1. Introduction

Scramjet engine has been viewed as a promising air-breathing propulsion system for aerospace applications. Ignition and sustaining combustion in the supersonic flow remains to be a critical issue in this engine system. At the takeover flight speeds with low Mach numbers (Ma<5), the low pressure and temperature in the supersonic flow will prohibit auto-ignition of the fuels. External sources must be applied for ignition in this situation.

In recent years, several plasma-based methods have been demonstrated for ignition in the supersonic flows, including plasma torches [1–4], discharges in the combustible medium [5–9] and laser-induced plasmas [10–12]. Compared with the electronic discharges, the laser-induced plasma (LIP) has several benefits for ignition, including easier control of the reaction position and timing, more flexible energy deposition for ignition, wider ignitable equivalence ratio range, feasibility of multi-point ignition, etc [13, 14]. With these benefits, the laser-induced plasma ignition (LIPI) has been extensively investigated in the last decade, and has been viewed as a promising ignition technique for natural gas engines [15], internal combustion engines [16], rocket engines [17, 18] and other advanced combustion systems. However, applications of LIPI in the supersonic combustion systems are rather limited [10–12]. Zudov et al. [10] investigated ignition of a homogeneous methane-air supersonic jet using a repetitive CO₂ laser generated plasma and demonstrated the possibility of LIPI in the supersonic flow. On a model hypersonic combustor fueled with hydrogen, Brieschenk et al. [11, 12] compared two LIP-based ignition schemes: direct LIPI in the shear layer of the inlet hydrogen injection, and ignition with laser plasma ionized fuel jet. Planar laser-induced fluorescence (PLIF) diagnostics showed that both ignition schemes could successfully generate hydroxyl (OH) radical, however, sustainable combustion was not achieved since no flameholding was applied. Compared with hydrogen, hydrocarbon fuels (ethylene, kerosene, etc.) are more difficult to ignite due to their longer ignition delays and narrower flammability limits. For liquid kerosene, the ignition also involves fuel atomization and evaporation processes, which makes successful ignition more challenging.

In this paper, we present cavity ignition of liquid kerosene in a supersonic flow of Ma 2.52 with a LIP on a model supersonic combustor equipped with dual cavities in tandem as flameholders. Successful ignition and sustainable combustion have been achieved. To our knowledge, this is the first time that the LIP is applied for successful ignition of liquid kerosene in the supersonic flow. Temporal evolution of the flame kernel generated by the LIP and its development into sustainable combustion have been obtained using the high-speed chemiluminescence imaging method. The correlation of the eventual ignition results with the experimental conditions is discussed considering the fuel atomization and evaporation, hydrodynamic and mixing conditions in the cavity where the ignitions are conducted.

2. Experimental setup

The experimental setup for cavity ignition of liquid kerosene in the supersonic flow with the LIP is shown in Fig. 1. The ignition experiments were conducted on a direct-connect test facility which consisted of an air heater (not shown), a supersonic nozzle (not shown), an isolator, a supersonic model combustor and a diffuser in the downstream. The air heater generated the vitiated air by simultaneous combustion of air, oxygen and ethanol. The total temperature, total pressure and oxygen concentration of the vitiated air could be controlled by adjusting the mass flow rate of each propellant. The mass flow rate of the vitiated air was 1 kg/s. A Ma 2.52 supersonic nozzle was adopted for the present ignition investigations, corresponding to the inlet flow condition of flight Ma 5.5 at 23 km altitude. A constant area isolator was installed between the supersonic nozzle and the supersonic model combustor to avoid thermal choking. The model combustor has a constant cross section of 50 mm width and 40 mm height. The static pressure in the combustor prior to the ignition was approximately 0.86 bar. Two open cavities (denoted as C1 and C2, respectively, along the main flow) in tandem were applied as flameholders, both with a depth of 11 mm and a length to depth ratio of 7. The dual-cavity configuration has been demonstrated to be better than the integrated injection/ignition/flameholding single cavity configuration for liquid kerosene [19]. The front walls of the cavities were perpendicular to the main flow, while the rear walls were inclined 45° towards the downstream.

 

Fig. 1 Experimental setup for cavity ignition of liquid kerosene in the supersonic flow with the laser-induced plasma.

