MCGA-assisted ignition process and flame propagation of a scramjet at Mach 2.0

Tiangang LUO,Jiajian ZHU,Mingbo SUN,Rong FENG,Yifu TIAN,Qinyuan LI,Minggang WAN, Yongchao SUN

Science and Technology on Scramjet Laboratory, National University of Defense Technology, Changsha 410073, China

KEYWORDS Cavity;Ignition process;Multi-Channel Gliding Arc(MCGA);Scramjet;Supersonic flow

Abstract The ignition process and flame propagation with ethylene fuel in cavity-stabilized scramjet by a Multi-Channel Gliding Arc(MCGA)at Mach 2.0 were investigated.Effects of equivalence ratios on the MCGA-assisted ignition process and flame propagation of the scramjet were recorded by two high-speed cameras from different view angles.The discharge characteristics of MCGA are also collected synchronously with the high-speed cameras.The distributions of temperature, velocity,and equivalence ratios in non-reactive flows of the cavity were simulated by Reynolds Averaged Navier-Stokes (RANS)model.The results show that MCGA can achieve reliable ignition with the Global Equivalence Ratios(GER)between 0.06 and 0.17.The ignition process is composed of flame kernel generation, flame development, and stable combustion.The time from flame kernel generation to the establishment of global flame decreases as GER decreases from 0.17 to 0.08.In the streamwise direction, the flame first develops to the Cavity Leading Edge (CLE) because of the influence of the cavity recirculation zone and then uplifts into the cavity shear layer, and finally develops to the Cavity Trailing Edge(CTE).In the spanwise direction, the flame width is less than 50% of the width of the cavity before developing to CLE and begins to develop towards the two sides of the combustor after reaching CLE, which is affected by the angular recirculation zone on both sides of CLE.The ignition processes by MCGA in the scramjet combustor are significantly affected by local distributions of equivalence ratios and velocity in the cavity.

1.Introduction

Scramjet has advantages of high speed, and simple structure,and is the core device of air-breathing aircraft that can achieve hypersonic flight.1In a scramjet combustor, ignition is very difficult on account of high-speed inlet flow conditions and the residence time of fuel is usually several milliseconds.2To realize the successful ignition of scramjet combustors,researchers have done extensive work in experimental3–5and numerical simulation6–8of reliable ignition process and combustion characteristics of scramjet combustors.The cavity flame-holder is usually used to make the ignition easier,but the selection of the ignition method is also very important.9–11Investigations on the plasma ignition in a scramjet were widely conducted.The thermal and chemical effects of plasma play an important part in decreasing ignition delay time and realizing rapid ignition.12–14Therefore, plasma-enhanced ignition technology is expected to achieve reliable ignition of scramjet combustors.Typical plasma sources include plasma jet,15laserinduced plasma,16spark,17and Gliding Arc (GA).18.

In the scramjet ignition process,the structure of the cavity,the non-reaction,and the reaction flow field have an important effect on the ignition.19–21An et al.22studied the laser and spark ignition progress with ethylene fuel at Ma=2.92,which shows that the flame kernel formed by laser ignition grows faster than spark ignition before reaching the Cavity Leading Edge (CLE), whereas they are almost the same in the rapid development stage.In terms of flame propagation, the initial flame generated by laser ignition is first brought back to CLE, then develops rapidly until it fills the cavity.Jia et al.23investigated the ignition of GA and found that it can broaden the blowout limit of the combustion, and the centralized discharge is superior to the single-channel discharge.In recent years, GA has a higher penetration than spark ignition, and the continuous releasing of the energy of GA also accelerates the ignition and reduces ignition time.15The numerical simulation method based on RANS can explain the experimental phenomenon reasonably and provide reliable ideas for experimental design.Wang et al.24reviewed the development of the cavity-stabilized scramjet using numerical methods.The influence of combustor, fuel injection, and equivalence ratio on combustion were studied by Two-Dimensional(2D)numerical simulation.25,26Huang et al.27used 2D numerical simulation to study the influence of injection scheme and equivalence ratio distribution on the flow field in the combustor and found that vertical injection is helpful to mode transition.Song et al.28studied single-orifice and multi-orifice injection schemes in a cavity-stabilized scramjet using 3D numerical simulation and found that multi-orifice has higher mixing efficiency and combustion efficiency.

