Electrostatically Assisted Fuel Injection and Charged Droplet Combustion. University of Illinois at Urbana-Champaign Urbana, IL USA

Transcription

1 ILASS Americas, 20 th Annual Conference on Liquid Atomization and Spray Systems, Chicago, IL, May 2007 Electrostatically Assisted Fuel Injection and Charged Droplet Combustion E. K. Anderson 1, A. P. Carlucci 2, A. De Risi 2, and D. C. Kyritsis 1* 1 Department of Mechanical Science and Engineering University of Illinois at Urbana-Champaign Urbana, IL USA 2 Department of Engineering for Innovation University of Lecce Lecce, Italy Abstract A gasoline fuel injector was modified to allow the induction of electrostatic charge to the spray as an additional means of controlling fuel dispersion. Fraunhofer diffraction measurements were made on the sprays, and comparing average droplet size for ten sprays, it was found that the standard deviation was cut in half by electrostatic assistance, and the relative span factor of an individual spray was reduced by approximately 10%. E10 droplets at various charge levels were ignited and the flame was recorded with a high speed camera. Morphological differences between different levels of charge were dramatic. * Corresponding author

2 Introduction The potential of stratified charge combustion to reduce the fuel consumption of spark ignited engines was one of the reasons for the development of Direct Injection Stratified Charge (DISC) engines. This engine design utilizes the fact that an engine can be operated with a stoichiometric air-fuel mixture near the spark [1] but leaner mixtures as the flame propagates away from the spark source [2]. However, it is still difficult to precisely control fuel stratification, which has hindered widespread use of such engines in automobiles. Accurate fuel stratification is made difficult by turbulent in-cylinder flow and the wide range of engine speeds and loads encountered by automotive engines, and poor control of fuel stratification can result in increased emissions and reliability problems. Current pressure driven fuel injectors suffer from the limitation that the mass of fuel injected, which is determined by the load on the engine, is coupled to the fuel momentum, which determines the spray pattern and is therefore crucial to achieving accurate stratification. A decoupling of fuel mass and momentum can be achieved by using electrostatic charge as an additional means of controlling the spray. As early as 1971, Thong and Weinberg [4] investigated electrical control of the combustion process. Later work by Kim and Turnbull [5] and Kelly [6] showed that electrostatic charge could be injected into low conductivity liquids, such as the liquid hydrocarbons present in gasoline, by using sharp electrodes. It is possible to use this technique to apply electrospray technology, which is already a mature technology and widely used in analytic chemistry [7-9], for fueling in the field of power generation. This was done by Gomez and collaborators who extensively studied fuel electrosprays [10,11] and used an electrospray in a liquid fuel battery which burns kerosene atomized by an electrospray technique [12,13]. The potential use of electrospray technology in micro combustors was also discussed by Gemci and Chigier [14]. Employing electrostatic effects as the sole means of fuel atomization, as was done in this line of work, simplifies fueling, but also limits the achievable fuel flow rate. In comparison, relatively little research interest has been directed toward the combination of electrostatic fuel atomization with pressure driven sprays, likely because of the assumption that inertial forces would dominate the electrostatic forces. Some recent research, however, suggests this is not necessarily true. A hybrid electrospray/air assisted injection system design was shown to achieve good droplet size control [15] and a paper by Hedrick and Parsons [16] discussed the application of electrospray in port fuel injection for an internal combustion engine. Computational work done by Thomas et al. [17] showed that electrostatic effects were significant during diesel injection, and a later paper described findings that soot formation was affected by electrostatic charging [18]. Shrimpton [19] showed computationally that fuel dispersion in a Direct Injection Spark Ignition (DISI) engine could be improved through the application of electrostatic charge to the fuel spray. Most of the above studies, however, are computational and lack experimental verification. The objective of our work was to experimentally determine the effect of electrostatic charge on gasoline sprays typical of those in automotive gasoline engines. A commercially available GDI injector was modified to apply charge to the fuel spray and installed in an optically accessible test chamber. Techniques described in [20,21] were employed to measure the effect of charge on fuel dispersion and droplet size. Additionally, combustion of electrostatically charged gasoline droplets was compared to combustion of noncharged droplets through the use of high speed video. Experimental Apparatus and Techniques The hardware and measurements are described in two separate sections below which focus separately on the non-reacting sprays and combustion of charged droplets. The fuel used for both experiments was the same pump 87 (R+M)/2 octane gasoline. regarding the fuel are available in [22]. Details Electrostatically Charged Sprays The configuration of the apparatus for obtaining Fraunhofer diffraction measurements of droplet size is shown in Fig. 1. The injector used was a Mitsubishi gasoline direct injector, p/n MR modified to inject electrostatic charge into the fuel. The design of the electrode placed over the injector tip, shown in Fig.2, utilizes the fact that the electric field strength is locally enhanced by sharp edges on surfaces with high electric potential. It features a conical tip at the orifice where fuel emerges. The electrode was made from copper and has 1 mm side walls and a 0.5 mm base thickness. The sharp cone protrudes 0.5 mm at an angle of 135 degrees with the base. The electrode generates a locally intense electric field that partially polarizes the fuel molecules and injects ions into it, thus transferring a net charge to the fuel. A polyester insert fits between the copper electrode and the injector body to electrically insulate the injector. To produce the charged sprays, 3 kv was applied to the copper electrode from a Fluke model 410B voltage generator. Details regarding the choice of angle for the sharp conical feature are available in [20,21].

