Experimental study on atomization phenomena of kerosene in supersonic cold flow

Transcription

1 Science in China Series E: Technological Sciences 2008 SCIENCE IN CHINA PRESS Springer tech.scichina.com Experimental study on atomization phenomena of kerosene in supersonic cold flow FEI LiSen, XU ShengLi, WANG ChangJian, LI Qiang & HUANG ShengHong Department of Mechanics and Mechanical Engineering, University of Science and Technology of China, Hefei , China Experiments were conducted to study the atomization phenomena of kerosene jet in supersonic flow. The kerosene jet was driven by compressed nitrogen. Meanwhile, the shadowgraph and planar laser-induced fluorescence (PLIF) were used to visualize the flow field in the case of different total pressure and jet pressure. The results imply the followings: The combination of shadowgraph and PLIF is a reasonable method to study the atomization phenomena in supersonic flow. PLIF can detect the distribution of kerosene droplets accurately. Shadowgraph can visualize the wave structure. Higher jet-to-freestream dynamic pressure initiates higher penetration height and the jet column will be easier to breakup and atomize, but it also induces stronger shock waves and aggravate total pressure lost. Three-dimensional, unsteady surface wave plays an important role in making the jet break up and atomize. Higher jet-to-freestream dynamic pressure will accelerate the development of surface wave and enlarge the amplitude of surface wave, while lower jet-to-freestream ratio will inhibit the development of surface wave. kerosene, supersonic flow, atomization, shadowgraph, planar laser-induced fluorescence 1 Introduction Scramjet which employs liquid hydrocarbon fuel is an attractive choice for hypersonic vehicle. Compared with gaseous fuel, liquid fuel has larger density and deeper penetration; it is safer and easier to deposit. But liquid fuel has complex breakup and atomization process before ignition. This process correlates incoming flow conditions, injection condition and structure of combustor and injector. A lot of results implied that liquid atomization process in supersonic flow was strongly non-linear and scale depended. It is a key problem in supersonic combustion which employs kerosene. Received July 31, 2006; accepted April 16, 2007 doi: /s Corresponding author ( forest@mail.ustc.edu.cn) Supported by the Fund for the Application of the Combination of Plif and Schlieren Methods in the Study of Mixing Enhancement by Newtype Cavity in Supersonic Combastion Field (Grant No ) Sci China Ser E-Tech Sci Feb vol. 51 no

