Study of laminar premixed flames in transient electric fields using PLIF and PIV techniques

Transcription

1 Study of laminar premixed flames in transient electric fields using PLIF and PIV techniques Johannes Kuhl, Gordana Jovicic, Lars Zigan *, Alfred Leipertz Lehrstuhl für Technische Thermodynamik (LTT), Universität Erlangen-Nürnberg, Germany, and Graduate School in Advanced Optical Technologies (SAOT), Universität Erlangen-Nürnberg, Germany * Corresponding author: Lars.Zigan@cbi.uni-erlangen.de Abstract The enhancement of the flame stability as well as the reduction of the pollutant emissions can be achieved by the use of the effect of electric fields; however, the ongoing mechanisms are not yet fully understood. In this experimental study laser based measurements such as planar laser induced fluorescence (PLIF) as well as particle image velocimetry (PIV) were used to reveal the basic mechanisms of the interaction as well as response times and its dependencies. Hence, a transient DC electric field was applied with different frequencies (1-200 Hz) to a laminar methane Bunsen type flame to reveal the flow and flame behavior. The PIV results show a deceleration of the flow field of the flame beginning at the flame root and growing downstream with time. The changes in the shape of the flame are calculated from the formaldehyde- and the OH-PLIF images in terms of the heat release rate. Thereby, the first flame response time was estimated to be 2-4 ms and different behaviors for higher frequencies were derived as the whole process takes more than 20 ms to stabilize. For 100 and 200 Hz the time between the on- and off-pulse is too short compared to the flame response time, so that a final steady state cannot be reached. 1. Introduction The basic mechanisms of the interaction between electric fields and hydrocarbon flames are still not fully discovered although the phenomenon itself is known for a long time [5,15]. In this time many experiments have been carried out for premixed and non-premixed flames. An early approach for a general description of the phenomenon can be found in the books of von Engel [11,12] and a substantial review on the influence of electric fields on flames is given by Bradley [4]. The two achievable effects with benefit to technical applications are enhanced flame stabilization and the reduction of pollutant emissions if the electric field is arranged longitudinally to the flame [6,7,27]. First attempts for a technical application at higher thermal load can be found in Weinberg et al. [32] and Altendorfner et al. [1]. However, fundamental work is necessary to understand this effect for the utilization in technical combustion systems and model improvements. Most of the early studies were carried out using conventional measurement techniques, e.g., exhaust gas analysis, temperature measurements or mechanical probing. However, they are highly invasive and introduce changes to the chemical reactions in the flame front as well as disturbance of the flow field. Therefore, non-invasive laser based techniques were used in the presented study. They allow a detailed insight into the complex interactions of the electric field and flame chemistry, flow pattern and may give an answer about the prevailing mechanism with a high temporal and spatial resolution. Therefore, the laminar Bunsen type flame was operated under various operating conditions [2,18] and two different species concentrations were visualized with planar laser-induced fluorescence (PLIF) to characterize the flame front structure. The flow field of the flame was simultaneously determined using particle image velocimetry (PIV). Previous laser based experiments [2,18] applying weak electric fields to laminar premixed Bunsen flames using a positively charged electrode above the flame gave strong indication that the ionic wind is mainly responsible for flame stabilization and the changes in the flame front. Other authors [16,17,25] assume that the ionic wind is not solely responsible for the strong flame changes when the burner itself was charged. For technical applications such as electric field assisted combustion it is necessary to know the response time of the flame in electric field. This time scale is required for model validation to - 1 -

