time-series spectra measurements from initial flame kernel in a...

17th International Symposium on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 07-10 July, 2014

- 1 -

Time-series Spectra Measurements from Initial Flame Kernel

in a Spark-Ignition Engine

Nobuyuki Kawahara1,*, Atsushi Inoue1 and Eiji Tomita1

1: Department of Mechanical Engineering, Okayama University, Okayama, Japan * correspondent author: [email protected]

Abstract Time-series of emission spectra from initial flame kernel were investigated in a spark-ignition engine. Spectral lines due to OH*, CH*, and C2* are related to the air-fuel ratio (AFR), and we chose to focus on the relative intensities of the spectral lines of these species. A spark plug sensor with an optical fiber was developed for SI engines, and the emission intensities of OH*, CH*, and C2* radicals were investigated for several equivalence ratios and used to estimate the AFR from the ratios of the intensity of the different spectral lines. Time-series of emission spectra from initial kernel in the practical SI engine can be obtained by using the fiber-optic spark plug and the spectrometer with ICCD camera. Detected area obtained using the developed fiber-optic spark plug is investigated using high-speed visualization image from initial flame kernel development from the spark plug. AFR could be evaluated from the chemiluminescence intensity of C2* spectral line to intensity of the CH* spectral line due to monotonic increase with the fuel quantity increase. Cycle-to-cycle variation of ignition delay could be observed by the spectral line of CH*. 1. Introduction Mixture composition of the unburned charge fundamentally influences the combustion process in spark-ignition engines, especially direct-injection spark-ignition (DISI) engines. Attention has been focused on spray-guided DISI (SG-DISI) engines to improve the thermal efficiency due to less wall-wetting fuel on the piston crown compared to current wall-guided DISI engines. The spray-guided system generates a stratified fuel concentration near the spark plugs because the fuel is aimed directly toward them. The mixture equivalence ratio near the spark plugs at the time of the spark discharge is especially important for the ignition to succeed. Moreover, evaluating the cycle-to-cycle fluctuation of combustion in direct-injection engines is desirable, as it is known that emission levels and thermal efficiency are related to fuel concentration in a spark-ignition engine. Cycle-to-cycle fluctuations of combustion in DISI engine are caused by inhomogeneity of fuel- air mixture around the spark plug, strong flow and higher turbulence inside cylinder, and much of exhaust gas recirculation(EGR) in engine cylinder. Therefore, it is important to investigate the relationship between local air/fuel ratio (AFR) around the spark plug and the initial flame kernel development from the spark plug [1]. A number of experimental methods have been developed to study the propagation characteristics of turbulent premixed flames [2-6]. A high-speed camera can be used to visualize the flame propagation and structure within an engine cylinder using an optically accessible engine setup, and laser-induced fluorescence (LIF) measurements provide information on the local and instantaneous fuel concentrations and flame structure [2,7-8]. However, both the high-speed camera and LIF measurements can only be carried out in an optically accessible engine setup, which typically has a lower engine speed than a commercially available automobile engine, and the durability is reduced because of the optically transparent materials. Moreover, the cylinder wall temperature and cavity configuration of an optically accessible engine differ from those of a vehicle engine. Chemiluminescence measurements using an optical fiber have been employed to characterize flames [2, 9-13]. The methodology is reliable, both for automobile engine applications and for providing information on engine engineering. To apply this technique to in-cylinder flame propagation, a test engine must be modified to provide optical access to the combustion chamber. Kawahara et al. applied a Cassegrain reflector to a test engine to investigate the flame structure. Chemiluminescence spectral lines have been observed from OH* at 306 nm, CH* at 431 nm, and C2* at 516 nm using an intensified charge-coupled device (ICCD) [12]. The flame front propagation speed and the flame thickness can be estimated from the CH* chemiluminescence