Download Full Size | PDF

The liquid kerosene at the ambient temperature was applied as the fuel. The kerosene was injected from three orifices with diameter of 0.2 mm located at the front wall of Cavity C1, parallel with the main flow. The orifices were 8 mm above the cavity bottom and 10 mm apart in the spanwise direction. The mass flow rate of the fuel and hence the overall equivalence ratio (ER) was controlled by adjusting the injection pressure.

To allow for laser incidence and optical diagnostics, the model combustor was equipped with quartz windows on the top and side walls. A Q-switched Nd:YAG laser (Spitlight 250, Innolas) with a repetition rate of 100 Hz was applied for the ignition experiments. The 1064 nm, ~8 ns output of the laser was expanded to a spot size of 25 mm in diameter, guided into the cavity and focused using a f = 100 mm plano-convex lens to generate the plasma in the spanwise center plane of the combustor. The LIPI experiments have been conducted in both cavities. The steering optics and the focal lens (enclosed in the dashed rectangle in the figure) were installed on a translation stage, to shift the ignition tests between the two cavities and to vary the ignition position in each cavity. A pulse energy of ~183 mJ was used throughout the ignition experiments. The laser pulse energy was measured using a joule meter (J50-MB-YAG, Coherent). From the pre-determined energy coupling of the LIP under the typical static pressure of the combustor, about 65% of the incident laser energy was coupled into the plasma.

The temporal evolution of the LIP-initiated flame kernel and its development into sustainable combustion were investigated by recording the chemiluminescence of the flame kernel with a high-speed camera (SA-5, Photron). The frame rate of the camera was 20,000 frames per second (fps). To analyze the ignition results and their correlation with the experimental conditions, qualitative distribution of the kerosene vapor in Cavity C2, in which both successful and failed ignitions were obtained (see context below), was investigated using the kerosene-PLIF technique [20]. The fourth harmonic generation of the Nd:YAG laser (266 nm, ~5 mJ) was shaped into a sheet with a combination of cylindrical and spherical lenses, and guided through the spanwise center plane of the cavity to excite the kerosene vapor. The PLIF signal of the di-aromatics components in the kerosene vapor was imaged using an ICCD camera (PIMAX-II, Princeton Instruments).

The direct-connect test facility, the ignition laser, the high-speed camera and the ICCD camera were synchronized using a digital delay generator (DG645, Stanford Research Systems).

3. Results and discussion

3.1 Ignition results and temporal evolution of the flame kernel

Under the same fuel injection scheme, the LIPI tests performed in Cavity C1 and Cavity C2 show quite different results. In Cavity C1, the ignition tests at different locations showed that it was unfeasible to achieve successful ignition in this cavity. This is consistent with the previous ignition test results conducted on the same facility using a high energy electric spark plug [19]. In that case, no ignition could be achieved in Cavity C1 using the spark plug, even with a pulse energy of up to 5 J [19]. We attribute the ignition failure to the imperfect fuel atomization and evaporation in Cavity C1. Indeed, during the ignition tests, white kerosene plume could be observed in Cavity C1, indicating abundance of large droplets in the cavity and bad atomization and evaporation of the fuel.

In Cavity C2, since it had a longer distance from the injection site, the fuel could have longer time to atomize and evaporate, which would create a more favorable condition for successful ignition. Therefore, more detailed ignition tests were performed in Cavity C2. A search for appropriate ignition position was performed first in this cavity. The overall ER was set to 0.24, a condition which had been tested as ignitable using the high energy electric spark in the previous investigations. The LIP was generated in the spanwise center plane of the cavity, 4 mm above the cavity bottom. The ignition position was gradually moved upstream from the rear ramp edge towards the cavity front wall. It was found that no successful ignition could be realized until a distance of 13.5 mm between the LIP and the cavity front wall was achieved.

The ignition properties of different overall ERs were then investigated with the LIP fixed at 13.5 mm downstream the cavity front wall. We denote this position as P_ig in the following context. The overall ERs were varied in the range 0.16-0.24. The ignition tests showed that successful ignitions could only be achieved with overall ERs of 0.18-0.24, while the ignitions failed when the overall ERs were less than 0.18.