2 identify influencing variables and mechanisms to enable a future application into technical systems, e.g. for the suppression of acoustic oscillations. Up to now, only few experimental data for the quantification of the momentum transfer between charge carriers and molecules are available [20,21,28,32]. Sun et al. [30] observed first luminosity changes (but no flame structure changes) 2 ms after applying an electric field on a flat-flame McKenna burner using high speed photography whereas the total flame response time took about 36 ms. Altendorfner et al. [2] showed local decelerations of about 1.5 m/s in the flow field of the flame for static fields when the electric field was applied. This deceleration clearly corresponds to the Ionic Wind effect (see below). In contrast, no clear change in the calculated flame front thickness was observed, i.e. the chemistry appears not to be influenced by the electric field. In the presented experiments this test rig was expanded with a high voltage, high frequency switch generating a transient DC-field in order to determine flame response times and changes in the flame structure. Therefore, the transient DC-field was applied with four different frequencies to different operating points. 2. Theoretical Background In general the two controversial discussed explanations of the interaction of electric fields with flames are both based on the existence of charge carriers in hydrocarbon flames [13]. They are produced in the flame front by chemo-ionization and consist mainly of electrons and positively charged ions (e.g. H 3 O +, CHO +, C 2 H 3 O + and C 3 H 3 O + ) and only a few negatively charged ions which again quickly decay into electrons and neutral molecules [20]. In case of a positively charged electrode above the burner the electrons are accelerated downstream whereas the positively charged ions are decelerated by the Lorenz Force and if the electric field strength is high enough even accelerated upstream towards the burner rim. According to their mass the electrons are faster accelerated whereas the ions transfer a higher momentum if they collide with other molecules. Due to the momentum transfer during multi collision processes the positively charged ions cause a hydrodynamic back pressure which often is referred to as Ionic Wind in literature. This Ionic Wind is the first explanation for the interaction phenomenon. The second explanation presumes a higher momentum transfer at very high electric field strengths which would be large enough to excite other molecules. This energy is then again transferred by intermolecular interactions to an enhancement of the production of radicals and thereby directly changes the chemistry of the flame. Moreover, the electric field withdraws the charge carriers from the flame front and thereby possibly changes the chemistry of the flame as well. In literature typical positive ion concentrations in the reaction zone vary between 10 9 and /cm 3 for premixed hydrocarbon-air flames [8-10,17,31]. It was shown that the strength of the local flow deceleration depends on the air-fuel mixture whereas the ionic wind effect was stronger in rich flames and thereby dependent on the number density of positively charged ions [18,19]. In summary, the effect is very complex and strongly dependents on the application of the electric field. Therefore, a theoretical calculation of the local strength of the ionic wind is difficult, since the local charge carrier concentration in the flame and their transport properties such as ion mobility are unknown and depend on local mixture composition and temperature. Due to this reason, only few numerical studies exist for prediction of local flame and flow changes, see, e.g. [3,23,29]. Most of those studies are purely based on electrohydrodynamic processes (i.e. the ionic wind) neglecting chemical changes introduced by the flow changes. Simplifications by only considering the main charge carrier, namely the H 3 O + -ion and electrons and assuming constant ion mobility are further sources of inaccuracy regarding the flame response behavior. Therefore, more experimental validation data is required. 3. Experimental Setup As discussed in the theoretical part it is not clear whether the electric field affects the flow field (Ionic Wind) or the flame chemistry or possibly both. Therefore two laser based measurement techniques were chosen to identify the dominant process, the particle image velocimetry (PIV) to determine the flow field of the flame and to reveal possible changes due to the Ionic Wind and the planar laser-induced - 2 -

3 fluorescence (PLIF) to detect specific flame molecule distributions. Therefore, two PLIF-techniques were chosen to excite the formaldehyde molecule and the OH radical from which the local heat release rate (HRR) can be calculated. The heat release rate again corresponds quite well with the actual flame front which otherwise could not be visualized easily [14]. The experimental test rig consists of a laminar Bunsen burner, the supply and implementation of the electric field and the optical components of the laser based measurement techniques, as schematically shown in Fig. 1. Fig. 1: Experimental setup including burner, electric field generation, lasers and cameras. The burner has an inner diameter of 13 mm and a length of 730 mm to ensure a fully developed laminar flow profile and is operated with methane-air mixtures. To study the chemical effects of the flame in an electric field different air-fuel ratios from λ = 1.0 to 1.2 were adjusted. The gas flows were adjusted with mass flow controllers and the combustion air was dried to a dew point of -80 C using an adsorption air dryer. For all operation points the flame was surrounded by a coaxial co-flow with a constant exit velocity of 1 m/s to shield from ventilation and dust of the ambience. 