- 2 -

spectra [13]. In this study, time-series of emission spectra from initial flame kernel were investigated in a spark-ignition engine. Spectral lines due to OH*, CH*, and C2* are related to the AFR, and we chose to focus on the relative intensities of the spectral lines of these species. A spark plug sensor with an optical fiber was developed for SI engines [14], and the emission intensities of OH*, CH*, and C2* radicals were investigated for several equivalence ratios and used to estimate the AFR from the ratios of the intensity of the different spectral lines. Time-series of emission spectra from initial kernel in a SI engine can be obtained by using the fiber-optic spark plug and the spectrometer with ICCD camera. Detected area obtained using the developed fiber-optic spark plug is investigated using high-speed visualization image from initial flame kernel development from the spark plug. Estimation of AFR measured by developed system is determined using homogeneous mixture in practical spark-ignition engine. 2. Experimental apparatus A schematic diagram of the experimental set-up is shown in Fig. 1. A four-stroke spark-ignition engine with a single cylinder was used to test our measurement technique. The bore and stroke were 96.0 and 62.1 mm, respectively, and the compression ratio was 12.0:1. Gasoline was injected into the intake port. The AFR is measured using O2 sensor in the exhaust pipe. A change in injection duration altered the A/F ratio. The engine speed was set at 7,000 rpm. The crank angle and top dead center (TDC) from a rotary encoder were used to change the spark and port-injection timing, and were recorded by the A/D converter. The in-cylinder pressure was measured with a pressure transducer set in the cylinder head. We developed M12- and M14-thread spark plug sensors with optical fibers to provide measurements. The spark plug sensors are shown in Fig. 2. Developing instrumentation for measuring the flame propagation characteristics within automobile engines is the goal of this work. It is possible to detect the light that is emitted from the flame near the spark plug during combustion by exchanging the standard spark plug for one with an optical fiber sensor fitted. The optical set-up consisted of a sapphire window, which is highly durable, and an optical fiber to relay the signal. Only spark plugs with an M14 thread could have an optical fiber that is protected from stress, as there was insufficient space with the M12 or smaller spark plugs to insert a probe, and instead the optical fiber must be directly installed. The optical fiber (Mitsubishi cable industries: STU800G) used had a numerical aperture of 0.21. The emitted light was measured using a spectrometer equipped with an ICCD (Andor: SR303i and Andor Tech. DH340). Light emitted from the flame or reaction zone was collected by the optical fiber and transferred to the spectrometer slit. To obtain chemiluminescence spectra for wavelengths in the range 200–800 nm, a grating with 150 lines per mm was used. The wavelength resolution was 1.0 nm. A mercury-vapor lamp and a He–Ne laser were used for wavelength calibration. By gating the intensifier synchronously with the engine, it was possible to collect the emitted light at specific crank angles. This made it possible to obtain an ensemble view of the emission spectra as a function of time. Moreover, time-series of emission spectra can be obtained using “fast kinetic modes” of ICCD. During these experiments, the cylinder pressure was recorded every 0.5 crank angle degree (CAD). 3. Experimental results Developed fiber-optic spark plug used standard optical fiber and sapphire window. Therefore the flame emission within the numerical aperture of optical fiber was detected. Detected area obtained using the developed fiber-optic spark plug should be investigated using high-speed visualization image from initial flame kernel development from the spark plug. In this experiment, a specially designed single-cylinder compression–expansion machine (CEM) that could only be fired once was used as engine experiment. The CEM was equipped with quartz window and a flat cylinder head with a 78-mm bore, a stroke of 85 mm, and compression ratio of 9; we could observe only a single combustion cycle in one experimental run. The CEM had one intake valve, and the exhaust gas was drawn out by a vacuum pump through the inlet valve following the experiment. Multi-hole injector is set on cylinder head for spray-guided DISI system. Iso-


- 3 -

octane was used as fuel.