Shown in Fig. 2 is the temporal evolution of the flame kernel of a typical successful ignition event performed in Cavity C2. In the figure, the supersonic main flow comes from left to right. The labels on the images are the temporal delays relative to the onset of the LIP. The bright light in the image with delay = 0 μs is due to the strong emission of the LIP. It can be seen that the flame kernel is generated within 50 μs following the LIP. After generation, the flame kernel is carried to the cavity leading edge by the recirculation flow in the cavity. Then the kernel develops in the leading edge area from delay of 200 μs to 1300 μs. From delay of 1400 μs, the flame begins to spread to the rear part of the cavity and fills the whole cavity around delay of 1700 μs. The flame continues to develop in Cavity C2 for about 8 ms, characterized with gradually enhanced chemiluminescence intensity and increased heat release. After that, the flame begins to propagate upstream towards Cavity C1, as shown in Fig. 3. The flame enters Cavity C1 at delay of approximately 14 ms, and fills the whole Cavity C1 at delay of about 16 ms. Full combustion is then established in Cavity C1, and much stronger chemiluminescence can be observed. Fluctuations in the flame front position could be observed during this upstream propagation process. The upstream propagation of the flame is probably realized through a boundary layer where the flow is in the subsonic regime, such that the downstream disturbances caused by the pressure build-up can propagate through this subsonic channel to the upstream.

 

Fig. 2 Typical temporal evolution of the flame kernel of a successful ignition event in Cavity C2. The fuel was injected from the front wall of Cavity C1. The bright light in the image with delay = 0 μs is due to the strong emission of the laser-induced plasma.

Download Full Size | PDF

 

Fig. 3 Upstream propagation of the flame from the downstream Cavity C2 to the upstream Cavity C1. The fuel was injected from the front wall of Cavity C1.

Download Full Size | PDF

In a typical failed ignition event, a flame kernel can also be generated by the LIP, however, the kernel generally extinguishes within 100 μs following its generation.

3.2 Discussion

The ignition experiments show that the final ignition results are closely correlated with the experimental conditions. For the liquid kerosene, several factors may affect the eventual LIPI ignition results, including fuel atomization and evaporation, hydrodynamic condition in terms of local strain rate and mixing condition in terms of local ER in the vicinity of the ignition site and in the leading edge area of the cavity where the flame kernel develops.

Bad fuel atomization and evaporation would cause strong scattering of the incident laser and even stop the laser from penetrating through the fuel and forming the plasma. The ignition failure in Cavity C1 is mainly due to bad fuel atomization and evaporation in this cavity.

Cavity C2 has a better atomization and evaporation condition, however the eventual ignition results are closely correlated with the ignition position and the overall ER. For spark ignition of turbulent flames, both the local hydrodynamic and local mixing conditions in the vicinity of the ignition position can affect the ignition results [21]. To realize a successful ignition, the local strain rate should not be too high to quench the LIP-generated initial flame kernel [21]. Meanwhile, the local ER should be within the flammability limit. The high-speed chemiluminescence shows that, for the failed ignitions initiated in the rear part of Cavity C2, the LIP-formed flame kernel is usually quenched within 100 μs following its generation. In this region, the shedding vortex that results from the impingement of the shear layer and the cavity aft ramp is too strong, which causes a high local strain rate. The high local strain rate then quenches the initial flame kernel and leads to eventual ignition failure.

For the ignitions performed with different overall ERs at the P_ig, since the mass flow rates of the fuel are much smaller than that of the main flow (1 kg/s), the flow field in Cavity C2 is mainly determined by the main flow. Thus, the local hydrodynamic conditions at the P_ig and in the leading edge area of the cavity are comparable with different overall ERs. Therefore, in this situation, the local ER is thought to be the main factor affecting the evolution of the flame kernels and the eventual ignition results.

To support the presumptions above, the qualitative distribution of the kerosene vapor in Cavity C2 in the cases of successful and failed ignitions was measured using the kerosene-PLIF technique. Shown in Fig. 4 is the comparison of the kerosene-PLIF images corresponding to successful and failed ignitions. The images are accumulation of 20 independent frames during the fuel injection process. The images are saturated at digital count of 2500 to enhance the difference in the regions where the ignition is conducted and where the flame kernel develops. The images are indexed to the same colormap scales and hence they can be directly compared. Although the kerosene-PLIF images are not calibrated to the absolute kerosene vapor concentration values, in consideration of the same excitation and detection optics, the same imaging parameters and similar static pressures and temperatures for the kerosene-PLIF investigations, the PLIF signal $S_{PLIF}$ is approximately proportional to the kerosene vapor concentration $n_{fuel}$, i.e., $S_{PLIF}\propto n_{fuel}$. Therefore, we can make a comparison of the kerosene vapor concentration distribution and hence the local ER distribution of successful and failed ignitions based on the kerosene-PLIF images.

