heptane/air φ=.8. Hot flame is eliminated. [MPa] nm (HCO) τ 1 τ 2 P: 1.39 MPa T: 595 K Fig nm: HCO band emission and blue coloration in compression-ignition cool and blue flames. No response for cool-flame onset. n-heptane/air φ=.8. Hot flame is eliminated. Emission profiles of 36.4 nm: OH band and nm: HCO band are shown in Figs. 5 and 6 respectively. There is no response from the cool-flame radiation on these emission traces. This feature characterizes blueflame occurrence. It is pointed out by Sokolik that the particular radiation of the excited HCO * radicals is a measure of blue flame as it is the HCHO * radiation for cool flames [5]: HCHO + OH HCO * + H2O (1) HCO * HCO + hυ (2) 4

5 [MPa] nm (OH) τ 1 τ 2 P: 1.1 MPa T: 58 K Fig nm: OH band emission and blue coloration in compression-ignition cool and blue flames. No response for cool-flame onset. n-heptane/air φ=.8. Hot flame is eliminated. When we examine the spectral emission records shown by Pipenberg and Pahnke [7] carefully, this information on HCO * could be derived, though OH is not shown on his records. In pre-cool and cool flame reactions an amount of the OH radical is produced by the oxidation of peroxides, but it will be consumed to oxidize the fuel and peroxides in a very short time after its generation. The dominant reactions in this regions are [8]: RH + O2 R + HO2 (3) OH + RH R + H2O (4) HO2 + HO2 H2O2 + O2 (5) H2O2 becomes unstable when the temperature exceeds 8 K: H2O2 + M OH + OH + M (6) [MPa] nm (HCO) P: 1.4 MPa T: 58 K Hot Flame Appears τ 1 τ 2 τ Fig. 7. Blue and red colorations and nm: HCO band emission in typical three-stage compression ignition. n-heptane/air φ=.8. 5

6 Formaldehyde will be attacked by the enriched amount of OH as described in reactions 1 and 2, and then carbon monoxide is produced. This stage is the blue flame appearance. OH band will be first observed definitely in this stage as can be seen from Fig. 5. Thus the nm: HCO band seems to be more useful than the OH band to distinguish blue flame from the cool flame. Hot-flame elimination is very critical. Figure 7 shows a case where the hot flame appears under almost the same condition as the case the hot-flame could be eliminated. A typical three-stage ignition is seen, in which no clear distinction is available between blue and hot flames on pressure profile if red-light emission trace is not observed. As shown here the red coloration is always strongly related to steep pressure rise due to the hot-flame onset. As the HCO and OH are found also during hot flame, this red coloration in the limited range of wave length near 6 nm is the only tool to identify the hot flame for the time being. This is useful practically, but it is naturally expected to clarify what the red-light emission originates from. H2O emission at nm [9] and soot which will be mentioned in the following section would be plausible for the cause of red coloration. 2) Soot formation in compression ignition Compression-ignition hot flame is always yellow or red [e.g., 1]. It has been pointed out that soot is formed quite readily from explosions [11]. Red coloration is true even of lean mixtures far from carbon formation limit of premixed flame. As the fuel pyrolysis and early oxidation process may differ from the case of flame propagation, it might be expected that the carbon would be formed more easily in ignition processes. The mixture used was n-heptane/air, equivalence ratio.8, in which the oxygen to carbon ratio O/C=3.92. Smoke limit in premixed propagating flames of paraffins lies in rich side near O/C=2.2, in equivalence ratio from 1.41 to Figure 8 shows the attenuation of transmitted laser light and emission of young soot at 68 nm monitored in the process developing to hot flame. Attenuation is shown upward, as the light intensities were recorded here to swing downward. Emission intensity originating in flames at 78 nm corresponding to the laser beam, is examined beforehand to be quite weak when equivalence ratio is.8. The 68 nm emission slightly precedes the hot-flame onset. The response of transmitted laser light somewhat fluctuates by a schlieren effect due to the fluid motion caused by piston compression. After the final hot-flame appears, laser-light attenuation increases exponentially with a time constant of the order of 1 ms. When the exponential curve is extrapolated to earlier time, it seems to start with the hot-flame. Before this point a temporary irregular attenuation can be seen corresponding to the appearance of low-temperature flames. Laser-light attenuation does not increase so fast at the time of hot-flame onset. It increases gradually after the onset. During this process, fluid motion caused by the hot ignition would be relaxing in the combustion chamber. The laser-light attenuation under this condition could be understood to be due to the solid carbon formation, not due to the schlieren effect on the fluid motion. At the time the hot flame appears the quantity of solid carbon is still small. [MPa] nm Laser Light Laser Light P: 1.3 MPa T: 543 K Attenuation [mv]. 1 1 Fig. 8. The attenuation of transmitted laser light and emission of young soot at 68 nm in compression ignition up to hot flame. Laser light decreases exponentially with time. n-heptane/air φ=.8. 6

