- Email: [email protected]
Combustion characteristics in a supersonic combustor with hydrogen injection upstream of cavity flameholder
Available online at www.sciencedirect.com
Proceedings of the
Proceedings of the Combustion Institute 34 (2013) 2073–2082
Combustion Institute www.elsevier.com/locate/proci
Combustion characteristics in a supersonic combustor with hydrogen injection upstream of cavity flameholder Hongbo Wang a, Zhenguo Wang a,⇑, Mingbo Sun a,⇑, Ning Qin b a
Science and Technology on Scramjet Laboratory, National University of Defense Technology, Changsha 410073, China b Department of Mechanical Engineering, University of Sheffield, Sheffield S1 3JD, England, UK Available online 27 June 2012
Abstract Combustion characteristics in a supersonic combustor with hydrogen injection upstream of a cavity flameholder were investigated both experimentally and numerically. The combustion was observed to be stabilized in the cavity mode around the shear layer via a dynamic balance and then spread into the main stream in the region around the jet centerplane where the flow was decelerated and turned to the main stream, supplying a favorable condition for the combustion to spread. The combustion spreading from the cavity shear layer to the main stream seemed to be dominated not only by the traditional diffusion process but also by the convection process associated with the extended recirculation flows resulting from the heat release and the interaction between the jet and the cavity shear layer. Therefore, the cavity-stabilized combustion appeared to be a strongly coupled process of flow and heat release around the cavity flameholder. Ó 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Cavity flameholder; Supersonic combustion; Flameholding
1. Introduction Scramjet engine lets the air stream enter into the combustor supersonically and organizes combustion within supersonic flow, where robust flameholding schemes are necessary due to the short combustor residence time. One promising candidate for such a flameholder is the wall cavity which has been shown to be effective in stabilizing the flame without excessively decreasing total pressure [1]. When used as an integrated fuel injection/ flameholding approach [2], cavity flameholders
⇑ Corresponding authors.
E-mail address: [email protected] (Z. Wang).
have become even more attractive in supersonic combustors and received more and more attention. Ben-Yakar et al. [3] used high-speed framing schlieren and OH-Planar Laser-Induced Fluorescence (PLIF) to investigate hydrogen transverse jet injected upstream of a cavity in air cross-flow simulating flight Mach 10 conditions, where auto-ignition was achieved and OH fluorescence appeared first in the recirculation upstream of the jet and extended along outer edge of the jet plume. O’Byrne et al [4] used OH-PLIF to investigate the supersonic combustion with horizontal injection from the aft wall of a cavity and found combustion occurred in the shear layer above the cavity rather than in the recirculating cavity flow.
1540-7489/$ - see front matter Ó 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.proci.2012.06.049
2074
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
However, Jeong et al [5] found the cavity acted as a flameholder for the case of upstream injection. Micka et al [6–9] investigated the combustion characteristics of a dual-mode scramjet combustor with cavity flameholder, and found that the combustion was anchored at the cavity leading edge at low stagnation temperature and stabilized a short distance downstream of the fuel injection jet in the jet-wake at high stagnation temperature. Gruber et al [10] studied the mixing and combustion in a supersonic flow using cavity flameholders. It was found that an imposed shock train had a significant impact on the mixing and chemical reaction processes that occurred in the cavity flameholder. This feature caused the cavity shear layer to separate, which effectively increased the volume of the cavity and the air entrainment. Rasmussen et al [11,12] investigated stability limits and flameholding mechanism of cavity-stabilized flames with direct cavity injection. When injected from the aft wall, the fuel came into immediate contact with hot combustion products from the reaction zone under the shear layer. Primary combustion occurred under the shear layer and in the aft region of the cavity volume. In contrast, when fuel was injected from the floor, a jetdriven recirculation zone of hot products near the upstream wall of the cavity served as a flameholder. The reaction then occurred on the underside of the shear layer. Sun et al [13,14] studied the combustion in a supersonic combustor with hydrogen injection upstream of cavity flameholders using OH-PLIF and hybrid ReynoldsAveraged Navier–Stokes (RANS)/Large Eddy Simulation (LES). It was shown that an approximately steady flame existed in the cavity shear layer and hot combustion products were transported into the injection jet by the vortex interaction of the jet-with-cavity shear layer, where the counter-rotating vortex induced by the jet and the cavity shear layer played an important role. Although several important phenomena and characteristics of the cavity-organized supersonic combustion are realized, there are still many open questions regarding the physical mechanisms of ignition and flameholding under various injection conditions. The present work focuses on flameholding characteristics in a supersonic combustor with hydrogen injection upstream of a cavity flameholder.