3 Figure 1: Experimental configuration for Malvern measurements. Gastight # ml syringe. The syringe pump was mounted vertically so that the syringe would point downwards. Downstream of the syringe, the fuel flowed through a steel 22 gauge syringe needle and then through a 5.2 cm long 0.18 mm outer diameter silicon carbide capillary. For each droplet combustion experiment, a droplet was fed from the syringe and suspended from the silicon carbide capillary. Ignition was provided by a spark generated by an automotive coil. A Pyramid Model PS-36KX power supply generating 12 volts was used to power a Bosch coil. The primary coil was triggered by a Gigavac model GR5LTA145 relay which received a timing signal from a Stanford Research Systems Model DG535 pulse generator. The droplet charging system consisted of a Fluke model 410B voltage generator which was set to 3 kv for charged droplet experiments and 0 kv for noncharged experiments for comparison. The high voltage lead from the voltage generator was connected to a Givac model GR5LTA145 high voltage relay to allow rapid switching of the high voltage. The relay was triggered by the same pulse generator described above. The high voltage switching system was necessary in order to apply high voltage only after the spark event to prevent discharge of the high voltage circuit through the ignition circuit. Imaging of the droplet combustion was provided by a Phantom v7 high speed camera. Three Nikon extension tubes totaling 82.5 mm were used to increase magnification along with a Tiffen 52 mm zoom lens. Camera exposure time was set to 70 µs and the zoom and f-stop settings were adjusted to provide a clear view of the entire flame. The camera was triggered by Figure 2: Injector electrode. Fraunhofer diffraction measurements were made using a Malvern Instruments Spraytec RTS 5000 particle analyzer. This instrument provided measurements of droplet diameter averaged across the line of sight of the instrument. For our injector sprays, the instrument line of sight was oriented perpendicular to the injector axis and measurements were made at several positions downstream of the injector. The results shown here were obtained at 2.8 bar injection pressure and 9 mm below the injector tip. Charged Droplet Combustion The experimental configuration for obtaining high speed video of charged and non-charged gasoline droplet combustion is shown in Fig.3. A KD Scientific Model 100 syringe pump was used to pump known quantities of ethanol blended gasoline from a Hamilton Figure 3: Droplet combustion apparatus.