7 Figure 8 shows the attenuation of transmitted laser light and emission of young soot at 68 nm monitored in the process developing to hot flame. Attenuation is shown upward, as the light intensities were recorded here to swing downward. Emission intensity originating in flames at 78 nm corresponding to the laser beam, is examined beforehand to be quite weak when equivalence ratio is.8. The 68 nm emission slightly precedes the hot-flame onset. The response of transmitted laser light somewhat fluctuates by a schlieren effect due to the fluid motion caused by piston compression. After the final hot-flame appears, laser-light attenuation increases exponentially with a time constant of the order of 1 ms. When the exponential curve is extrapolated to earlier time, it seems to start with the hot-flame. Before this point a temporary irregular attenuation can be seen corresponding to the appearance of low-temperature flames. Laser-light attenuation does not increase so fast at the time of hot-flame onset. It increases gradually after the onset. During this process, fluid motion caused by the hot ignition would be relaxing in the combustion chamber. The laser-light attenuation under this condition could be understood to be due to the solid carbon formation, not due to the schlieren effect on the fluid motion. At the time the hot flame appears the quantity of solid carbon is still small. The same signal responses are monitored in the cool and blue flame periods, during and after which the hotflame is eliminated as carried out in previous section. Figure 9 is the result. It is recognized that the laser-light attenuation occurs even in the hot-flame-eliminated low-temperature flame originated from a lean mixture. It is initiated in τ2 region; at the time when the cool flame has degenerated. The attenuation and the time required to reach a certain equilibrium value decrease as the compression temperature increases and hence the pre-cool-flame period, τ1 is reduced. The amount of attenuation is smaller in hot-flame-eliminated case compared with when the the hot flame is associated. It would be quite difficult to accept that this attenuation is due to the "solid" carbon formation because of the insufficient temperature for thermal dehydrogeneration. Attenuation curve in this case is, however, very similar to the one shown in Fig. 8. It will be unreasonable to consider that the attenuation comes only from the laser-beam refraction caused by schlieren effect. It can be appreciated that the precursor of carbon is formed, though it is unknown whether the precursor is gaseous or partially liquefied. Since the precursor of carbon has been formed already in τ2 region, though it is a little, it would radiate redcolor thermal emission when heated up by the steep hot-flame heat release. However as mentioned above, the radiation at wave length 78 nm is not observed during hot ignition and in the solid carbon formation period, which would lead to the conclusion that the thermal radiation from carbon particle is not dominant for the red coloration. H2O emission at nm rises again first in line of candidates. Concluding Remarks Cool, blue, or hot flame is identified by the HCO band emission with blue and red colorations as described by the following algorithm: [Cool flame] = [Blue color] AND [Red color] AND [HCO] [Blue flame] = [Blue color] AND [Red color] AND [HCO] [Hot flame] = [Blue color] AND [Red color] AND [HCO] Overbar means the false:; the inverse of true:1 in logical algorithm. Cool- and blue-flame emissions appear superimposed over the CO continuum emission in visible blue region, which makes it difficult to distinguish each low-temperature flame from the other nm in the HCO Vaidya band is almost the only tool for identifying the blue flames. Visible red spectrum near 6 nm is most effective to perceive the hot-flame appearances separately from the accompanying low-temperature flames. Autoignited hot flame turns yellow or red, even if the original mixture is quite lean. The H2O emission at nm and thermal radiation of soot are considered as the reason of red coloration. Carbon formation starts in the compression-ignition low-temperature flame as early as t2 region; just after the cool flame has degenerated. Acknowledgment The first author would like to thank the Nitto Foundation, Aichi, Japan for its support and aid. 7