alcohol combustion to simulate flight Mach six conditions, resulting in a Mach 2.52 and mass flow 1 kg/s stream in the combustor entrance. The run time is 5–6 s for the air heater and 0.8–1 s for the fuel injection, and the non-cooling supersonic combustor walls are directly exposed to the atmosphere. Detailed flow conditions at the nozzle exit and fuel jet exit are listed in Table 1, where the fuel is 99.5–99.8% pure H2. Schematic of the test section is shown in Fig. 1. The combustor has a constant width of 50 mm and height of 40 mm. A cavity with depth D ¼ 8 mm, length-to-depth ratio L=D ¼ 7 and aft angle of 45° is mounted on the bottom side, and an injector with orifice diameter 2 mm is fixed 10 mm upstream of the cavity leading edge. Quartz glass windows can be mounted on the top and side walls to allow optical access. OH-PLIF, flame luminosity and schlieren are introduced to characterize flow and combustion. The exposure time of PLIF is 50 ns. The exposure time and frame rate of flame luminosity and schlieren are 0.25 ms and 4000 frame/s, respectively. The PLIF system includes an Nd:YAG laser, a dye laser, a frequency doubler and an intensified charged-coupled device (ICCD) system. The output beam from Nd:YAG laser is at 532 nm. It pumps the dye laser to generate the laser beam at 567.106 nm. Then the output from the dye laser is transferred into the doubler, which transforms it into an Ultraviolet (UV) laser beam at 283.553 nm and 18 mJ/pulse. The UV beam is adjusted to a laser sheet by a group of lenses. The thickness of the plane beam is about 0.2 mm and the width is about 50 mm. The PLIF image is photographed by an ICCD camera. PLIF measurements on y–z planes are calibrated from the original images where the camera is arranged angled to the plane beam. 2.2. Numerical treatment A hybrid RANS/LES method [15] blending the Spalart–Allmaras RANS model [16] (used for near-wall regions) and Yoshizawa sub-grid scale (SGS) model [17] (used for regions away from the wall) is used as turbulence model. An assumed sub-grid PDF (Probability Density Function) Table 1 Experimental conditions.
2. Experimental and numerical descriptions 2.1. Experimental setup The experiments are carried out at National University of Defense Technology in a recently developed direct-connected rig. The model scramjet combustor is installed behind the nozzle of the air heater, which heats the air by means of air/O2/
Parameter
Air
Jet
T 0, K P 0 , MPa Ma Y O2 , % Y H2 O , % Y CO2 , % Y N2 , %
1486 1.6 2.52 23.38 6.22 10.16 60.24
300 0.6/1.2/1.8 1.0 0.0 0.0 0.0 0.0
Y H2 , %
0.0
100.0
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2075
Fig. 1. Schematic of test section.