In our previous work,Multi-Channel Gliding Arc(MCGA)could be used to restrain the mode transition of combustion at Ma=2.92.29The transition of GA discharge mode and electrode temperature have been studied and result showed that spark discharge was important in ignition.30The ignition processes of single-channel GA and MCGA were also compared and the result showed that the MCGA on ignition is more remarkable with the growth of Global Equivalence Ratio(GER).31The energy of GA on ignition was controlled and it was found that ignition energy was very important to the success of ignition at the initial flame formation stage.32Direct ignition and reignition were also found in the ignition process.33.

Ignition at high total temperature and high Mach number of combustion chamber inlet flow under supersonic conditions have been studied extensively.However, it is more difficult to ignite the scramjet combustor with inlet airflow at a low total temperature and a low Mach number.It is possible to achieve ignition under these conditions with MCGA.This paper investigates the ignition process by MCGA under different GERs at Ma=2.0 and a total temperature of 940 K.The dynamic evolution process of flame kernel generation,initial flame development, and flame establishment are also studied.High-speed CH* chemiluminescence images from the side and top of the cavity and the discharge characteristic of MCGA was collected synchronously to study the ignition process.Combined with temperature, velocity, and equivalence ratio distribution of non-reactive flows simulated by the Reynolds Averaged Navier-Stokes (RANS), the experimental phenomenon was explained reasonably.

2.Experimental and simulation descriptions

2.1.Structure of model scramjet combustor

The scramjet model consists of the air heater,the Laval nozzle,the isolation section, the combustor, and the expansion section.The experimental study was carried out on a scramjet model at the National University of Defense Technology(NUDT).34The Mach number of airflow into the combustor is 2.0,the total pressure is 1.08 MPa and the total temperature is 940 K.Inlet flow parameters of the combustor are achieved by burning the mixture of oxygen, alcohol, and air in the air heater.Alcohol has been completely burned in the heater,and the detailed combustor inlet air flow parameters are shown in Table 1.

The structure of the model scramjet combustor is shown in Fig.1.The inlet of the combustor cross-section is 40 mm(height)and 50 mm(width).There are four 1 mm diameter vertical fuel injectors located 15 mm from CLE.The cavity depth is 20 mm,the width is 50 mm,the length-depth ratio is 4.5,the closeout angle is 45°,and the expansion angle is 2°.The igniter of MCGA is on the central axis of the cavity bottom wall and 20 mm away from CLE.Streamwise(x),spanwise(y),and bottom wall-normal direction (z) are also marked in Fig.1.

As shown in Fig.2(a), the structure of MCGA igniter is composed of tungsten needles, ceramic, and iron casing.The iron casing is 40 mm in diameter, and the ceramic is 35 mm in diameter.The high voltage anode is composed of six tungsten needles with a diameter of 1 mm, which are connected to power.The cavity wall is grounded and acts as cathodes.The MCGA is produced by achieving a breakdown between the six high voltage anodes and the nearest cavity bottom wall.The instantaneous image of the MCGA discharge in the combustor is shown in Fig.2(b), and a six-channel gliding arc can be seen.The MCGA can be blown to move along the iron casing of the igniter by the cavity flow in a supersonic flow.

2.2.Schematic of measurement system

An AC power supply (CG-10000F) was used to generate the MCGA.The input voltage of the power supply can be adjusted from 220 V to 380 V.The oscilloscope (Tektronix DPO4104)was used to record the characteristics of the MCGA discharge in real-time.The current and voltage of MCGA were recordedby voltage (P6015A) and current (Pearson 6600) probe.Camera A (FASTCAM SA-X2) and Camera B (NAC Memrecam HX-7S) were synchronized to record the ignition process of the scramjet ignited by the MCGA from the two-angle view(the side and the top)of the scramjet combustor.The two cameras were triggered synchronously by a digital delay/pulse generator (DG535).

Table 1 Combustor inlet flow parameters in test.

Fig.1 Schematic of model scramjet combustor (there are four uniformly distributed injectors with a diameter of 1 mm,which has been marked in red).