4 the same pulse generator controlling spark and the high voltage circuit. Results and Discussion Fraunhofer Diffraction (Malvern) Measurements Results of the Malvern measurements are shown in Fig 4. There, droplet size results are presented as a function of time during spray evolution. The data were taken at a distance of 9 mm below the injector tip and for an injection pressure of 2.8 bar. It should be noted that because of the way the Malvern technique operates, the data are integrated in the radial direction in the spray, i.e. the measurements provide an average size across the radial direction in the spray. At each time after injector actuation, ten measurements were taken for both the charged (3 kv) and the non-charged sprays and the mean value and the standard deviation of these samples are reported in Fig. 3. The two plots of Fig. 4 both show D v50, or median droplet diameter, for charged and non-charged sprays. D vxx is defined for what follows as the droplet diameter for which XX% of the injected mass is composed of smaller diameter droplets. For these sprays, the D v50 value is actually higher for the charged sprays, indicating that the charging is not helpful in generating smaller droplets. However, in addition to the D v50, Fig. 4 also shows the standard deviation of the D v50 for 10 separate charged sprays and 10 separate non-charged sprays to give some indication of spray-to-spray variation. The standard deviation is significantly reduced (by approximately a factor of two) for the charged sprays, indicating that the application of electric charge makes the median droplet diameter much more consistent between sprays. This can be potentially beneficial for automotive engines where obtaining low cycle to cycle variation is important. In the second plot of Fig.4, the relative span factor (RSF) of the charged and non-charged sprays is reported instead of standard deviation. RSF is computed for each individual spray as RSF D D ) D ( v90 v10 / v50 Defined in this way, the RSF is indicative of the variation of droplet size within an individual spray. Figure 4 shows that although difference in RSF is not as dramatic as that of the standard deviation, there is a significant reduction in the droplet size variation within individual sprays when electrostatic charging is applied. The narrower droplet size distribution for the charged sprays is in general agreement with the results of electrospray experiments which often report monodisperse droplet size distributions [8, 10-11]. Charged Droplet Combustion Figure 4. Median droplet diameter (Dv50), standard deviation of Dv50, and relative span factor (RSF) plotted against time after injector actuation. If an electrostatically assisted injection system such as the one described in the preceding sections were to be installed on an engine, the droplets injected into the cylinder would carry some charge and in addition there would be some electric field generated by the potential difference between the charged injector tip and the electrically grounded walls of the cylinder. At the same time, flames do generate charged species because of the high temperatures which lead to the generation of a very dilute plasma [24]. In order to provide preliminary results on electrostatic effects on the flames of electrostatically charged fuels, experiments investigating the ignition and combustion of ethanol blended gasoline droplets have been conducted. The mixing of ethanol to the gasoline fuel, in addition to providing a link to the emerging field of biofuels, serves the purpose of enhancing the electrical conductivity of the fuel because of the polar -OH bond. Using the experimental apparatus described in the previous section, charged and non-charged droplets

5 were generated, ignited, and recorded with a high speed video camera. In Fig. 5, several images from the high speed videos of the droplet combustion are presented. For all of the images presented, a fuel droplet of approximately 10% ethanol and 90% gasoline of 1±0.2 mm diameter was generated by the syringe pump. Once a stable droplet was formed, it was ignited with the spark electrodes visible in the figure as horizontal lines coming from the lower left and right sides of each image. The high speed camera began recording 40 ms before the ignition event for each video. In Fig. 5, one frame from the high speed video of seven different droplets is shown. The charge on the capillary in each video varied from 3 kv to + 3 kv. All of the images were captured 100 ms after the ignition event, so the only experimental parameter changed between images is the applied voltage. Figure 5 shows that the flame is attracted toward the electrically grounded spark electrodes for positive voltages. In addition, for positive voltages, the flame is also attracted towards the ground plate, which is visible at the bottom of the images for positive voltages, and just below the field of view for the rest of the images. This is due to the fact that the camera was moved downward for the positively charged droplet combustion so that the entire flame could be captured by the camera. In contrast, the luminous soot for the negatively charged droplets formed a taller and narrower visible flame, with no apparent attraction to the electrically grounded spark electrodes or ground plate. These results suggest that the luminous soot carries some positive electrostatic charge, causing it to be attracted to the lower-potential nearby objects, which are the spark electrodes and ground plate for the positively charged droplets, and the negatively charged capillary for the negatively charged droplets. The unusually small flame of the -2 kv image is due to the large error in producing droplets of consistent diameter with the syringe pump. Fuel was observed to continue being ejected from the capillary even after the syringe pump stopped moving, suggesting that some sort of valve is necessary to improve fuel metering. For this reason, these are clearly preliminary results that establish that the presence of electrostatic charge does result in significant morphological changes in the visible flame. Further investigation is necessary, which might include improving fuel metering. In addition, improved visualization of the fuel droplet itself during combustion is desired in order to allow a measurement of the droplet diameter and, therefore, a meaningful comparison of burning rates for charged and non-charged droplets. It should be noted that the current transferred to a nearby ground plate as fuel was fed through the charged Figure 5: Images of ethanol blended gasoline combustion at various levels of droplet charge. Images were obtained from high speed video of droplet combustion and show the flame 100 ms after ignition. droplet combustion apparatus was measured. Results showed that the droplet combustion apparatus generated charge densities around 130 nc/ml, compared to charge densities of around 2 nc/ml for the non-reacting, charged gasoline sprays. This is due both to the configuration of the particular apparatus and the increased conductivity of the ethanol containing fuel.