50 mm above the burner rim the ring-shaped electrode with a diameter of 20 mm was mounted with a ceramic holder to ensure electrical insulation to the setup. In order to close the electric circuit the burner was grounded. For the cooling of the electrode de-ionized water was used. The applied voltage was generated by a special high power voltage supply. A constant supply voltage of 6 kv which equals electric field strength of 120 kv/m was applied to all operating points whereas the supply voltage was kept in the linear current voltage regime below saturation and secondary ionization, such as corona discharge. Moreover, to avoid any possible damage to the electrode by plasma discharge a 12 MΩ resistance was mounted which limited the current to 3.33 ma at a supply voltage of 40 kv. The high speed, high voltage switch (Behlke) to interrupt the constantly applied DC voltage was mounted between the resistance and the power supply and consequently monitored with a 1:1000 high voltage probe (LeCroy) an displayed on an oscilloscope. To determine the flow field particle image velocimetry (PIV) was applied. The air-fuel mixture was seeded with a blend of titaniumdioxide particles with an average diameter of 1 µm. A double-pulsed Nd:YAG laser with a wavelength of 532 nm was used to generate two laser pulses with a short fixed temporal delay of 50 µs. The light sheet for PIV was formed with a focal lens of 1000 mm and a telescope of two lenses with focal lengths of 63 and 150 mm. The signal was detected with a nonintensified double shutter CCD camera. A narrow band pass filter centered at 532 nm was included in order to suppress flame luminosity and the radiation of the other two excitation lasers for the PLIF measurements as well as the fluorescence signals. Commercially available software package was used for the vector field calculation. The interrogation area was set to 16 x 16 pixels with 50% overlap and multi-pass. Planar laser-induced fluorescence (PLIF) was used to measure the two dimensional distributions of the formaldehyde molecule (CH 2 O) as well as the OH radical through the flame center. A broadband - 3 -

4 Nd:YAG laser at 355 nm was used to excite formaldehyde, which acts as a precursor for the flame front indicating the beginning of the preheating zone. The fluorescence signal was detected with an intensified CCD camera in a range from 380 to 500 nm with several band pass and notch filters. The light sheet was formed simultaneously with the laser beam for PIV after overlaying both laser beams using a dichroid mirror but focused with a separate 1000 mm focus lens. For the excitation of the OH radicals, which are a good marker for the flame front and the post oxidation zone, a dye laser was pumped with another Nd:YAG laser at a wavelength of 532 nm. First, the wavelength was changed to 566 nm in the dye laser using Rhodamin 6G and then doubled to 283 nm afterwards. For these experiments, the narrow band dye laser was tuned to nm to excite a specific OH line. The laser beam was first focused with a lens of 1000 mm focal length and then homogenized and expanded using a beam homogenizer and a Fourier lens of a focal length of 500 mm [22]. Finally, it was overlapped with the two other light sheets using another dichroid mirror. The detection of the fluorescence signal of the OH radicals was done at a wavelength of 308 nm with an intensified CCD camera and several band pass filters to suppress scattered light. To enable the observation of the same region of interest with all three cameras, the PIV camera and the formaldehyde camera were positioned using the Scheimpflug principle [24], whereas the camera for the OH radical was adjusted perpendicular to the detection plane. After simultaneous acquisition of the three signals, all single images were evaluated separately with a commercially available software package, averaged over 50 images and finally mapped and overlaid using equidistant cross-pattern images. 