2 A lot of work has been done on liquid jet. Forde et al. [1] studied deformation and trajectory of liquid jet in supersonic flow. Schetz et al. [2] studied breakup of liquid jet column and the role of surface wave in breakup of jet column. Nejad et al. [3,4] studied the effects of properties and location in the plume on droplet diameter for injection in supersonic flow. Xu et al. [5] studied unsteady transverse kerosene jet in supersonic flow filed. The above work mainly employed traditional optical diagnostic method such as high speed camera, schlieren/shadowgraph, etc. These methods can visualize the shock wave structure induced by the jet and qualitatively identify the breakup, atomization and mixing of the jet. Because of the integration effects, it is impossible to accurately identify the boundary of the jet or study the details of diffusion of the fuel. So, before making substantially progress on theoretical model, it is so necessary to develop a subtle optical diagnostic method that we can understand the kerosene atomization in supersonic flow deeply. In the present investigation, PLIF was applied to kerosene atomization in supersonic flow. Kerosene is a kind of complex mixture, which would emit fluorescence under excitation [6]. Fluorescence from kerosene droplets can be captured to represent the distribution of kerosene droplets, so the details about jet boundary and kerosene diffusion could be achieved. It is the key to study atomization of kerosene in high-speed flow by PLIF. Recently, PLIF has been used broadly in measurement of gaseous or liquid diffusion and mixing [7 9]. In most of these works, tracers were added to visualize the position of material of interest, such as NO in gaseous material and acetone in liquid. In our investigation, we chose kerosene itself as the tracer to simplify the experimental method and avoid non-uniform effects of the tracer in mixture and the effects of different physical and chemical properties of the tracer. The exposure time of PLIF was very short (nano second order), so PLIF has high temporal resolution and the motion of flow could be frozen effectively. PLIF uses laser sheet technology whose thickness is less than 1 mm to avoid integration effects, so PLIF has high spatial resolution. PLIF is a non-intrusive technology, it is suitable for study of supersonic combustion. Shadowgraph method was also used to study the wave structure in the flow field. The comparison between results from shadowgraph and PLIF could be made to study the atomization and mixing phenomena of kerosene in supersonic flow. 2 Apparatus and instruments Our experimental system consisted of wind tunnel, kerosene jet system, shadowgraph system and PLIF system. The blow down wind tunnel included: 1) gas tank whose volume was 90 m 3 and the maximum pressure was up to 1.2 MPa; 2) 2D nozzle whose Mach number was 2.2; 3) test section that had a rectangular test section measuring 50 mm wide, 56 mm high and 500 mm long. The test section had quartz windows in both side walls as well as the top wall for laser diagnostics and flow visualization. A small orifice whose diameter was 0.3 mm was drilled 70 mm away from the entrance. Kerosene jet was driven by pressurized nitrogen, the jet was controlled by electromagnetic valve whose maximum working pressure was up to 4 MPa. A schematic diagram of the nozzle and the test section is shown in Figure 1. The PLIF system consisted of three parts: 1) the light source system which included YAG laser device, dye laser device and frequency doubler. Nd:YAG laser device pumped the dye laser device at wavelength 532 nm and 800 mj/pulse. The dye laser device employed Rhodamine 6G as the dye whose fluorescence ranged from 559 to 576 nm with peak at 566 nm. The dye laser from 146 FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no

3 dye laser device was at wavelength 284 nm and 30 mj/pulse. 2) The beam path system. The dye laser was expanded to a laser sheet by cylindrical lens whose focal length was 20 mm and converged by convex lens whose focal length was 500 mm to a thin sheet measuring 60 mm wide and 0.3 mm thick. 3) The signal acquisition and processing system consisted of ICCD, time delayer (DG535), computer and image processing software. The optical alignment is shown in Figure 2. The laser sheet entered the test section from the side wall (Figure 2(a)) and top wall (Figure 2(b)) respectively. The flash lamp of the Nd:YAG laser device was triggered by external signal, the output Q-switch signal triggered DG535, the ICCD was triggered by DG535 after preset delay time to ensure that the shutter of ICCD was open during the lifetime of fluorescence. In order to increase signal-to-noise ratio, we adopted a high pass filter whose pass band ranged from 300 to 550 nm to avoid the effects of light signal besides fluorescence from kerosene. Figure 1 Schematic of the nozzle and the test section. 1, Nozzle; 2, test section; 3, quartz windows; 4, kerosene jet unit. Figure 2 Schematic of the optical alignment of the PLIF system. (a) Laser sheet entered the test section from the side wall; (b) laser sheet entered the test section from the top wall. 1, Nd:YAG laser device; 2, dye laser device; 3, reflector; 4, cylindrical lens; 5, slot; 6, convex lens; 7, test section; 8, ICCD; 9, filter. In this investigation, we changed the total pressure of the wind tunnel and the jet pressure, so FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no