 

Fig. 4 Comparison of the kerosene planar laser-induced fluoresce images corresponding to successful and failed ignitions. The ignition position P_ig and the leading edge area of the cavity are indicated as the solid star and the dashed circle, respectively. The dashed line indicates the ignition test line. The images are saturated to digital count of 2500 to enhance the difference in the regions where the ignition is conducted and where the flame kernel develops.

Download Full Size | PDF

From Fig. 4, we can see clear differences in the relative kerosene vapor concentrations at the P_ig (indicated as the solid star) and in the leading edge area of the cavity (indicated as the dashed circle). The relative kerosene vapor concentration and therefore the local ER of the successful ignition event is obviously higher than that of the failed one, indicating the important effect of the local ER on the final ignition results.

We can also see from the upper panel PLIF image corresponding to successful ignition event that, in the rear part of Cavity C2, the relative kerosene vapor concentrations along the ignition test line (shown as the dashed line) are of similar values to that at the P_ig. The local mixing condition at these positions should also allow for a successful ignition, however, the ignitions eventually failed. This fact demonstrates the equally important effect of the local hydrodynamic condition on the ignition results. As we have pointed out above, the too large local strain rates at these positions quench the initial flame kernel within a short time before it can reach the leading edge area of the cavity, and lead to ignition failures.

Therefore, based on the qualitative distribution of the kerosene vapor in the cavity, both the hydrodynamic condition in terms of local strain rate and the mixing condition in terms of local ER value can affect the development of the flame kernel and the eventual ignition result in Cavity C2.

Our ignition results demonstrate important merits of LIPI over conventional electronic discharge plasmas, especially easier control of the ignition position and more flexible energy deposition. Meanwhile, the LIPI proposes more reasonable technical requirements on the laser system (a pulse energy of about 180 mJ at 100 Hz), when compared with the high pulse energy (up to 5 J) of the electric spark plug [19]. The requirements on the laser system can be further reduced by applying other advanced LIPI schemes, such as the laser ablation ignition and double-pulse ignition. The laser ablation ignition generates the plasma for ignition by ablating solid targets instead of gas breakdown. The minimum laser pulse energy for ignition has been demonstrated to be reduced by one order of magnitude in the laser ablation ignition scheme [22]. The double-pulse ignition applies two laser pulses with adjustable inter-pulse delay to generate the plasma for ignition. Using the same amount of energy as a single-pulse ignition system, the double-pulse ignition can result in a longer plasma lifetime and better ignition characteristics [23]. In practical engine systems, the LIPI system needs to be miniaturized. In this case, a burst-pulse laser may be the solution. This laser system can achieve high repetition and high pulse energy output simultaneously in a short period while minimizing thermal load to the gain media [24]. Therefore, high ignition success probability and reliability can be achieved with a compact LIPI system. Further work in these aspects are in progress now.

4. Conclusion

In conclusion, we have achieved successful cavity ignition and sustainable combustion of liquid kerosene in a supersonic flow of Ma 2.52 using a LIP on a model supersonic combustor equipped with dual cavities in tandem as flameholders. High-speed chemiluminescence imaging shows that for a typical successful ignition event, the flame kernel is generated within 50 μs following the LIP and then carried to the leading edge area of the cavity where the initial kernel develops into flame covering the whole cavity. The flame initiated in the downstream Cavity C2 then propagates upstream into Cavity C1 and establishes full sustainable combustion.

The correlation of the final ignition results with the ignition position and the overall ER indicates the combined effects of the fuel atomization and evaporation, local hydrodynamic and local mixing conditions on the flame kernel evolution and its development into sustainable combustion. Bad fuel atomization and evaporation lead to ignition failures in Cavity C1. The qualitative distribution of kerosene vapor in Cavity C2 indicates that for the ignitions at the P_ig with different overall ERs, the ignition results are mainly determined by the local ERs in the vicinity of the ignition position and the lead edge area of the cavity. However, the ignition failures at the rear part of the cavity characterized with similar local mixing conditions to that of the P_ig also demonstrate the equally important effect of the local hydrodynamic conditions on the flame kernel evolution and the eventual fate of the ignition.

This work demonstrates the feasibility of LIPI in ignition of hydrocarbon fuels, especially liquid kerosene, in a scramjet combustor. We believe the results presented in this work can serve as references for the growing laser ignition community, both in the aspect of applications of LIPI and in the aspect of developing laser systems for the LIPI applications.