closure model [18] is used for turbulence-chemistry interaction. The 9 species 19 step chemistry mechanism [19] for H2–air combustion is adopted. More details can be found in Ref. [18]. The width of the computational domain is reduced to 30 mm and periodic boundary condition is adopted in the spanwise direction, which saves much computing cost but does not significantly change the combustion characteristics since the fuel jet is basically confined in the central volume under the present conditions. Two meshes containing grid points of about 2.5 million and 4 million are tested for Pjet = 0.6 MPa. Since only minor differences between the mean and fluctuating results on the two meshes, all the calculations presented are carried out on the smaller grid, consisting of a 326 121 51 grid above the cavity and 151 61 51 in the cavity. 3. Results and discussion 3.1. Experimental and numerical observations The combustion characteristics under three injection conditions Pjet = 0.6, 1.2 and 1.8 MPa, resulting in equivalence ratios of 0.038, 0.076 and 0.11 based on the experiment conditions, are considered. It is found in the experiment that stable combustion cannot be obtained without a cavity, indicating the cavity acts as a flameholder and the combustion is stabilized in the cavity mode rather than in the jet-wake mode. Figure 2 shows the experimental average flame luminosity images, where a combustion iso-luminosity contour of 10 counts is used to characterize the combustion zone outline. The combustion zone spreads into the main flow at an approximately constant angle from the fixed stabilization location though the stabilization location is pushed upstream with increasing injection pressure. The spreading angles are 17°, 27° and 30° for Pjet = 0.6, 1.2 and 1.8 MPa, respectively. Not surprisingly, the jet is ignited
earlier for higher injection pressure due to an earlier combustion spreading location and a larger spreading angle. The flamebase (defined as the axial position of the combustion zone leading edge calculated for y/D = 0–0.1 based on the experimental flame luminosity imaging at 4000 Hz) oscillates considerably and may move up beyond the cavity leading edge intermittently for Pjet = 1.8 MPa, as shown in Fig. 3, which may suggest a heavy separation of the upstream boundary layer and an enlarged cavity recirculation. The experimental schlieren images are shown in Fig. 4. For Pjet = 0.6 MPa, the reflection wave of the bow shock is weak and interacts with the hydrogen jet in the very downstream region of the cavity, thus the interaction between the shock waves and cavity-stabilized combustion is weak. For Pjet = 1.8 MPa, the bow shock becomes stronger and adverse pressure gradients are larger due to the intense heat release around the cavity. As a result, the upper wall boundary layer is massively separated, and the separation shock wave is very strong and interacts with the hydrogen jet above the cavity, inducing a strong coupling with the cavity-stabilized combustion. For Pjet = 1.2 MPa, the separation shock wave interacts with the hydrogen jet around the cavity aft wall. Figures 5 and 6 show the instantaneous contours of temperature and density gradient dq=dy together with sonic line in the centerplane from the calculations. The average position of bow shock incidence on the upper wall boundary layer is shown in Fig. 7. The calculated combustion and wave structures agree well with the experimental observations, lending to confidence in the calculations. In general, the combustion is mainly confined in the subsonic regions around the cavity or the lower wall. The freestream keeps supersonic all the while, indicating the combustor operates at scramjet mode. For Pjet = 0.6 MPa, the hydrogen jet is not completely ignited until it is convected to the
2076
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
Fig. 2. Experimental average flame luminosity images. Dashed line is iso-luminosity contour and dashdotted line defines combustion spreading angle.
Fig. 3. Histories of flamebase oscillations from experimental flame luminosity images.
downstream region of the cavity, but large-scale structures formed around the jet edge may be ignited intermittently around the cavity aft wall. The heat release around the cavity is weak and no obvious separation is detected in the upper wall boundary layer. For Pjet = 1.2 MPa, the hydrogen jet is ignited around the cavity aft wall and an obvious separation is detected in the upper wall boundary layer. For Pjet = 1.8 MPa, the
intense heat release around the cavity shear layer lifts the jet greatly and pushes the sonic line deeply into the main stream. Furthermore, the hot products around the cavity leading edge are not confined below the corner but may intrude deeply into the main stream, which is consistent with the experimental observations, indicating a considerable extension of the recirculation around the jet centerplane.
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2077
Fig. 4. Experimental schlieren images.
Fig. 5. Calculated instantaneous contours of temperature together with sonic line in centerplane.
OH-PLIF images displaying the instantaneous combustion structures are taken for Pjet = 1.2 and 1.8 MPa, as shown in Figs. 8 and 9. The OH-PLIF results also demonstrate a large combustion spreading angle in the centerplane and OH radical
may spread to the upstream region beyond the cavity leading edge especially for Pjet = 1.8 MPa. From the end views, one can clearly see that the OH radical spreads from the cavity shear layer to the main stream and then ignites the whole
2078
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
Fig. 6. Calculated instantaneous contours of density gradientdq=dy together with sonic line in centerplane.
