Investigations of the formation and oxidation of soot inside a direct injection spark ignition engine

M. Rossbach2, A. Velji1, U. Wagner1, U. Spicher1, R. Suntz2, H. Bockhorn2

1: Institute for Reciprocating Engines, Karlsruhe Institute of Technology, Germany 2: Institute for Chemical Technology and Polymer Chemistry, Karlsruhe Institute of Technology, Germany

Abstract: In this work the formation and oxidation of soot inside a direct injection spark ignition engine at different injection and ignition timing was investigated. In order to get two-dimensional data during the expansion stroke, the RAYLIX-technique was applied in the combustion chamber of an optical accessible single cylinder engine. This technique is based on the quasi-simultaneous detection of Rayleigh-scattering, laser-induced incandescence (LII) and extinction which enables simultaneous measurements of temporally and spatially resolved soot concentrations, mean particle radii and number densities. These investigations show that in our test engine the most important source for soot formation during combustion are pool fires, i.e. liquid fuel burning on the top of the piston. These pool fires were observed under almost all experimental conditions.

Keywords: soot, direct injection spark ignition engine, LII, Rayleigh scattering

1. Introduction

Modern direct injection spark ignition engines, especially when they are operated in stratified mode, have to deal with the problem of soot emission in the exhaust gas, because fuel rich combustion takes place in the vicinity of the fuel injection jet, leading to soot formation. Later, a fraction of this soot is emitted by Diesel- as well as direct injection spark ignition engines in terms of fine and ultra-fine particles. Fine dust, having a particle diameter less than 10 micrometers (PM 10) can partly penetrate into the lung because the filtering function of the nasal-pharyngeal space is not adequate for these particles. Aggravated allergy symptoms, the increase in asthmatic attacks and even cardiovascular illnesses have been quoted as the possible consequences of the increasing concentrations of fine dust in the air we breathe [1-4]. Adverse health effects are apparently not caused by the mass, but rather primarily by the surface of the particles. Particles that are formed from combustion processes are apparently more relevant than, for example, ground particles or tire wear particles. The disadvantage of particle emission is well known from Diesel engines. As a consequence, a variety of

technical solutions to reduce soot emissions are developed. Measures targeting fuel additives generally have the disadvantage that they simultaneously increase the nitrogen oxide emissions. External measures like exhaust gas treatment systems e.g. particle filters, considerably reduce particle emissions but are associated with difficulties in their application (e.g. regeneration of the particle filters) and generally also lead to an increase in fuel consumption [5]. Classical methods for the detection of soot, such as the determination of the blackening index (SZ according to Bosch or filter smoke number FSN) or gravimetric methods, to investigate e.g. the effects of load change, are very difficult to use or even fail because of the lack of temporal resolution. Furthermore, with these methods emission behaviour cannot be measured at the location where particle formation takes place, i.e. directly inside the combustion chamber. In contrast, optical techniques are capable to take measurements in the combustion chamber and in the exhaust system [6-12]. Additionally, high temporal and spatial resolution of the phenomenon under investigation can be achieved.

2. Experimental Setup

2.1 Engine and test bed specifications The test engine was a single cylinder direct injection spark ignition research engine with spray guided combustion (Fig. 1), thus possible to operate in stratified mode.

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 1/8

Figure 1: Arrangement spark plug / injector The injector was placed in the center of the roof of the combustion chamber. One of the two exhaust valves was removed in order to position the spark plug and the cylinder head was modified to ensure a narrow arrangement of the spark plug and injector. The injector is equipped with a twelve-hole nozzle. The main technical data of the engine are listed in Table 1. Table 1: Technical data of the test engine Swept Volume 652 cm3

Bore 100 mm Stroke 83 mm Compression ratio 11.0 :1 - Valves 2 intake valves

1 exhaust valve -

Optical access was achieved by a fused silica glass ring beneath the roof of the combustion chamber as shown in figure 2. However, the piston completely blocks the glass ring between 30 crank angle degrees (CAD) before and after top dead center (TDC), respectively.