8 [MPa] Laser Light Attenuation [mv].5 P: 1.38 MPa T: 547 K [MPa] Laser Light Attenuation [mv].5 P: 1.38 MPa T: 547 K Fig. 9. The attenuation of transmitted laser light and blue coloration in hot-flame-eliminated low-temperature flames. n-heptane/air φ=.8. References [1] Ohta, Y. and Takahashi, H. Homogeneity and Propagation of Autoignited Cool and Blue Flames. Progress in Astronautics and Aeronautics; Dynamics of Flames and Reactive Systems, (ed. J. R. Bowen, N. Manson, A. K. Oppenheim, and R. I. Soloukhin), 95, (1984), pp , AIAA. [2] Ohta, Y., Kadowaki, S., Terada, K. and Takahashi, H. Effect of Turbulent Fluid Motion on Low- Temperature Autoignition of Fuel-Air Mixture under Piston Compression. Presented in 12th Int'l Colloquium on Dynamics of Explosions and Reactive Systems, Ann Arbor, MI, (1989), and to appear in Progress in Astronautics and Aeronautics; Dynamics of Reactive Systems and Explosions, (199), AIAA. [3] Ohta, Y., Hayashi, A. K., Takahashi, H. and Fujiwara, T. Consequence of Temperature and - Time History for Autoignition. Progress in Astronautics and Aeronautics; Dynamics of Reactive Systems, Flames and Configurations, (ed. J. R. Bowen, J. -C. Leyer, and R. I. Soloukhin), 15, (1986), pp , AIAA. [4] Coat, C. M. and Williams, A. Investigation of the Ignition and Combustion of n-heptane-oxygen Mixtures. 17th Symp. (Int'l) on Comb., (1979), pp , Combustion Institute. 8

9 [5] Sokolik, A. S. Self-Ignition, Flame and Detonation in Gases, Israel Program for Scientific Translation, (1963), p [6] Sheinson, R. S. and Williams, F. W. Chemiluminescence Spectra from Cool and Blue Flames: Electronically Excited Formaldehyde. Comb. and Flame, 21, (1973), pp [7] Pipenberg, K. J. and Pahnke, A. J. Spectrometric Investigations of n-heptane Preflame Reactions in a Motored Engine. Ind. Eng. Chem., 49-12, (1957), pp [8] Cox, R. A. and Cole, J. A. Chemical Aspects of the Autoignition of Hydrocarbon-Air Mixtures. Comb. and Flame, 6, (1985), pp [9] Gaydon, A. G. The Spectroscopy of Flames, Chapman and Hall, (1974), p [1] Sheppard, C. G. W. and Ibrahim, E-S. A. A. S.I. Engine Ion Probe Diagnostics. Instrumentation for Combustion and Flow in Engines, (ed. D. F. G. Durão, J. H. Whitelaw, and P. O. Witze), NATO ASI Series E-154, (1989), pp , Kluwer Academic Publishers. [11] Gaydon, A. G. and Wolhard, H. G. Flames, Their Structure, Radiation and Temperature, Chapman and Hall, (197), p