Fig. 7. Position of bow shock incidence on the upper wall boundary layer.
hydrogen jet. It is apparent for Pjet = 1.2 MPa that OH radical above the cavity is clustered around the centerplane at x = 10 mm and spreads to a wider region at x = 40 and 75 mm, suggesting that the combustion spreading from the cavity to the main stream primarily occurs around the centerplane under the jet. This is also a good proof that the interaction between the jet and the cavity shear layer plays an important role in the flameholding. For Pjet = 1.8 MPa, the OH distribution has already been rather wide in the spanwise direction at x = 10 mm, which suggests a wider range of combustion spreading around the centerplane, resulting from a stronger interaction between the jet and the cavity shear layer and a larger extension of the cavity recirculation, as can be seen from the numerical results below. Figure 10 shows the instantaneous and timeaveraged contours of OH distribution together with sonic line and axial velocity stagnation line from the calculations. With increasing injection pressure, the spanwise dimension of the subsonic region around the hydrogen jet increases and intense combustion occurs earlier. Also, the recirculation region becomes larger and extends
deeper into the main stream around the centerplane with increasing injection pressure, as a result of which the hot products can spread to the main stream more upstream and ignite the jet more readily. Moreover, the extension of the recirculation definitely increases the residence time of the reactants around the cavity, which is beneficial to reduce the distance for combustion to be completed. Here, the interaction between the jet and the cavity shear layer can tear a gap intermittently on the shear layer and enhance the mass and momentum exchange between the fluid in and out of the cavity. In addition, the heat release around the cavity and the counter-rotating vortex induced by the jet promote the extension of the cavity recirculation even more. This intermittently enlarged recirculation may explain the observations of experiments and calculations that hot products including OH around the cavity leading edge may intrude deeply into the freestream or even intermittently appear upstream of the cavity. The large combustion spreading angle may also be related to this recirculation extension, as will be analyzed below.
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2079
Fig. 8. OH-PLIF images of side view in centerplane and end views in three axial locations: Pjet = 1.2 MPa.
Fig. 9. OH-PLIF images of side view in centerplane and end views in three axial locations: Pjet = 1.8 MPa.
3.2. Flameholding mechanism Figure 11 shows the schematic image showing the possible flame holding and spreading. Basically, the combustion is expected to be stabilized
around the cavity shear layer via a dynamic process. First, the horseshoe vortex formed around the jet exit and the vortex interaction of the jet-with-cavity shear layer tend to transport some fuel into the shear layer, making the fluid there
2080
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
Fig. 10. Calculated oblique views of axial slices at x = 12, 7.6, 27.1, 46.7, 66.2, 85.7, 105.3, 124.8, 144.4 and 164 mm; instantaneous (left) and time-averaged (right) contours of OH distribution together with sonic line (purple) and axial velocity stagnation line (black). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 11. Schematic of combustion spreading around the cavity.
partially premixed and combustible. Second, the hot products within the cavity recirculation and the intense turbulence in the shear layer are beneficial to enhance ignition and combustion around the shear layer. Third, the shear layer supplies a favorable fluid dynamics condition for the possible flamebase to be stabilized. Due to the velocity distribution in the shear layer, there exists a transverse location where the flame speed equals the convection speed and the flamebase can be stabilized. Once a disturbance destroys this balance, the flamebase can adjust its position to achieve a new balance as long as the disturbance is a sustainable one; otherwise the flame is blown off.
Notably, the flamebase may appear upstream of the cavity leading edge intermittently for high injection pressure due to heavy separation of the upstream boundary layer and the enlarged cavity recirculation. However, the combustion is stabilized in the cavity mode at most time and cannot be stabilized in the jet wake without the cavity flameholder. That is, the intermittent jet-wake combustion is unstable and must rely on the stable cavity-stabilized combustion. Once the combustion is stabilized around the cavity shear layer, it needs to spread to the main flow and ignite the jet so that the flameholding process can be accomplished. Flameholding in
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2081
Fig. 12. Evaluated flow conditions in the spanwise direction.