Figure 2: Optical accessible test engine 2.2 Soot measurement techniques In order to get two dimensional data of in-cylinder soot concentrations, mean particle radii and particle number densities during the expansion stroke, the

RAYLIX-technique was applied (Fig. 3). In addition, an AVL Smoke Meter 415S was used for measuring the Filter Smoke Number (FSN) in the exhaust gas. The measuring principle is based on the classical filter paper method.

3. RAYLIX-technique

In the following a brief description is given how the soot particle properties are obtained from the measured signals. The physical background has been described previously [13-15], therefore only a brief outline is given here.

Figure 3: Experimental setup of the RAYLIX-technique The RAYLIX-technique is based on the simultaneous detection of the Rayleigh scattering, the laser-induced incandescence (LII) combined with the determination of integral extinction of the laser light intensity. The two laser pulses of a modified Nd:YAG double pulse laser (PIV laser) are expanded into light-sheets passing the object under investigation in a temporal sequence. The temporal separation of both beams is < 100 ns, which is more than two orders of magnitude smaller compared to the turbulent time scales. The first pulse possessing a low energy density to avoid a significant heating of the soot particles, induces the Rayleigh-scattering. This low energy beam also serves for the estimation of the integral extinction. The second high energy laser beam heats up the soot particles to temperatures near the threshold of evaporation (≈ 4000 K). As a consequence the thermal radiation of the particles increases far beyond the thermal radiation encountered with flame temperatures. The Rayleigh-scattered light and the LII-signal are detected perpendicular to the propagation direction of the laser light sheet by two ICCD (intensified

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 2/8

charge coupled device) cameras. By calibrating the LII-image with the integral extinction, two-dimensional maps of the soot volume fractions are obtained. The Rayleigh-images are calibrated by the Rayleigh-signal from ambient air. The two-dimensional maps of the LII- and the Rayleigh-signal can be obtained with high spatial (< 200 µm) and temporal (< 200 ns) resolution. 3.1 Extinction In the Rayleigh-regime (soot particles are small compared to the wavelength of the scattered light) and if a log-normal size distribution of the particles is assumed, the coefficient of scattering Kext is given by

( ) V23

m2

22

ext Nσ5,4expr2m1mIm

λπ8K ⋅⋅⋅⎟⎟

⎠

⎞⎜⎜⎝

⎛+−

⋅−=

[1], with the wavelength of the light λ, the complex refractive index of the soot particles m, the mean particle radius rm, the standard deviation of the log-normal size distribution σ and the particle number density NV. For σ and m 0.34 [16-18] and 1.73-0.6i [19] is assumed, respectively. 3.2 Rayleigh Scattering Considering the assumptions described above, the scattering efficiency QV is given by:

( ) V26

m

2

2

2

4

4

Vsca Nσ18expr2m1m

λπ16Q~I ⋅⋅⋅

+−

⋅⋅

=

[2] The detection system has to be calibrated with the known scattering strength of air in order to obtain QV from the scattering intensity Isca. 3.3 Laser Induced Incandescence (LII) The detected LII signal shows a quantitative correlation to the soot volume fraction fV or the soot concentration which is soot volume fraction multiplied by the density of soot [20-22]. To obtain absolute values for fV the relative LII signal is calibrated with the integral extinction [23]. The calibration constant for the LII signal was determined from measurements at large (> 67) CAD based on images with small and compact soot clouds, respectively. As a consequence of the contamination of the glass ring during the measurements and, to a minor extend, by fluctuations in the pulse energy of the laser the experimental error is up to 40 %. Unfortunately, we have to admit a statistical error of up to 40% for this value as consequence of contamination of the glass ring during the measurements and, in minor amount, fluctuations in laser energy output. Additionally, effects of agglomeration of the primary particles are neglected here. It was shown, that in polydisperse fractal