4 we could study the penetration, deformation of the cross section and the effects of surface wave on breakup and mixing of the jet. 3 Results and discussion 3.1 Fluorescence from kerosene Kerosene is a kind of mixture, whose emission spectra is also complex. The kerosene was pumped by 284 nm light, generated by combination of high voltage mercury arc lamp and monochrometer, and a high pass filter (light of wavelength shorter than 340 nm was denied) was used to avoid the effects of pump light. The emission spectra are shown in Figure 3. As can be seen from Figure 3, the red curve represents the background signal; the peaks correspond to the characteristic peaks of mercury arc lamp. The green curve represents the signal from kerosene. By comparison with the background signal, we can find that the emission peaks of kerosene are located at 350 and 380 nm. Figure 4 shows the fluorescence of atomized kerosene in quiescent air that was pumped by laser sheet of wavelength was 284 nm. The laser sheet entered the droplets group at 100 mm away from the injector. In order to avoid the effects the scattered laser, a filter with pass band ranging from 300 to 550 nm was located in front of the ICCD to deny the pump laser whose wavelength was 284 nm. As can be seen from Figure 4, strong fluorescence signal could be generated under excitation, which also represents the distribution of kerosene droplets. Figure 3 Emission fluorescence spectrum of kerosene. Figure 4 Image of fluorescence from kerosene droplets. 3.2 Atomization phenomena of kerosene in supersonic flow Results by shadowgraph are given in Figure 5, where the total pressure of wind tunnel was 0.3 MPa and the jet pressure was 1.0 MPa. As can be seen from Figure 5, the incoming flow from right to left was supersonic, so shock wave emerged in front of the jet column. Because the single orifice was very small, no evident boundary layer separation could be observed, the circulation region was so small that the detachment shock was very weak even decayed to compressive wave. Near the exit of the orifice, the luminance was low, indicating that the jet did not diffuse sufficiently in this area, the concentration of the kerosene was high. The luminance increased gradually along the flow direction, this indicated that the jet had begun to diffuse. The jet penetration 148 FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no

5 height increased along with the distance from the orifice gradually, but more and more slowly. In contrast to Figure 5, Figure 6 shows the distribution of the kerosene fluorescence at the cross section that was normal to the jet axis. The laser sheet entered the test section 90 mm away from the orifice (the dark line in Figure 5(a)), the ICCD was located at the same side of the laser sheet. As can be seen from Figure 6, the bright stripes and spots were fluorescence from the kerosene on the quartz windows and the facula at center was fluorescence from kerosene jet. The shape of the cross section was irregular, obvious wave structure could be observed at the left edge of the cross section. The luminance at the cross section was not uniform but was high at the center and decreased along the radial direction gradually. This indicated that the concentration of the kerosene decreased along the radial direction, and high concentration core existed in the jet column. The facula diffused to the bottom wall indicated that there was some kerosene flux at the bottom wall. Figure 5 Shadowgraph image of the flow field (incoming flow was from right to left). (a) The entire image; (b) partial enlarged drawing. Figure 6 Kerosene fluorescence at the cross section that was normal to the jet axis. a, Kerosene fluorescence from the kerosene that was adhered to the side window; b, kerosene fluorescence from the jet. Figure 7 shows the distribution of kerosene fluorescence at the cross section along flow direction. The laser beam was expanded by cylindrical lens and converged at the thickness direction. A laser sheet measuring 200 mm long and less than 0.3 mm thick was generated and entered the test section. As can be seen in Figure 7, surface wave appeared on the jet surface. Vortex structure behind the jet column was an important reason of the formation of surface wave [10]. Because of the vortex structure, the pressure at the opposite sides of the jet column was imbalanced to make the jet column to oscillate periodically and generated the surface wave. The surface wave was three dimensional as can be seen from Figures 6 and 7. The amplitude along the axial direction was larger than that along endwise direction. These axial, large amplitude waves initiated at the position where the jet began to bend (point 1 in Figure 7). The surface waves enlarged the area where the freestream interacted with the jet, increased the interaction between the jet and freestream, so it made the jet strongly unstable. Therefore, surface wave was the dominant mechanism in jet decomposition. The luminance of the fluorescence far away from the injector was low, which indicated that the kerosene concentration was low. This implied that the jet column had begun to break up, even FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no