4. Results and Discussion In this experimental study transient electric fields are applied to estimate response times of laminar premixed flames. Therefore the air-fuel ratio, the exit velocity and the frequency of the DC-fields were varied in order to clarify the dynamic interactions. Previous experimental studies with static fields showed that the strength of the ionic wind depends on the amount of charge carriers accelerated in the electric field and the number of collisions with the fresh gas molecules. Thus, the variation of the airfuel ratio shows a stronger impact on ionic wind than the variation of the exit velocity [18]. However, in the presented study the response time and the response behavior of the flame at different frequencies and operating points are of concern. Therefore, the changes of the electric field excited flame are observed over time at the on- and the off pulse for a period of 20 ms. After the first reaction of the flame it will take some time until the process reaches a steady state. Thereby, the time range from the first flame response until the fully developed steady state will limit the applicable frequency which is very important to know for further technical application. For higher frequencies than the limiting frequency the flame will start to respond but cannot complete as the field is deactivated before the flame reaches the steady state. Therefore the flame will start to oscillate in a quasi stationary condition where the phase may be shifted to the applied electric field and the amplitude of the oscillation will get smaller for higher frequencies. The repetition rate of the high voltage is therefore increasingly applied with 1, 10, 100 and 200 Hz. Thereby, for 1 and 10 Hz the flame response is scanned for 20 ms in 2 ms increments during the rising and the falling edge of the pulse as the electric field strength was kept constant at 120 kv/m (with a supply voltage of 6 kv) as well as the symmetry of the pulses is kept at 50:50 for all operating points. For higher frequencies, where the period is only 10 or 5 ms respectively, the whole period was scanned with twenty equidistant time steps. In Fig. 2 and 3 exemplary differential PIV images are shown from an operating point of λ=1 and an exit velocity of 2 m/s. In a differential image the changes are highlighted as constants cancel out. For the differential PIV images the initial flow field of the flame is always subtracted from the current flow field revealing the changes under the influence of the electric field. Moreover, the calculated heat release rate is overlaid in white to mark the flame front. Fig. 2 shows the changes of the flame for the rising edge

5 2 ms 6 ms 18 ms Fig. 2: Development of the flow deceleration of a laminar flame with λ=1.0, with a transient electric field (rising edge at 0 ms, 50 % pulse symmetry; 120 kv/m). A deceleration around the flame root already occurs within the first 2 ms when the flame is subjected to the electric field. The maximum deceleration in this region is reached at 6 ms while the deceleration still develops further downstream until the complete flame reaches its final flow velocity and structure at 18 ms. The flame is constricted at its root point and its tip is stretched. Moreover, a disturbance is developed, (a wave-like oscillation) which is initialized at the flame root point and propagating in axial direction. However, the flame front disturbance is less pronounced compared to ref. [26] where turbulent flame front wrinkling was reported using different electric field geometry. In Fig. 3 the flame behavior for the falling edge of the electric field is shown. 502 ms 506 ms 510 ms Fig. 3: Development of the local deceleration of a laminar flame with λ=1.0, with a transient electric field (falling edge at 500 ms, 120 kv/m). The developed flow field changes disappear after deactivation of the electric field and the former flow field begins to establish again. While the deceleration is still fully developed until the electric field is switched off at 500 ms its starting point at the flame root lifts off within the first 2 ms. The strength of the deceleration decreases while the region propagates further downstream (see e.g. at 506 ms) until it finally disappears after 10 ms. It can be noted that the initial flow field without electric field is faster developed during the off-pulse than the deceleration for the on-pulse, see Fig. 2. In general the deceleration behavior reported in Figures 2 and 3 stays the same for other operating points. Only the strength of the local deceleration changes with the air-fuel ratio [19]. In contrast the calculated flame front (i.e. the heat release rate) does not show any clear broadening or any other changes in the flame front from which a changed chemical reaction could be derived in these macroscopic images. Therefore, the resolution has to be improved for future experiments. Nevertheless the flame stretching and the disturbance are clearly visible in the OH fluorescence images, therefore, the OH fluorescence images will be analysed for further characterisation of the transient phenomenon. In Fig. 4 an OH image is shown where on the right side of the left image with electric field the initial signal is subtracted to point out the developed changes. The constriction at the flame root point and the - 5 -