As shown in Fig.3,the simultaneous recording of the Cameras A and B is achieved using an oscilloscope and a pulse generator.The MCGA discharge triggers the oscilloscope,and the Camera A placed at the side of combustor is trigged by the oscilloscope.The Camera A triggers the pulse generator, and finally the Camera B suspended from the top of combustor is triggered by the pulse generator.The camera gate signals of the two cameras are also recorded by the oscilloscope at the same time to judge whether the current and voltage waveform is synchronized with the real-time photography.

Camera A is equipped with an f/5.6 Tokina lens and Camera B with an f/11 Nikon lens.Both cameras are equipped with a band filter(430±10 nm)to collect CH*chemiluminescence images in the ignition process.The exposure time of both cameras was 40 μs and the shooting frequency was 20 kHz.The fuel pre-injection pressure and GER used in the experiment are shown in Table 2.The GER is defined as the ratio of the amount of air required by the theory of complete combustion of all ethylene injected into the combustor from fuel injector to the real amount of all air supplied.The Local Equivalence Ratio (LER) is defined as the ratio of the amount of air required by the theory of complete combustion of local ethylene in the combustor to the real amount of local air supplied.Five tests with different GERs were studied, and each set of the tests was repeated at least two times.

2.3.Computational method

Fig.3 Schematic of experimental device location (red line is oscilloscope triggering two cameras synchronously).

Table 2 Fuel pre-injection pressure and global equivalence ratio in test.

To further investigate the mechanism of MCGA ignition process in scramjet combustor, distributions of temperature,velocity, and equivalence ratios in the non-reactive flows were simulated by using the method of three-dimensional numerical simulation based on RANS and Shear Stress Transport (SST)k-ω turbulence model.The code used was developed in-house by NUDT and the detailed procedure can be found in Refs.35–37.The Advection Upstream Splitting Method(AUSM) + -up spatial discretization method was also used and Courant-Friedrichs-Levy (CFL) number was 0.1.This simulation program has been used in our previous studies.29,31,38,39The scramjet model size and airflow parameters in the simulation are consistent with the experimental setup.

The grid independence verification is shown in Fig.4.The selected streamwise plane(x-z)is 7 mm to the left of the central axis of the cavity,which is the fuel injector section.The results show that the pressure distribution of the scramjet under the non-reactive flows of different grid cells was reasonable consistency.Therefore, in order to save the calculation time and reduce the amount of calculation, the medium grid is selected for calculation.

Fig.2 Schematic of MCGA igniter and discharge in static air.

Fig.4 Simulated pressure distribution on bottom wall of scramjet combustor under non-reactive flows in Test 1 (position of CLE is set as 0 mm).

3.Results and discussion

3.1.Characteristics of MCGA discharge in ignition

To study the discharge characteristics of MCGA during ignition, high-speed photography and discharge waveforms are simultaneously collected.Fig.5 shows the voltage and current waveform of MCGA within 600 μs at GER = 0.14.During the discharge period from 0 μs to 600 μs, many sharp peaks appear in the current and voltage waveform, showing the occurrence of spark-type discharges because of the turbulence affected in the cavity.40At 0 μs, as shown in Fig.5, a current peak appears in the range of the camera gate.Figs.6(a)and(b)show CH* chemiluminescence images in the ignition process used by MCGA from the side and the top of the combustor,which corresponds to voltage and current waveforms shown in Fig.5.At 0 μs, as shown in Fig.6, a bright GA appears at this time, which is due to the breakdown between the electrodes.During the ignition time from 0 μs to 600 μs, the maximum instantaneous peak voltage of the GA is 15.4 kV and the maximum instantaneous peak current is 12.2 A.

3.2.Ignition process by MCGA discharge

Fig.5 Voltage and current waveforms of MCGA discharge from 0 μs to 600 μs at GER = 0.14.