6 atomize to fine droplets. At the wall near the orifice there existed a dark area, it indicated that the kerosene concentration was low near the lee side of the jet column. A brighter area existed behind the dark area, indicating the jet impinged on the bottom wall under the effects of the high speed incoming flow. Figure 7 Kerosene fluorescence at the cross section along the flow direction. In contrast to the results by PLIF, the jet boundary obtained by shadowgraph was smooth (Figure 5); no evident surface wave could be observed. The reason was the exposure time of shadowgraph was long (4 ms), so that the flow could not be frozen. The integration effect was another reason. Three-dimensional information was overlapped on a two-dimensional image, so it is difficult to identify the jet boundary accurately. There existed a great difference in identifying the jet boundary by PLIF and shadowgraph. At the position 90 mm away from the injector, the penetration heights obtained from Figures 5 7 were 6.2 mm (Figure 5), 12.1 mm (Figure 6) and 12.6 mm (Figure 7) respectively. Obviously, shadowgraph method underestimated the penetration height. The reason is that the shadowgraph method is sensitive to second derivative of the density in the flow field. When the concentration of the droplets changed gradually, the luminance would relatively uniform, so it became difficult to identify the distribution of the droplets. This phenomenon became more evident at the jet boundary. In contrast to shadowgraph, PLIF used the signal from kerosene droplets, so it could visualize the kerosene distribution even the concentration was very low. The difference between the results from Figure 6 and Figure 7 arose from the effects of the unsteady surface wave on the jet, so the penetration height at the same position would vary with time, and the variation correlated with the amplitude of the surface wave. 3.3 Effects of total pressure and jet pressure In this investigation, we changed the total pressure of the wind tunnel P t and the jet pressure P j. Images were obtained by shadowgraph (Figure 8) and PLIF (Figure 9) where R was the 2 2 jet-to-freestream dynamic pressure ratio which was defined as R = ρjvj / ρ V. Figure 8 Shadowgraph images at different total pressures and jet pressures. (a) P t =0.6 MPa, P j =1.0 MPa, R=3.24; (b) P t =0.6 MPa, P j =0.5 MPa, R=1.55; (c) P t =0.3 MPa, P j =1.0 MPa, R=6.48; (d) P t =0.3 MPa, P j =0.5 MPa, R=3.11. As can be seen from Figure 8, the penetration height increased as R increased. When R increased from 1.55 to 6.48, the penetration height 90 mm away from the injector also increased 150 FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no

7 from 4.9 to 6.2 mm. But it is not practical to increase the penetration height by only increasing R. When R was high, the surface wave could be observed, but when R was low, the jet boundary seemed smoother. The luminance at the boundary of the jet was high. This implied that the jet column had already broken and atomized, so it is difficult to distinguish the jet boundary. Wave structures were also affected by R. The shock angle was 37 when R was The shock was a typical bow shock. The shock angle was 31 when R was 1.55 and the shock was almost oblique shock. The reason was the higher penetration height at higher jet-to-freestream dynamic ratio. It could induce stronger shock wave and cause more total pressure lost. This is a real problem in designing a combustor. Compared Figure 8 to Figure 9, it is implied that the penetration height increased with increasing R more clearly. When R increased from 1.55 to 6.48, the penetration increased from 6.2 to 12.1 mm at the position 90 mm away from the injector. The jet deformed under the effects of freestream. The higher the R, the more evident the deformation. When R was relatively low (Figure 9(a) and (b)), the cross section of the jet was almost elliptic. When R was relatively high (Figure 9(c) and (d)), the amplitude of the surface wave increased obviously, the cross section became more irregular, and the diffusion region was larger than that at low jet-to-freestream ratio. It implied that the jet column had already broken up seriously. Figure 9 Kerosene fluorescence at the cross section that was normal to jet axis. (a) P t =0.6 MPa, P j =1.0 MPa, R=3.24; (b) P t =0.6 MPa, P j =0.5 MPa, R=1.55; (c) P t =0.3 MPa, P j =1.0 MPa, R=6.48; (d) P t =0.3 MPa, P j =0.5 MPa, R=3.11. In contrast to Figure 9, fluorescence images at the cross section that was along the flow direction can be seen from Figure 10. Figure 10 implied that penetration height increased as R increased. When R increased from 1.55 to 6.48, the penetration height at the position 90 mm away from the injector increased from 5.5 mm (Figure 10(b)) to 12.6 mm (Figure 10(c)). When R was low (Figure 10(a) and (b)), the fore part of the jet was smooth, no obvious surface waves appeared; when R was high (Figure 10(c) and (d)), surface waves appeared near the jet orifice and developed downstream rapidly. These implied the jet was more stable when R was low. The in- FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no