aggregates the Rayleigh-Debye-Gans (RDG) theory provides better results, because soot particle agglomerates do not rigorously satisfy the Rayleigh-limit criteria [24-25]. However, the error is of minor amount, especially when a moderate laser fluxes are used [26]. 3.4 Determination of Mean Particle Radii and Particle Number Densities Using the different nonlinear dependence of the LII-signal (ILII ~ r3) and the Rayleigh-scattered light (IRay ~ r6), the mean particle radii rm and the particle number densities NV can be obtained by dividing both signals by each other. According to Jones [27], multiple scattering should not occur provided an extinction less than 63%. For cases of highest soot concentrations this limit is always reached. However, there is experimental evidence that multiple scattering effects can be observed even at an extinction of 10% [28]. In conclusion we have to admit an error based on multiple scattering in the same order of magnitude as the error based on the extinction measurements.

4. Experimental Results

The engine was operated at several inhomogeneous operating conditions, which differ in the injection and ignition timing. Additionally, one homogeneous operating condition with increased fuel and decreased air mass in order to achieve a relative air-fuel ratio of λ = 1 was studied. Engine speed was 1500 rpm, injection pressure was 130 bar, the injection strategies of the homogeneous and one stratified injection mode are listed in Table 2. Table 2: Engine operating conditions Mode SOI Ignition IMEP λ FSN CAD

before Compression TDC

CAD before Compression TDC

bar - -

homo. 340 30 2.2 1.0 0.07 strat. 24 8 1.9 3.9 0.55

Figure 4 shows images obtained from a high speed camera while engine was operated in homogeneous mode. The left side shows the position of the injector (multi-hole nozzle) as well as the individual jets of the fuel spray after start of injection. The position of the piston is also indicated. As can be clearly seen on the right side of the figure, the fuel spray, injected during inlet stroke, wets the piston (left side of figure) and is still present when combustion starts, leading to pool fires (right side of figure). To quantify the effects of pool fires on soot formation, results

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 3/8

obtained by the RALIX-technique will be presented and discussed in the following.

4.2 Stratified combustion mode The lower two-dimensional maps of the Figures 5a to 5c show the results during stratified engine operation. Soot is first detected in the combustion chamber volume at 40 CAD ATDC. In Figures 5a to 5c a representative result is shown (43 CAD ATDC). Thereafter, soot clouds begin to shift towards the piston surface, followed by an increasing soot concentration and cloud volume. At about 65 CAD ATDC, peak soot amount is reached. In general, the highest soot concentrations are detected close to the piston as can be seen by fV (5a). As well as in homogeneous mode, this fraction of the soot cannot be detected at 69 CAD or later, resulting in an proceeding separation of soot clouds. Nevertheless, soot can be detected at 69 to 80 CAD without an exception in comparatively high concentrations. In figure 5, a representative two-dimensional map at 75 CAD ATDC is presented. FSN in the exhaust was 0.55, which is quite consistent with the results discussed above.

Figure 4: Images from high-speed film As can be seen in Table 2 the FSN in stratified mode is much higher than in the homogeneous mode although the global λ-value is significantly lower. To get a deeper insight into this expected behavior, the RAYLIX results for these two operating points will be now discussed in detail. Figure 5 shows the soot volume fraction (5a), the mean soot particle radii (5b) and the soot particle number density (5c) for the homogeneous and stratified combustion mode during the expansion stroke at three characteristic times. In all figures the position of the piston is indicated by a white shaded line, while the upper and lower edge of the glass ring is indicated by a solid white line, respectively. Each two-dimensional map was obtained by averaging 20 individual images.

4.1 Homogeneous combustion mode The upper two-dimensional maps of the Figures 5a to 5c show the results during homogeneous engine operation, respectively. At 52 CAD ATDC from the two-dimensional maps of fV it becomes obvious that soot exits on top of the piston. The highest soot amount was detected at 55 CAD ATDC. While the piston moves downwards (65 CAD ATDC) soot volume fraction decreases, indicating that soot oxidation exceeds soot formation. At 69 CAD ATDC, the piston is below the lower ring of the detection window. As a consequence, the detection of soot at the top of the piston - as observed at earlier CAD’s - fails. Nevertheless, small discrete soot clouds were observed in some cases, as shown in Figures 5a to 5c at 79 CAD ATDC, but total soot amount was almost zero. From the very low number density NV (Figure 5c) it becomes obvious that the particles have really been oxidized. Obviously, the soot is more or less completely oxidized in the later phase of combustion leading to low soot emission as can be seen from the low Filter Smoke Number (FSN = 0.07).