The two-angle images of the MCGA ignition process at GER = 0.14 are shown in Fig.6, which were recorded by the two cameras placed at the side and the top of the scramjet combustor.All images are normalized by dividing the pixels in the image by their maximum pixel value.The unit of the legend is arbitrary unit (a.u.).At 0 μs, a bright GA can be found.At 100 μs, the generation of a flame kernel can be observed, and the flames attach to the MCGA are defined as flame kernels.33The CH* chemiluminescence of MCGA is stronger than that of the flame kernel in 0–500 μs.Thus, the yellow regions represent MCGA and the surrounding blue regions represent the flame kernel in Fig.6.At 300 μs,the flame is separated with MCGA to form the initial flame.The generated initial flame mainly develops along the flow direction to CLE.The initial flame is marked by the dotted line in Fig.6(P7).At 400 μs,the generated initial flame develops to CLE and begins to develop towards the Cavity Trailing Edge (CTE).Meanwhile,a new initial flame is formed around the MCGA, which is marked with a dotted line in Fig.6(P9).At 500 μs, the flame reaches the side wall near CLE and begins to develop downstream along main stream direction.At 800 μs, the flame fills the cavity and a global flame is established.

To further study the ignition process at a higher GER, the ignition process at GER = 0.17 was studied.As shown in Fig.7, a bright GA can be observed at 0 μs.The generation and development of a flame kernel are observed at 100 μs.At 200 μs, an initial flame is separated from the MCGA, and the initial flame is closer to the sidewall,as indicated by dotted lines in Fig.7(P3).At 300 μs,the flame develops to CLE along with airflow and forms a resident flame.Then the flame begins to develop towards the walls on both sides of the cavity.At 600 μs, the flame fills the whole cavity.At 800 μs, the flame in cavity angular backflow zone begins to disappear and develops toward the mainstream flame.Compared with GER=0.14,the flame at GER=0.17 can fill the whole cavity faster.With the increase of GER, more ethylene fuel is injected into the cavity.Due to the effect of turbulence on fuel mixing,fuel distribution in the cavity is more extensive,and so LER in the cavity is also increased correspondingly.These factors play an important role in accelerating the ignition process.

To further study the ignition process at a lower GER, the ignition process at GER=0.06 was investigated.Fig.8 shows the normalized flame area at GER=0.06.The flame area during the ignition process is summed from the total pixels in each image and the effects of the emission from the MCGA are excluded.The normalized flame area is calculated by dividing the flame area of each image by the average global flame area.The global flame area is defined as the averaged flame area in the stable combustion stage.33The MCGA ignition process is divided into three stages,which have been marked in Fig.8(a).The first stage is the flame kernel generation, the second stage is the flame development,and the third stage is the stable combustion.These three stages also exist in other GER, and the biggest difference between them lies in the different periods in the first stage.

In the stage of flame kernel generation(0–6.4 ms),the flame kernel can be continuously generated around the MCGA.But due to the influence of unfavorable environments of LER,the flame kernel is difficult to develop into the initial flame.Fig.8(b) is a partially enlarged view from 6.0 ms to 7.6 ms of the first and second stages of the ignition process in Fig.8(a).The CH* chemiluminescence images of the eight points S1-S8 marked in Fig.8(b) have been shown in Fig.9.At 6.3 ms, the flame kernel formation can be seen around the MCGA in Fig.9(S1) and then blown out due to the effect of turbulence at 6.4 ms, the phenomenon of which can be seen in Fig.9(S2).In this stage, the flame kernel is continuously formed around the MCGA,but it is hardly self-sustained after leaving the MCGA.

Fig.6 Ignition process with MCGA at GER = 0.14 (fuel injection location has been marked with red dots, and white dotted line in image represents visual field from top of cavity).

Fig.7 Ignition process with MCGA at GER = 0.17.

Fig.8 Flame area fluctuations with time during ignition process at GER= 0.06(image of the maximum flame area has been marked,and red dotted line indicates stage from initial flame generation to the maximum flame area).

Fig.9 Flame development and change process at GER = 0.06.

In the stage of flame development (6.5–7.5 ms), the flame directly generated at CLE is observed at 6.5 ms in Fig.9(S3).This may be because the heat carried by the combustion products makes the local temperature rise, and that MCGA constantly ignites new flame kernel.At the same time, part of the unburned fuel and combustion products with high temperatures gather at CLE.This is due to the presence of cavity recirculation zone,which forms a favorable condition for ignition.27Therefore, self-ignition is realized at CLE.The generated flame moves to CTE under the action of airflow at 6.9 ms in Fig.9(S6).The flame that develops from CLE ignites the fuel near the cavity close-out ramp in Fig.9(S7).At this time,the flame area is half that of the stable combustion stage.The flame, part of the fuel, and part of the high-temperature combustion products near the cavity close-out ramp move towards CLE under the effect of the cavity recirculation zone, which provides favorable conditions for the establishment of the global flame.23As shown in Fig.9(S8), the maximum flame area is reached at 7.5 ms.In the stage of stable combustion, the flame in the combustor always fluctuates due to the unstable and turbulent flow, which will lead to the oscillation of the flame area.