��!��

��

� ��

��

��

��

Fig. 1 Experimental set-up

Fig.2 Spark-plug sensor with optical fiber Figure 3 show the time-series of emission spectra from initial flame kernel obtained by using developed fiber-optic spark plug. Chemiluminescence was analyzed using a spectrometer equipped with an ICCD. By synchronizing the gating of the intensifier with the engine, we were able to collect light in time-series. The chemiluminescence intensities of the OH*, CH*, and C2* spectral lines were clearly observed at 306 nm, 431 nm, and 516 nm. OH* can be detected until 5.13 ms after the spark timing, on the other hand, CH* and C2* were detected until around 3.0 ms. CH* and C2* were formed inside the flame reaction zone of initial flame kernel, however, OH* can be survived inside burned gas region. In order to understand the detected area of developed fiber-optic spark plug, high-speed visualization images were captured with flame emission spectra simultaneously. Figure 4 showed flame emission spectra at 2.73 and 4.10 ms after the spark timing and visualized flame images. Red circles in images showed the detected area in consideration with numerical aperture of optical fiber. At 2.73 ms after the spark timing, the chemiluminescence intensities of the OH*, CH*, and C2* spectral lines were clearly observed at 306 nm, 431 nm, and 516 nm. Initial flame kernel arrived at the detected area of the developed fiber-optic spark plug. OH*, CH*, and C2* were formed in flame thickness of initial flame kernel. At 4.10 ms, OH* emission line and broadband emission spectra from 350 to 600 ns in wavelength were detected. At this timing, premixed flame already passes through the detected area. Broadband emission spectra from 350 to 600 ns in wavelength is attributed to CO-O


- 4 -

recombination [15], which occurs in the region of burned gas area due to final oxidation.

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

Fig. 3 Time-series of emission spectra from initial flame kernel

��

��

7.5�

5.0�

2.5�

��

4.10ms��

��

7.5�

5.0�

2.5�

2.73ms�

��

��

10mm��

Fig. 4 Simultaneous measurement of emission spectra and high-speed visualization of initial flame kernel.


- 5 -

Next, estimation of AFR measured by developed system is determined using homogeneous mixture in practical spark-ignition engine. Figure 5 shows the time-series of flame spectra of chemiluminescence in the practical SI engine. Experiments were done at an engine speed of 7,000 rpm, an A/F ratio of 14.7, and throttle position of 20 %. The spark timing was 58° before TDC (BTDC). The flame spectrum of OH*, CH* and C2* could be detected around 0.5 ms after spark timing. These spectrum intensities increased with the CA, and CH* and C2* reached a peak around 0.8msecond after spark timing. After these periods, the CO-O recombination spectrum at 300~600 nm was observed. When the flame passed the measurement location, no significant noise source was observed, which means that this method detects light only from the flame front of initial flame kernel. We did not see any background noise attributable to the spark before the light from the flame front arrived.

��

��

��

�

��

��

��

��!��

��"��#��

��

��

��

Fig. 5 Time-series of emission spectra of chemiluminescence in a spark-ignition engine

Figure 6 shows the relationship between preset AFR and intensity ratio of C2*/CH* from emission spectra of initial flame kernel. We propose to use the ratio of the intensity of the C2* spectral line to the intensity of the CH* spectral line for AFR evaluation, as the CH* and C2* spectral lines could be observed at the flame front of initial flame kernel. The available time period to observe the chemiluminescence intensity of the OH* spectral lines is therefore longer than that for CH*. The ratio of the intensity of the C2* spectral line to the intensity of the CH* spectral line decreased monotonically with the AFR, and so we can determine the AFR based on this metric by exploiting a time series of spectra. Figure 7 shows the cycle-to cycle variation of ignition delay by the comparison with time-series spectra and rate of heat release (ROHR). This experiment implemented under the same conditions, and we could observe two types of spectra. One is possible to observe radical emissions from forth timing data (0.576 ms after spark timing). The other is possible to observe radical emissions from fifth timing data (0.768 ms after spark timing). Under earlier timing that can observe radical emissions, the start timing of ROHR is also early timing. Therefore, we can evaluate the cycle-to-cycle variation of ignition delay by observing emission spectra from initial flame kernel using the developed spark-plug sensor.


- 6 -

��

��

��

��

�� '"!$&#��%�� )$��

��'��

��

� � ��

��

� %� $�%(�#�

%�!��&��

Fig.6 Relationship between preset AFR and intensity ratio of C2*/CH*.