the traditional sense (diffusive preheating and seeding reactants with combustion products) appears impossible due to low flame speeds (10 m/s) in comparison to combustor velocity (1000 m/s) under scramjet conditions [20]. The large combustion zone spreading angle observed in the present study also indicates the spreading mechanism for the cavity-stabilized supersonic combustion may be different. Take Pjet = 1.2 MPa for example. The combustion spreading angle from the cavity shear layer to the main stream is around 27° and the freestream velocity is 1340 m/s. If one supposes that a flame spreads to the main stream under the flow condition like the freestream, the corresponding flame speed is up to 600 m/s, which seems impossible. So it is believed that there must be quit different flow conditions where the combustion spreads from the cavity shear layer to the main stream. Still take Pjet = 1.2 MPa for example. In order to obtain the flow conditions in the spanwise direction, one can integrate velocities in the spanwise planes just above the cavity (x/D = 0–7 and y/D = 0– 1 are used here) based on the calculation. If one presumes the stabilization is achieved by a flame spreading, according to the observed spreading angle a ¼ 27 , the integrated velocity magnitude U , the velocity angle to the axial direction b, and the schematic in Fig. 11, the required flame speed S f at each spanwise position can be evaluated. The results are shown in Fig. 12, where z0 denotes the spanwise distance away from the centerplane. It can be seen that the flow is well decelerated (450 m/s) and turns to the main stream (b ¼ 24) around the centerplane, where the flame speed (21 m/s) required to achieve the observed spreading angle is the least and reasonable. This may be attributed to the enlarged recirculation resulting from the heat release and the interaction between the jet and the cavity shear layer. Though the effects of auto-ignition behavior on the combustion cannot be clarified, these analyses indicate the possibility of the existence of a flame. Away from the centerplane, the velocity of the main stream becomes higher and the required flame speed becomes larger and unpractical. Therefore, no matter the combustion
is dominated by auto-ignition or flame propagation, it is believed that the combustion spreading from the cavity shear layer to the main stream mostly occurs around the jet centerplane where the flowfield supplies a favorable flow condition for the combustion to spread. Meanwhile, the combustion spreading is achieved not only by the diffusion process but also by the convection process around the centerplane, which is very different from that in the traditional sense where it is mainly controlled by the mass and heat diffusion. Therefore, the flameholding is a strongly coupled process of flow and heat release around the cavity flameholder, where the extension of the cavity recirculation and the interaction between the jet and the cavity shear layer play an important role. 4. Conclusions Combustion characteristics in a supersonic combustor with hydrogen injection upstream of a cavity flameholder were investigated both experimentally and numerically. The combustion and shock wave structures were clarified based on the experimental observations using OH-PLIF, flame luminosity and schlieren. Large eddy simulation was also carried out to provide insight into the flameholding mechanism. Basically, the combustion was stabilized in the cavity mode around the shear layer via a dynamic balance controlled by the local flow conditions. For high injection pressure, however, the flamebase or combustion zone might be pushed upstream of the cavity intermittently due to the heavy separation of the upstream boundary layer and enlarged cavity recirculation, resulting from the intense heat release around the cavity and the interaction between the jet and the cavity shear layer. The combustion then spread into the main stream in the region around the jet centerplane, where the flow was decelerated and turned to the main stream, supplying a favorable condition for the combustion spreading. The combustion spreading from the cavity shear layer to the main stream seemed to be dominated not only by the traditional diffusion process but also by the
2082
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
convection process associated with the extended recirculation flows resulting from the heat release and the interaction between the jet and the cavity shear layer. Therefore, the cavity-stabilized combustion appeared to be a strongly coupled process of flow and heat release around the cavity flameholder.
Acknowledgements This work is supported by the National Natural Science Foundation of China under Grant Nos. 50906098 and 91016028.
References [1] T. Mathur, M. Gruber, K. Jackson, et al., J. Propul. Power 17 (6) (2001) 1305–1312. [2] A. Ben-Yakar, R.K. Hanson, J. Propul. Power 17 (4) (2001) 869–877. [3] A. Ben-Yakar, R.K. Hanson, Aerospace Sciences Meeting and Exhibit, AIAA 99-0484, 1999. [4] S. O’Byrne, I. Stotz, A.J. Neely, et al., International Space Planes and Hypersonics Systems and Technologies. AIAA 2005-3357, 2005. [5] E. Jeong, S. O’Byrne, I.-S. Jeung, et al., J. Propul. Power 24 (6) (2008) 1258–1268. [6] D.J. Micka, J.F. Driscoll, Joint Propulsion Conference and Exhibit, AIAA 2008-5071, 2008.