To compare the flame combustion condition in detail, the intensity of the flame at different equivalence ratios was compared.The total pixels value of the image can be used as a marker of the heat release or the change of combustion since the chemiluminescence comes primarily from excited species CH*.30Fig.10 shows the normalized flame intensity with time at different GER.The time when the MCGA discharge is generated is set as t = 0 μs.The flame experiences 800 μs when reaching the maximum intensity value and its maximum flame intensity value is 3.8 times that of the average stable combustion stage at GER = 0.14.As the equivalence ratio increases,the flame experiences 700 μs when it reaches its maximum intensity value at GER=0.17.The effect on the time of flame development stage is small with the GER being decreased from 0.17 to 0.14.As the GER decreases from 0.14 to 0.08,the flame reaches the maximum flame intensity value after the 1250 μs at GER = 0.11.As the GER continues to decrease, the flame intensity value reaches its maximum after 2500 μs at GER = 0.08.It can be seen that the influence of the GER on the ignition process is mainly in the flame kernel generation stage.

3.3.Flame propagation trajectories

Fig.10 Normalized flame intensity for MCGA ignition process under different GERs.

To compare the flame propagation trajectory of the scramjet combustor at different GERs, the images ranging from 0 μs to 5000 μs are integrated along the vertical direction.Each row represents an image, and a black row is inserted between two adjacent image rows in order to distinguish them.The CLE is defined as 0 mm,while the ordinate represents the time process.Fig.11(a) shows the flame development trajectory in the time range of 0 μs to 5000 μs at GER = 0.17.The flame kernel is generated within the width range of 20–45 mm, and the generated initial flame moves towards CLE under the action of airflow in the cavity.At 400 μs, the flame reaches CLE, and after 100 μs, the flame begins to develop towards CTE in the streamwise direction.A row of pixels represents the flame fills the entire cavity at 600 μs, which is consistent with Fig.7.The flame reaches its maximum area and spreads beyond CLE towards the fuel injection position at 700 μs and forms a stable mainstream flame after 2000 μs.Fig.11(b)shows the flame propagation trajectory at GER = 0.14 and it is almost consistent with the flame movement trajectory in Fig.11(a).The flame propagation trajectories at GER = 0.11 and GER = 0.08 are shown in Figs.11(c) and(d), respectively.Compared with GER = 0.17 and GER = 0.14, the time from flame kernel generation stage to flame development stage is longer, and costs 1200 μs and 2500 μs respectively.The flame generated by MCGA is always within the range from 15 mm to 45 mm of the cavity,which is the front part of the igniter.

To study the flame trajectory observed from the top of the scramjet combustor, the same method is used to process images.Fig.12 shows the CH* chemiluminescence images integrating along the spanwise direction from 0 μs to 2500 μs.Its ordinate represents the width of the cavity,the central axis of the cavity is 0 mm,the abscissa represents the time change, and the dotted line represents the edge of the field of view of the observation window at the top.Fig.12(a) shows the flame trajectory recorded from the top of the combustor at GER = 0.17.Before the flame reaches CLE, there is no obvious flame propagation phenomenon in the spanwise direction.The flame is distributed in the range of-5 mm to 15 mm of the cavity and the flame width is less than 20 mm in the spanwise direction, occupying less than 1/2 of the width of the cavity.This may be due to the generation of recirculation zone on both sides of CLE.Under the influence of the angular recirculation zone of CLE,the initial flame is gathered near the central axis of the cavity.It is not until 400 μs later that the flame begins to develop towards the side of the combustor.Combining with Fig.7(b), the flame just reached CLE at 400 μs.In the subsequent flame propagation process,the cavity is always full of flame in the spanwise direction.Fig.12(b)shows the flame trajectory observed from the top of the cavity at GER = 0.14.the generated initial flame is mainly distributed in the range of about 10 mm along the cavity central axis of the streamwise direction.The flame width is about 1/2 of the width of the cavity and wider than GER = 0.17 at the same time.