6 broadening of the post-oxidation zone becomes clearly visible. In Fig. 4 (right) extracted profiles at 5 mm height are shown where the temporal changes become evident too. The steep slope marks the position of the flame front corresponding to the temperature distribution. After the peak the OH molecule signal decays in the exhaust gas region. The local broadening of the OH region indicates the flame response to the changed flow field. In order to quantify the response behavior the OH-LIF images were evaluated at horizontal lines at the heights of 5 and 15 mm above the burner rim. Thereby the signal was normalized to each maximum and the inner flame radius was taken at the position of half maximum signal intensity. Fig. 4: Comparison of OH-fluorescence signal images (left) with 6 kv supply voltage and the differential with the initial flame (0 kv) at the right half image. Normalized OH-fluorescence profiles from which the inner flame radii were calculated over time (right). The build-up time of electric fields between the ring-electrode and the burner is already considered in the displayed function of the rising edge and was controlled by the high-voltage probe head and the oscilloscope. Therefore, the presented time scale is the actual flame response time when the electric field is completely developed. For the presented results the Bunsen type flame was operated with λ=1.2 and an exit velocity of 1.5 m/s. Moreover, the derived radii are normalized to the one calculated from the initial OH fluorescence signal taken from the uninfluenced flame. In Fig. 5 the development of the inner flame radius (R) is shown at the heights of 5 and 15 mm above the burner for applied frequencies of 1 and 10 Hz. Moreover, the applied field is indicated as recorded from the high-voltage probe. Fig. 5: Response behavior of the flame shown by the inner flame radius at 5 and 15 mm for the rising edge of the transient DC field (left) and the falling edge (right) for frequencies of 1 and 10 Hz. The temporal behavior at 5 mm, i.e. in the region of the flame root point, where the first noticeable changes in the flow occurred and at 15 mm are shown using the calculated flame radius. Until the rising edge at 0 ms the inner flame radius stays constant. At 4 ms an increase of the radius at both heights is - 6 -

7 evident. Whereat the radius at 15 mm keeps constantly increasing until 8 ms the flame radius at 5 mm drops below its initial representing the constriction of the flame root. After the minimum at 8 ms the flame radius at 5 mm increases slightly again but stays below its initial. At the height of 15 mm the enlarged radius also drops after 8 ms below its initial reaching a local minimum at 14 ms. The disturbance created by the constriction at the flame root travels the 10 mm between the two extracting profiles within 4 ms, or with a speed of 2.5 m/s which is about 1.0 m/s faster than the average exit velocity of the uninfluenced flow. This is in agreement with the increased flow velocity in the fresh gas because of the constriction at the flame root as shown in Fig. 2. At the falling edge first changes become evident at 4 ms after the electric field is switched off at 500 ms. The radius at 15 mm drops to a local minimum at 8 ms and begins to oscillate because of the flame front changes. The oscillations at 5 mm are smaller and in opposite direction. At both heights the initial flame radius is not reached after 20 ms. Anyhow, the temporal behavior of the flame for both, the on- and offpulse are very similar at 1 Hz and 10 Hz. In Fig. 6 and 7 the development of the inner flame radius at 100 and 200 Hz is shown. Because the pulse duration at such frequencies is shorter than 20 ms the on- and off-pulse were not scanned separately. Hence, always the whole period was scanned with ten equidistant time increments (i.e. 1 ms step size for 100 Hz and 0.5 ms step size for 200 Hz). Fig. 6 shows the changes in the inner flame radius for a frequency of 100 Hz at 5 and 15 mm on the left and for 200 Hz on the right. Again, the applied voltage is indicated as recorded from the high-voltage probe. Fig. 6: Response behavior of the inner flame diameter at 5 and 15 mm for a frequency of 100 Hz (left) and 200 Hz (right) of the DC field. The temporal behavior of the flame proves an oscillating character. Steady initial condition cannot be reached anymore. However, the amplitude of the oscillation is much larger than with 1 or 10 Hz (Fig. 5). The inner flame radius at 5 mm oscillates below its initial with smaller amplitude than at 15 mm. The mean of the oscillation at 15 mm lays below the initial radius at this position whereas the peak of the oscillation exceeds it about 3 ms. Moreover, the oscillations at 5 and 15 mm are peak to peak phase shifted by 3-4 ms which is a little bit faster than the disturbance speed that was calculated for 1 and 10 Hz respectively. In