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 4/8

Figure 5a: Soot volume fraction (fV); homogeneous versus stratified

Figure 5b: Mean soot particle radii (rm); homogeneous versus stratified

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 5/8

Figure 5c: Soot particle number density (NV); homogeneous versus stratified From these investigations it becomes obvious that in SI-engines with direct injection the main source of soot emission in the homogeneous mode are pool fires. In the stratified mode soot is firstly formed in the combustion chamber volume due to locally fuel rich mixture regions and later on, at the piston due to pool fires. The higher amount of soot formed during the stratified mode in combination with a slower combustion (globally lean fuel mixture (λ > 1)) leads to an incomplete oxidation of soot particles and consequently to higher soot emission. 4.3 Constance of particle size Figure 5b shows a relative constant particle size of ca. 10 nm diameter. Significant variations are observed near the piston due to reflections. In homogeneous mode, where reflections from the far side of the glass ring are not absorbed by soot clouds in the combustion chamber volume, this is even more pronounced. Furthermore, in stratified mode a region where dirt was deposited during the measurement shows apparently increased particle sizes (right side at 65 CAD and 75 CAD after top dead centre. In Figure 6 the local mean particle radii are averaged (weighted based on the local soot concentration) are shown as function of CAD. Unfortunately, this value

is not representative, if the overall soot amount is very low, as can be seen in homogeneous mode. Hence, measurements with an overall soot amount above 1e-4 mm3 are accentuated. This value is obtained by the estimation of the volume detected by the camera. The pixel area of the two-dimensional maps is determined from the magnification scale of the camera (0.141 mm/pixel). This pixel area is multiplied by the thickness of the laser light sheet (9 mm) to obtain the volume detected by each pixel. The sum of the soot volume fractions in the map multiplied with this volume is the total amount of soot detected in the combustion chamber. If a soot amount obtained by this procedure exeeds 1e-4 mm3, it is used for further data evaluation in figure 6.

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 6/8

0

2

4

6

8

10

12

14

16

18

20

20 30 40 50 60 70 80 90

CAD

mea

n pa

rticl

e di

amet

er [n

m]

homogenous

homogeneous &soot>1e-4 mm³stratified

stratified &soot>1e-4 mm³

Figure 6: fV-weighted, averaged mean particle diameter It is obvious that the particle diameters are between 6 and 12 nm. Moreover in stratified mode particle sizes are between 8 and 10 nm in diameter. This is in quite good agreement with the in-cylinder snatch sampling method, applied inside an HCCI engine [29].

5. Conclusion

For the first time the RAYLIX-technique was applied in a single cylinder spark ignition engine with direct injection and spray guided combustion to analyze soot formation and oxidation. It was observed that the main cause of soot formation in the homogeneous mode are pool fires. In the stratified mode soot is both formed in the combustion chamber volume due to local rich mixture regions and on the piston due to pool fires. Due to a lack of time the higher amount of soot cannot be fully oxidized, despite overall oxygen concentration is higher.

6. Acknowledgement

This study is funded within the research program CRC 606 (“Non-stationary Combustion: Transport Phenomena, Chemical Reactions, Technical Systems”) by the German Research Foundation, Federal Ministry of Education and Research, Germany

.

7. References

[1] J. P. Longwell: “Soot in Combustion Systems and its Toxic Properties”, Lahaye and Prado Editors, Plenum Press, New York, 1983.