[7] D.J. Micka, J.F. Driscoll, Proc. Combust. Inst. 32 (2009) 2397–2404. [8] D.J. Micka, S.M. Torrez, J.F. Driscoll, International Space Planes and Hypersonics Systems and Technologies, AIAA 2009-7362, 2009. [9] D.J. Micka, Combustion Stabilization, Structure, and Spreading in a Laboratory Dual-Mode Scramjet Combustor, Ph.D. thesis, The University of Michigan, Michigan, USA, 2010. [10] M.R. Gruber, A.S. Nejda, T.H. Chen, et al., J. Propul. Power 11 (2) (1995) 315–323. [11] C.C. Rasmussen, J.F. Driscoll, K.-Y. Hsub, Proc. Combust. Inst. 30 (2005) 2825–2834. [12] C.C. Rasmussen, S.K. Dhanuka, J.F. Driscoll, Proc. Combust. Inst. 31 (2007) 2505–2512. [13] M.B. Sun, Z.G. Wang, J.H. Laing, et al., J. Propul. Power 24 (4) (2008) 688–696. [14] M.B. Sun, H.B. Wang, X.S. Bai, et al., Joint Propulsion Conference and Exhibit, AIAA 20095187, 2009. [15] H.B. Wang, N. Qin, M.B. Sun, et al., Proc. IMechE Part G: J. Aerospace Engineering (2011). [16] P.R. Spalart, S.R. Allmaras, Aerospace Sciences Meeting and Exhibit, AIAA 92-0439, 1992. [17] A. Yoshizawa, K. Horiuti, J. Phys. Soc. Jpn. 54 (8) (1985) 2834–2839. [18] H.B. Wang, N. Qin, M.B. Sun, et al., Sci. China Ser. E: Tech. 54 (10) (2011) 2694–2707. [19] C.J. Jachimowski, An Analytical Study of the Hydrogen–Air Reaction Mechanism With Application to Scramjet Combustion, Report No. TP-2791, NASA, 1988. [20] T. Cain, C. Walton, Joint Propulsion Conference and Exhibit, AIAA 2002-3877, 2002.
Proceedings of the
Proceedings of the Combustion Institute 34 (2013) 2073–2082
Combustion Institute www.elsevier.com/locate/proci
Combustion characteristics in a supersonic combustor with hydrogen injection upstream of cavity flameholder Hongbo Wang a, Zhenguo Wang a,⇑, Mingbo Sun a,⇑, Ning Qin b a
Science and Technology on Scramjet Laboratory, National University of Defense Technology, Changsha 410073, China b Department of Mechanical Engineering, University of Sheffield, Sheffield S1 3JD, England, UK Available online 27 June 2012
Abstract Combustion characteristics in a supersonic combustor with hydrogen injection upstream of a cavity flameholder were investigated both experimentally and numerically. The combustion was observed to be stabilized in the cavity mode around the shear layer via a dynamic balance and then spread into the main stream in the region around the jet centerplane where the flow was decelerated and turned to the main stream, supplying a favorable condition for the combustion to spread. The combustion spreading from the cavity shear layer to the main stream seemed to be dominated not only by the traditional diffusion process but also by the convection process associated with the extended recirculation flows resulting from the heat release and the interaction between the jet and the cavity shear layer. Therefore, the cavity-stabilized combustion appeared to be a strongly coupled process of flow and heat release around the cavity flameholder. Ó 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Cavity flameholder; Supersonic combustion; Flameholding
1. Introduction Scramjet engine lets the air stream enter into the combustor supersonically and organizes combustion within supersonic flow, where robust flameholding schemes are necessary due to the short combustor residence time. One promising candidate for such a flameholder is the wall cavity which has been shown to be effective in stabilizing the flame without excessively decreasing total pressure [1]. When used as an integrated fuel injection/ flameholding approach [2], cavity flameholders
⇑ Corresponding authors.
E-mail address: [email protected] (Z. Wang).
have become even more attractive in supersonic combustors and received more and more attention. Ben-Yakar et al. [3] used high-speed framing schlieren and OH-Planar Laser-Induced Fluorescence (PLIF) to investigate hydrogen transverse jet injected upstream of a cavity in air cross-flow simulating flight Mach 10 conditions, where auto-ignition was achieved and OH fluorescence appeared first in the recirculation upstream of the jet and extended along outer edge of the jet plume. O’Byrne et al [4] used OH-PLIF to investigate the supersonic combustion with horizontal injection from the aft wall of a cavity and found combustion occurred in the shear layer above the cavity rather than in the recirculating cavity flow.