both heights it becomes apparent that the flame can adopt the frequency of the electric field at 100 Hz. Yet, the oscillations are shifted in time in comparison to the rising- and the falling-edge of the electric field. In addition, the large amplitude and the time shift suggests that some flow amplification processes are overlaid. At 200 Hz the temporal behavior of the flame reveals an oscillating character again and also the pulsation of the flame equals the frequency of the electric field. In comparison to 100 Hz the amplitude of the oscillation is much smaller and the oscillation at the two heights is almost in phase. There seem to be two waves of disturbance travelling through the flame front at the same time. This could explain the coincidental in phase character of the two curves. In total, the flame radius is always smaller than its initial at this position and the flame still adapts the impressed frequency despite the low flame response time of 2-4 ms (see Fig.5)

8 Therefore, 200 Hz seem to be close to the limiting frequency applying pulsed electric fields for flame control. Since at 100 Hz the maximum flame structure changes were identified this operating point is further discussed in detail. In Fig. 7 a combination of the OH-fluorescence images (left) with overlaid flow deceleration from the PIV results (right half images) are shown to highlight the mentioned processes. Fig. 7: Response behavior of the flame (λ=1.2 and an exit velocity of 1.5 m/s) marked by the OH-fluorescence in combination with the calculated flow deceleration for 100 Hz over time Although the moving disturbance is better visible in the outer exhaust gas region it corresponds well with the trend of the inner flame radius in Fig. 6. At the height of 5 mm the inner radius is already reduced after the rising edge at 0 ms. The maximum radius is reached at 3 to 4 ms and afterwards decreasing again until reaching its minimum at 9 ms. In Contrast, at a height of 15 mm the inner flame radius provides a minimum at 3 ms and consequently increases to the maximum at 8 ms which is even above the initial radius. A region of deceleration grows from the flame root beginning at 2 ms. The strength of the deceleration increases up to 1.1 m/s until the electric field is deactivated at 5 ms again. At 6 ms the region of deceleration detaches at the flame root region and starts to disappear by propagating downstream and dissipates until the field is switch on again. Then a region of flow deceleration begins to establish at the flame root again. This process still corresponds well to the one shown in Fig. 2 and 3 although a steady state is not reached any more and the maximum deceleration reached at 100 Hz (1.1 m/s) equals the maximum for 1 and 10 Hz at the same operating point. The large amplitude of the flame radius changes at 100 Hz can be explained by the magnitude of the flame response velocity, due to the ionic wind and the local flame convection velocity (i.e. the transportation of the surface waves) at this operating point. Although the time for complete development of the maximum flame response is longer (~8 ms at 1 Hz) the flame can adopt the electric field frequency. The induced flow disturbance travels with about 2.5 m/s (convective flow) which is faster than the maximum velocity change at the flame root point due to the ionic wind. Therefore, two surface waves are moving through the flame leading to its distinct curvature and stretching. 5. Conclusion In this study the response behavior of laminar Bunsen flames on transient weak DC-electric fields were studied using laser based measurement techniques. The application of PLIF and PIV techniques allowed a detailed observation of the complex ongoing processes of electric field induced flow field and flame structure changes. In the applied electric field the flame is pushed towards the burner and is constrained, which can be attributed to the momentum transfer by the ionic wind. From the OH-PLIF-images temporal information of the flame structure can be calculated, which is more distinct compared to the scalar velocity field, since the flow direction also changes. Moreover, when the electric field is applied a disturbance is evident propagating downstream through the flame and the exhaust gas region broadens which corresponds to deceleration of the flow. Therefore, the inner radius of the OH-fluorescence at two - 8 -