[2] U. Heinrich, I. Mangelsdorf, M. Aufderheide, “Durchführung eines Risikovergleichs zwischen Dieselmotoremisssionen und Ottomotoremissionen hinsichtlich ihrer kanzerogenen und nicht-kanzerogenen Wirkung”, Umweltbundesamt, Berichte Bd.2/99, Erich Schmidt Verlag Berlin, 1999.

[3] H. Schulz: “Cardiovascular effects of nanoparticles”, 8th ETH Conference on Combustion Generated Nanoparticles, Zurich, Switzerland, 2004.

[4] A. Peters: “Effects of Fine and Ultra-fine Particles on the Heart”, 9th ETH Conference on Combustion Generated Nanoparticles, Zurich, Switzerland, 2005.

[5] U. Wagner: “Experimentelle Untersuchungen außer- und innermotorischer Maßnahmen zur Rußminimierung bei Dieselmotoren”, PhD-Thesis, Universität Karlsruhe (TH), 2006.

[6] S. Schraml, C. Heimgärtner, S. Will, A. Leipertz, “Application of a new Soot Sensor for Exhaust Emission Control Based on Time Resolved Laser Induced Incandescence (TIRE-LII)”, SAE Technical Paper 2000-01-2864, Society of Automotive Engineers Inc, 2000.

[7] S. Schraml, C. Heimgärtner, C. Fettes, A. Leipertz: “Investigation of In-Cylinder Soot Formation and Oxidation by Means of Two-Dimensional Laser-Induced Incandescence (LII)”, 10th International Symposium on Application of Laser Techniques to Fluid Mechanics, Lisboa, Portugal, 2000.

[8] B. F. Kock, T. Eckhardt, P. Roth: “In-Cylinder Sizing of Diesel Particles by Time-Resolved Laser-Induced Incandescence (TR-LII)”, Proceedings of the Combustion Institute, 29, 2, Elsevier Inc, 2002.

[9] B. F. Kock, B. Tribalet, C. Schulz, P. Roth: “Two-color time-resolved LII applied to soot particle sizing in the cylinder of a Diesel engine”, Combustion and Flame, 147, Elsevier Inc, 2006.

[10] D. R. Snelling, G. J. Smallwood, R. A. Sawchuk, W. S. Neill, D. Gareau, L. W. Chippior, F. Liu, L. Ö. Gülder, W. D. Bachalo: “Particulate Matter Measurements in a Diesel Engine Exhaust By Laser-Induced Incandescence and the Standard Gravimetric Procedure”, SAE Technical Paper 1999-01-3653.

[11] G. J. Smallwood, D. R. Snelling, R. Sawchuk, W. S. Neill, D. Gareau, W. L. Chippior, F. Liu, W. D. Bachalo, O. L. Gulder: “In-Situ Real-Time Characterization of Particulate Emissions From a Diesel Engine Exhaust By Laser-Induced Incandescence”, SAE Technical Paper 2000-01-1994.

[12] G. J. Smallwood, D. R. Snelling, Ö. L. Gülder, D. Clavel, D. Gareau, R. A. Sawchuk, L. Graham: “Transient Particulate Matter Measurements From the Exhaust of a Direct Injection Spark Ignition Automobile”, SAE Technical Paper 2001-01-3581.

[13] H. Geitlinger, T. Streibel, R. Suntz, H. Bockhorn: “Two-dimensional imaging of soot volume fractions, particle number densities, and particle radii in laminar and turbulent diffusion flames”, Twenty-

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 7/8

seventh Symposium (International) on Combustion, Proceedings, 1, The Combustion Institute 1998.

[14] M. Stumpf, A. Velji, U. Spicher, B. Jungfleisch, R. Suntz, H. Bockhorn, “Investigations on soot emission behaviour of a common-rail diesel engine during steady and non-steady operating conditions by means of several measuring techniques”, SAE Technical Paper 2005-01-2154, Society of Automotive Engineers Inc, 2005.