8 tensity of fluorescence was low far downstream to represent the kerosene droplets concentration was also very low. This obviously indicated that the jet column had already broken up even atomized to fine droplets. Figure 10 Kerosene fluorescence at the cross section along the flow direction. (a) P t =0.6 MPa, P j =1.0 MPa, R=3.24; (b) P t =0.6 MPa, P j =0.5 MPa, R=1.55; (c) P t =0.3 MPa, P j =1.0 MPa, R=6.48; (d) P t =0.3 MPa, P j =0.5 MPa, R= Conclusions (1) The combination of PLIF and shadowgraph is a reasonable method to study the atomization phenomena in supersonic flow. PLIF can detect the distribution of kerosene droplets accurately and afford details of the jet. It is helpful to deepen the understanding of breakup, atomization and diffusion phenomena of liquid jet. Shadowgraph can visualize the wave structure which is a good supplement to PLIF. (2) Jet-to-freestream dynamic pressure ratio plays an important role in determining the penetration height and the breakup and atomization process. The penetration height is higher and the breakup and atomization process is faster at high jet-to-freestream dynamic pressure ratio, but stronger shock would be induced to make the total pressure lost higher. (3) The surface wave is the dominant factor in breakup of the liquid jet. Low jet-to-freestream dynamic pressure ratio would inhibit the development of the surface wave. The wave appears after a long smooth part. When the ratio is high, surface wave appears where the jet begins to bend downstream. The surface wave is three-dimensional and unsteady, the amplitude of the wave becomes large when the ratio is high. In order to understand the details of the atomization of liquid jet, quantitative study on distributionof droplet diameters is necessary and the work is in progress. 1 Forde J M, Molder S, Szpiro E J. Secondary liquid injection into a supersonic airstream. AIAA J, 1966, 3(8): Schetz J A, Kush E A, Joshi P B. Wave phenomena in liquid jet breakup in a supersonic crossflow. AIAA J, 1980, 18(7): Wu P K, Kirkendall K A, Fuller R P, et al. Spray structures of liquid jets atomized in subsonic crossflows. J Propul Power, 1998, 13(2): Nejad A S, Schetz J A. Effects of viscosity and surface tension on a jet plume in supersonic crossflow. AIAA J, 1984, 22(4): Xu S L, Archer R D, Milton B E, et al. Unsteady transverse injection into a supersonic flow. Sci China Ser E-Tech Sci, 2000, 43(2): Schmidtl G, burkert A, Triebel W. A measurement system to investigate droplet size distribution in dense kerosene sprays. Technisches Messen, 2003, 70 (1): [DOI] 7 Yip B, Miller M F, Lozano A, et al. A combined OH/acetone planar laser-induced fluorescence imaging technique for visualizing combusting flows. Exp Fluids, 1994, 17: [DOI] 8 Gauba G, Klavuhn K G, McDaniel J C, et al. OH planar laser-induced fluorescence velocity measurements in a supersonic combustor. AIAA J, (4): Lozano A, Yip B, Hanson R K. Acetone: a tracer for concentration measurement in gaseous flows by planar laser-induced fluorescence. Exp Fluids, 1992, 13: [DOI] 10 Kamotani Y, Greber I. Experiments on a turbulent jet in a crossflow. AIAA J, 1972, 10(8): FEI LiSen et al. Sci China Ser E-Tech Sci Feb vol. 51 no