Fig.11 Longitudinal integral diagram of images taken from side of cavity at different equivalence ratios.

Fig.12 Transverse integral diagram of image is taken from top of cavity at different equivalence ratios.

3.4.Mechanism of MCGA ignition process and flame development

MCGA can generate six channels of gliding arcs, which increases the contact area between the discharges and the fuel/air mixture.Meanwhile,MCGA can produce many flame kernels,and the different flame kernels can be merged to form a large initial flame, thus accelerating the ignition process.32The thermal and chemical effects of MCGA plasma can enhance the mixing and heat release of fuel,and further accelerate the ignition process.33,41

Fig.13 Velocity magnitude distribution of non-reactive flows.

Fig.13 shows the distribution of velocity magnitude in the cavity at Ma=2.0.The selected spanwise plane(x-y)is 20 mm away from the bottom wall of the cavity.As shown in Fig.13(a), when the airflow reaches the close-out ramp of the cavity along the mainstream, part of the airflow enters the cavity and moves towards CLE.Then the airflow is lifted and reenters the main stream to complete a cycle.The flow forms cavity recirculation zone, and the velocity in the cavity is lower than mainstream, which is more suitable for the generation and development of flame.The simulated flow trajectory is similar to the experimental results of the flame moves toward CLE at the beginning in Fig.11.The experimental results in Fig.7 show that the flame moves upward after moving to CLE,then develops near the cavity shear layer, and finally develops to CTE.The simulated results are similar to the experimental results in Fig.7.As shown in Fig.13(b), the flow that enters the cavity recirculation zone will generate angular recirculation zone near both sides of CLE.The experimental results in Fig.12 show that the flame generated by MCGA is initially gathered near the central axis of the cavity and the flame width in the spanwise direction is only about 50%of the width of the cavity.The flow on both sides will move towards the central axis of the cavity in the simulation results, which is similar to the experimental results in Fig.12.

The selected streamwise plane(x-z)in Fig.14 is 7 mm to the left of the central axis of the cavity, which is the fuel injector section.The simulated temperature magnitude distribution in the cavity is shown in Fig.14(a).In the supersonic nonreactive flows, the temperature is significantly higher in the cavity than that in the mainstream, and the temperature near CLE is the highest.This explains the experimental phenomenon in Fig.9 in which the flame is generated directly at CLE.Figs.14(b) and (c) show the LER distribution in cavity at GER=0.17 and GER=0.08,respectively.It is found that the LER in the cavity is higher near the cavity close-out ramp and the cavity bottom wall.The decrease of GER will lead to the decrease of the LER in the cavity,which is the main reason for the different times from flame kernel generation to flame development stage in Fig.10.

4.Conclusions

In conclusion, investigations of the MCGA-assisted ignition process and flame propagation of a scramjet at Ma=2.0 were carried out by using simultaneous two-angle CH* chemiluminescence imaging,electrical measurements,and RANS simulations.Effects of the equivalence ratios on the ignition and propagation characteristics were revealed, and the following conclusions can be drawn:

Fig.14 Magnitude distribution of non-reactive flows in streamwise plane.

(1) Successful ignitions of the scramjet combustor at Ma = 2.0 with the GER between 0.06 and 0.17 were achieved by using MCGA, showing that the MCGA has strong ignition ability and can achieve reliable ignition of a scramjet.

(2) The MCGA ignition process is composed of flame kernel generation stage, flame development stage, and stable combustion stage.With the increase of GER from 0.08 to 0.14,the development time from the flame kernel generation to the flame development stage gradually decreases, but with the GER increasing from 0.14 to 0.17, the development time does not improve much.

(3) In the streamwise direction,the initial flame develops to CLE due to the influence of airflow in the cavity, then uplifts into cavity shear layer,and global flame is finally formed along the direction of the main stream.In the spanwise direction,the flame is distributed near the central axis of the cavity and its width is less than 50% of the width of the cavity before reaching CLE.Then the flame begins to develop towards two sides of combustor after reaching CLE, which is affected by the angular recirculation zone on both sides of CLE.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (Nos.12172379, 11925207, and 91741205) and the Foundation for Outstanding Young Scholars of National University of Defense Technology, China.