�� !�

�) **,) ��

��

��

��

��

��

��

�

�� !��

��

��- $ &!+"��&%�

��

��

�

�&+ &*#+.��,��

��

��

�

�&+ &*#+.��,��

θ*+��/��

0+�

0+��#!&#+#'&�� $�.�

��

��

��

��

��

��

��%($#&!�+#% ��%*�

��%($#&!�+#% ��%*�

Fig. 7 Cycle-to cycle variation of ignition delay by the comparison with time-series spectra and rate of heat release (ROHR) 4. Conclusions Time-series of emission spectra from initial flame kernel were investigated in a spark-ignition engine. A spark plug sensor with an optical fiber was developed for SI engines, and the emission intensities of OH*, CH*, and C2* radicals were investigated for several AFR and used to estimate the AFR from the ratios of the intensity of the different spectral lines. These results can be summarized as follows: (1) The spark plug sensor with optical fiber can be applied to practical engine under the high engine speed of

7000 rpm and can observe flame kernel characteristics such as radicals emission of OH*, CH* and C2*. (2) The AFR could be evaluated from the chemiluminescence intensity of C2* spectral line to intensity of the

CH* spectral line due to monotonic increase with the fuel quantity increase. (3) Cycle-to-cycle variation of ignition delay could be observed by the spectral line of CH*.


- 7 -

References 1. Heywood, J.B., Internal Combustion Engine Fundamentals, McGraw-Hill, Inc., (1998). 2. H. Zhao, and N. Ladommatos, Engine Combustion Instrumentation and Diagnostics, Society of Auto

Motive Engineers, Inc.,(2001) 3. Bowditch, F.W., A New Tool for Combustion Research: A Quartz Piston Engine, SAE Paper

No.150B,(1960) 4. Nakanishi, K., Hirano, T., Inoue, T., and Ohigashi, S., The Effects of Charge Dilution on Combustion

and Its Improvement-Flame Photograph Study, SAE Paper No.750054, (1975) 5. Bates, S., Flame Imaging Studies of Cycle-by-Cycle Combustion Variation in an SI Four-Stroke Engine,

SAE Paper No.892086, (1989) 6. Aleiferis, P.G., Taylor, A.M.K.P., Whitelaw, J.H., Ishii, K, Urata, Y., Cyclic Variation of Initial Flame

Kernel Growth in a Honda VTEC-E Lean-Burn Spark-Ignition Engine, SAE Paper No. 2000-01-1207(2001)

7. Brian P, David L.R, Volker S, High-speed imaging analysis of misfires in a spray-guided direct injection engine, Proc. of Combustion Institute, Volume 33, pp.3089-3096(2011)

8. R.S. Barlow, G.-H. Wang, P. Anselmo-Filho, M.S. Sweeney, S. Hochgreb, Application of Raman/Rayleigh/LIF diagnostics in turbulent stratified flames, Proc. of Combustion Institute, Volume 32, pp.945-953(2009)

9. Witze, P., Hall, M., and Wallace, J., Fiber-Optic Instrumented Spark Plug for Measuring Early Flame Development in Spark Ignition Engines, SAE paper No.881638, (1988)

10. Spicher, U., Schmitz, G. and Kollmeiner, H., Application of a New Optical Fiber Technique for Flame Propagation Diagnostics in IC Engines, SAE Paper No.881673, (1988)

11. Y.Hardalupas, M.Orain, C.S. Panoutsos, A.M.K.P Taylor, Jimmy Olofsson, Hans Seyfried, Mattias Richter, Johan Hult, Marcus Aldén, Fredrik Hermann, Jens Klingmann, Chemiluminescence sensor for local equivalence ratio of reacting of fuel and air (FLAMESEEK), Applied Thermal Engineering, Volume 24, Issues 11-12, pp. 1619-1632(2004)

12. Kawahara, N., Tomita, E., Arimoto, S., Ikeda, Y., Nishiyama, A., In-spark-plug Sensor for Analyzing the Initial Flame and Its Structure in an SI Engine, SAE Paper No.2005-01-0644(2005)

13. Kawahara, N., Tomita, E., Arimoto, S., Ikeda, Y., Nishiyama, A., Measurement of Flame Propagation Characteristics in an SI Engine Using Micro-Local Chemiluminescence Technique, SAE Paper No. 2005-01-0645(2005)

14. Kawahara, N., Inoue, A., and Tomita, E., In-Cylinder Observations of Chmiluminescence in Turbulent Premixed Flames Using a Spark Plug Sensor with an Optical Fiber, SAE Paper No. 2013-01-2578, (2013)

15. Gaydon, A.G., The Spectroscopy of Flames, Chapman and Hall, (1974).

time-series spectra measurements from initial flame kernel in a...

Documents