9 flame heights was chosen to describe the local response behavior of the flame front. The maximum OH-intensity is localized at the flame front and shows a displacement when the electric field is applied. The results show a similar behavior of the flame radius for frequencies of 1 and 10 Hz. First changes occurred 2-4 ms after the rising and the falling edge whereas first decelerations were measured already after 2 ms. After the first response of the flame it takes more than 20 ms to reach stationary conditions. Moreover, from the delay between the two observed heights the travelling speed of the disturbance through the flame front was calculated to be 2.5 m/s which corresponds to the convective velocity. A different behavior occurred for higher frequencies. Hereby, the duration where the electric field is applied is too short to reach stationary conditions. Therefore, the transient phenomenon establishes a quasi-stationary condition with an oscillating character. The oscillations are phase shifted for 100 Hz whereas the peak to peak distance at the two heights equals the one observed from the lower frequencies. Yet the amplitude especially at 15 mm is larger than for lower frequencies. Therefore, it is assumed that the flow processes interfere with each other. The region of deceleration reaches until 100 Hz its maximum strength of about 1.1 m/s and reveals a similar behavior for lower frequencies. At 200 Hz the amplitude of the radius oscillation is smaller. The response time which is essential for active flame stabilization by electric fields was estimated to be about 2-4 ms, which is faster than theoretical values provided in the literature for the integral flame. However, no clear indication or negation for changed flame chemistry could be detected. Therefore, this effect needs further analysis focusing on the temperature and species field by application of additional measurement techniques. Acknowledgements The authors gratefully acknowledge financial support by the German Research Foundation (DFG) for parts of this work and for funding the Erlangen Graduate School in Advanced Optical Technologies (SAOT) within the framework of the German Excellence Initiative

10 6. Literature [1] Altendorfner F., Beyrau F., Leipertz A., Hammer T., Most D., Lins G., Branston D.W. (2010), Chemical Engineering & Technology 33 (4) [2] Altendorfner F., Kuhl J., Zigan L., Leipertz A. (2011), Proceedings of the Combustion Institute, 33 (2) [3] Belhi M., Domingo P., Vervisch P. (2010), Combustion and Flame [4] Bradley D., Weinberg F.J. (Ed.) (1986), Advanced Combustion Methods, Academic Press, London, pp [5] Brande T.W. (1814), Philosophical Transactions of the Royal Society of London [6] Calcote H.F., Berman C.H. (1989), ASME Fossil Fuels Combustion Symposium, Houston, pp [7] Calcote H.F., Pease R.N. (1951), Industrial and Engineering Chemistry 43 (12) [8] Calcote H.F. (1962), Proceedings of the Combustion Institute 8 (1) [9] Calcote H.F. (1963), Proceedings of the Combustion Institute 9 (1) [10] Calcote H.F. (1965), Ionization in hydrocarbon flames, Report No. TP-119, AeroChem Research Laboratories, New Jersey, pp [11] von Engel A., Steenbeck M.( 1932), Elektrische Gasentladungen Ihre Physik und Technik, vol. 1, Julius Springer Verlag, Berlin, p [12] von Engel A., Steenbeck M. (1934), Elektrische Gasentladungen - Ihre Physik und Technik, vol. 2, Julius Springer Verlag, Berlin, p [13] Goodings J.M., Bohme D.K., Sugden T.M. (1977), Proceedings of the Combustion Institute 16 (1) [14] Gökalp I. (1987), Combustion and Flame 67, [15] Hauksbee F. (1708), Philosophical Transactions of the Royal Society of London [16] Kim M.K., Chung S.H., Kim H.H. (2011), Proceedings of the Combustion Institute, 33 (1), [17] Kim M.K., Ryu S.K., Won S.H., Chung S.H. (2010), Combustion and Flame [18] Kuhl J., Jovicic G., Zigan L., Leipertz A. (2012), Transient electric field response of laminar premixed flames, 34 th International Symposium on Combustion, Warsaw, Poland. (accepted for presentation) [19] Kuhl J., Jovicic G., Zigan L., Leipertz A. (2011), Proceedings of the 5 th European Combustion Meeting, Cardiff, Wales

11 [20] Lawton J., Weinberg F.J. (1969), Electrical Aspects of Combustion, 1st ed., Clarendon Press, Oxford, p [21] Payne K.G., Weinberg F.J. (1962), Proceedings of the Combustion Institute 8 (1) [22] Pfadler S., Beyrau F., Löffler M., Leipertz A. (2006), Optics Express 14 (22) [23] Prager J., Riedel U., Warnatz J. (2007), Proceedings of the Combustion Institute 31 (1) [24] Prasad A.K., Jensen K. (1995), Applied Optics 34 (30) [25] Ryu S.K., Kim Y.K., Kim M.K., Won S.H., Chung S.H. (2010), Combustion and Flame [26] Schmidt J., Kostka S., Lynch A., Ganguly B. (2011), Experiments in Fluids 51 (3), [27] Sepp V.A., Ulybyshev K.E. (1997), High Temperature 35 (5) [28] Starikowskii A.Y. (2005), Proceedings of the Combustion Institute [29] Starikowskii A.Y., Skoblin M., Hammer T. (2008), 17 th International Conference on Gas Discharges and their Applications, Cardiff [30] Sun L., Lins G., Hammer T. (2005), Interaction of a low-pressure flat flame with an electric field, Proceedings of the European Combustion Meeting, Louvain-La-Neuve [31] Weinberg F. (1986), Advanced Combustion Methods, Academic Press, London, [32] Weinberg F.J., Carleton F., Dunn-Rankin D. (2008), Combustion and Flame