[15] H. Bockhorn, H. Geitlinger, B. Jungfleisch, T. Lehre, A. Schön, T. Streibel, R. Suntz: “Progress in characterization of soot formation by optical methods”, Physical Chemistry Chemical Physics, 15, 4, RSC Publishing, 2002.

[16] G. Wannemacher: “Untersuchungen zur Russbildung in brennerstabilisierten Kohlenwasserstoff-Sauerstoff-Flammen”, PhD-Thesis, Technische Hochschule Darmstadt, 1983.

[17] H. Bockhorn, F. Fetting, A. Heddrich, G. Wannemacher: “Investigation of the surface growth of soot in flat low pressure hydrocarbon oxygen flames”, Proceedings of 20th Symposium (International) on Combustion 1984, Pittsburgh, Pa., USA, 1982, 20, 1, The Combustion Institute, 1984.

[18] A. Heddrich: “Untersuchungen zur Bildung von höheren Kohlenwasserstoffen und zum Teilchenwachstum von Russ in flachen laminaren Kohlenwasserstoff-Sauerstoff-Flammen”, PhD-Thesis, Technische Hochschule Darmstadt, 1986.

[19] H. Chang, T. T. Charalampopoulos: “Determination of the wavelength dependence of refractive indices of flame soot”, Proceedings of the Royal Society of London, Series A: Mathematical, Physical and Engineering Sciences, 430, 1880, The Royal Society, 1990.

[20] L. A. Melton: “Soot diagnostic based on laser heating”, Applied Optics, 23, 13, Optical Society of America, 1984.

[21] D. L. Hofeld: “Real-time soot concentration measurement technique for engine exhaust streams”, SAE Technical Paper 930079, Society of Automotive Engineers Inc, 1993.

[22] J. Appel, B. Jungfleisch, M. Marquardt, R. Suntz, H. Bockhorn: “Assessments of Soot Volume Fractions from Laser Induced Incandescence by Comparison with Extinction Measurements in Laminar Premixed Flat Flames”, Proceedings of the Combustion Institute / Symposium (International) on Combustion, 26, 2, Elsevier Inc, 1996.

[23] B. Yang, U. O. Koylu: “Detailed soot field in a turbulent non-premixed ethylene/air flame from laser scattering and extinction experiments”, Combustion and Flame, 141, 1-2, Elsevier Inc, 2005.

[24] T. L. Farias, Ü. Ö. Köylü, M. G. Carvalho: “Range of validity of the Rayleigh–Debye–Gans theory for optics of fractal aggregates,” Applied Optics 35, 33, 6560 (1996).

[25] Ü. Ö. Köylü: “Quantitative analysis of in situ optical diagnostics for inferring particleyaggregate parameters in flames: implications for soot surface

growth and total emissivity”, Combustion and Flame, 109, 4, 488, (1996).

[26] P. O. Witze, S. Hochgreb, D. Kayes, H. A. Michelsen, C. R. Shaddix: “Time-resolved laser-induced incandescence and laser elastic-scattering measurements in a propane diffusion flame”, Applied Optics 40, 15, 2443 (2001).

[27] A. R. Jones: “Scattering of electromagnetic radiation in particulate laden fluids”, Progress in Energy and Combustion Science, 5, 2, Elsevier Ltd, 1979.

[28] M. Kerker: “The Scattering of Light and Other Electromagnetic Radiation”, Academic Press, New York, 1969.

[29] S. Mosbach, M. S. Celnik, A. Raj, M. Kraft, H. R. Zhang, S. Kubo, K. O. Kim: “Towards a detailed soot model for internal combustion engines”, Combustion and Flame, 156, 6, Elsevier Inc, 2009.

8. Glossary

ATDC: After Top Dead Centre BTDC: Before Top Dead Centre CAD: Crank Angle Degree FSN: Filter Smoke Number IMEP: Indicated Mean Effective Pressure LII: Laser Induced Incandescence rpm: Revolutions per Minute SOI: Start of Injection TDC: Top Dead Centre

The spark ignition engine of the future – December 2nd & 3rd, 2009 – Strasbourg INSA Page 8/8