6 Conclusions The addition of an electrode to the tip of a commercially available fuel injector enabled significant electrostatic effects to be observed experimentally in a commercially available fuel injector. The use of a sharp conical feature on the electrode was important in achieving an acceptable level of charge on the low conductivity gasoline. Malvern measurements of droplet size distribution showed that the application of electrostatic charge significantly increased spray repeatability, as seen by a roughly 50% reduction in the standard deviation of the observed average droplet size computed for ten sprays. In addition, the relative span factor, an indicator of droplet size distribution within individual sprays, was smaller for the charged sprays. These two improvements came at the cost of a slight increase in average droplet size. Video of positively and negatively charged burning droplets suggests that the soot generated carries a positive charge. For the positively charged droplets, soot luminosity was observed extending very little in the vertical direction, but stretched towards nearby electrical grounds. For negatively charged droplets, the luminous soot was stretched upward in the vertical direction and appeared to be attracted more strongly to the negatively charged capillary than to the nearby electrical grounds. Acknowledgements The authors would like to thank the Donors of the American Chemical Society Petroleum Research Fund for support of this research (Grant # PRF G9-Dr. David L. Shutt, Acting Program Director) and the support of the U.S. Department of Energy Graduate Automotive Technology (GATE) Center of Excellence at the University of Illinois at Urbana-Champaign (Grant DOE DE-FG26-05NT42622). 9. Law, S.E. (Ed.), Journal of Electrostatics 54:217 (2002). 10. Tang, K.Q., and Gomez, A., Journal of Colloid and Interface Science 184: (1996). 11. Gomez, A., and Tang, K.Q., Physics of Fluids 6: (1994). 12. Kyritsis, D.C., Roychoudhury, S., McEnally, C., Pfefferle, C., Gomez, A., Experimental Thermal and Fluid Science 28: (2004). 13. Kyritsis, D.C., Coroton, B., Faure, F., Roychoudhury, S., Gomez, A., Combustion and Flame 139:77-89 (2004). 14. Gemci, T., Chigier, N., AIAA Paper Nr (2003). 15. Rickard, M., Dunn-Rankin, D., SAE Paper (2002). 16. Hedrick, R.E., Parsons, M.H., SAE Paper (1997). 17. Thomas, M.E., DiSalvo, R., Makar, P., SAE Paper (2002). 18. Bellan, J., Harstad, K., Atomization and Sprays 8: (1998). 19. Shrimpton, J.S., International Journal for Numerical Methods in Engineering 58: (2003). 20. Anderson, E.K., Carlucci, A.P., de Risi, A., Kyritsis, D.C., Atomization and Sprays 17:1-25 (2007). 21. Anderson, E.K., Carlucci, A.P., de Risi, A., Kyritsis, D.C., International Journal of Vehicle Design in press (2007) oline.pdf. 23. Adrian, R.J., Experiments in Fluids 39: (2005). 24. Lewis, B., von Elbe, G., Combustion, Flames, and Explosions of Gases, 2 nd ed., Academic Press, References 1. Takagi, Y., Proceedings of the Combustion Institute 27: (1998). 2. Ricardo, H.R., U.S. Patent 1,271,942 (1918). 3. Shiga, S., Ozone, H., Machacon, T.C., Karasawa, T., Nakamura, H., Udea, T., Jingu, N., Huang, Z., Tsue, M., and Kono, M., Combustion and Flame 129:1-10 (2002). 4. Thong, K.C., and Weinberg, F.J., Proceedings of the Royal Society of London 324: (1971). 5. Kim, K., and Turnbull, R.J., Journal of Applied Physics 47: (1976). 6. Kelly, A.J., Journal of the Energy Institute 57: (1984). 7. Fenn, J.B. (Nobel lecture), Angewandte Chemie International Edition 42: (2003). 8. Fernandez De La Mora, J., The Journal of Fluid Mechanics 243: (1996).