1540-7489/$ - see front matter Ó 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.proci.2012.06.049
2074
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
However, Jeong et al [5] found the cavity acted as a flameholder for the case of upstream injection. Micka et al [6–9] investigated the combustion characteristics of a dual-mode scramjet combustor with cavity flameholder, and found that the combustion was anchored at the cavity leading edge at low stagnation temperature and stabilized a short distance downstream of the fuel injection jet in the jet-wake at high stagnation temperature. Gruber et al [10] studied the mixing and combustion in a supersonic flow using cavity flameholders. It was found that an imposed shock train had a significant impact on the mixing and chemical reaction processes that occurred in the cavity flameholder. This feature caused the cavity shear layer to separate, which effectively increased the volume of the cavity and the air entrainment. Rasmussen et al [11,12] investigated stability limits and flameholding mechanism of cavity-stabilized flames with direct cavity injection. When injected from the aft wall, the fuel came into immediate contact with hot combustion products from the reaction zone under the shear layer. Primary combustion occurred under the shear layer and in the aft region of the cavity volume. In contrast, when fuel was injected from the floor, a jetdriven recirculation zone of hot products near the upstream wall of the cavity served as a flameholder. The reaction then occurred on the underside of the shear layer. Sun et al [13,14] studied the combustion in a supersonic combustor with hydrogen injection upstream of cavity flameholders using OH-PLIF and hybrid ReynoldsAveraged Navier–Stokes (RANS)/Large Eddy Simulation (LES). It was shown that an approximately steady flame existed in the cavity shear layer and hot combustion products were transported into the injection jet by the vortex interaction of the jet-with-cavity shear layer, where the counter-rotating vortex induced by the jet and the cavity shear layer played an important role. Although several important phenomena and characteristics of the cavity-organized supersonic combustion are realized, there are still many open questions regarding the physical mechanisms of ignition and flameholding under various injection conditions. The present work focuses on flameholding characteristics in a supersonic combustor with hydrogen injection upstream of a cavity flameholder.
alcohol combustion to simulate flight Mach six conditions, resulting in a Mach 2.52 and mass flow 1 kg/s stream in the combustor entrance. The run time is 5–6 s for the air heater and 0.8–1 s for the fuel injection, and the non-cooling supersonic combustor walls are directly exposed to the atmosphere. Detailed flow conditions at the nozzle exit and fuel jet exit are listed in Table 1, where the fuel is 99.5–99.8% pure H2. Schematic of the test section is shown in Fig. 1. The combustor has a constant width of 50 mm and height of 40 mm. A cavity with depth D ¼ 8 mm, length-to-depth ratio L=D ¼ 7 and aft angle of 45° is mounted on the bottom side, and an injector with orifice diameter 2 mm is fixed 10 mm upstream of the cavity leading edge. Quartz glass windows can be mounted on the top and side walls to allow optical access. OH-PLIF, flame luminosity and schlieren are introduced to characterize flow and combustion. The exposure time of PLIF is 50 ns. The exposure time and frame rate of flame luminosity and schlieren are 0.25 ms and 4000 frame/s, respectively. The PLIF system includes an Nd:YAG laser, a dye laser, a frequency doubler and an intensified charged-coupled device (ICCD) system. The output beam from Nd:YAG laser is at 532 nm. It pumps the dye laser to generate the laser beam at 567.106 nm. Then the output from the dye laser is transferred into the doubler, which transforms it into an Ultraviolet (UV) laser beam at 283.553 nm and 18 mJ/pulse. The UV beam is adjusted to a laser sheet by a group of lenses. The thickness of the plane beam is about 0.2 mm and the width is about 50 mm. The PLIF image is photographed by an ICCD camera. PLIF measurements on y–z planes are calibrated from the original images where the camera is arranged angled to the plane beam. 2.2. Numerical treatment A hybrid RANS/LES method [15] blending the Spalart–Allmaras RANS model [16] (used for near-wall regions) and Yoshizawa sub-grid scale (SGS) model [17] (used for regions away from the wall) is used as turbulence model. An assumed sub-grid PDF (Probability Density Function) Table 1 Experimental conditions.
2. Experimental and numerical descriptions 2.1. Experimental setup The experiments are carried out at National University of Defense Technology in a recently developed direct-connected rig. The model scramjet combustor is installed behind the nozzle of the air heater, which heats the air by means of air/O2/
Parameter
Air
Jet
T 0, K P 0 , MPa Ma Y O2 , % Y H2 O , % Y CO2 , % Y N2 , %
1486 1.6 2.52 23.38 6.22 10.16 60.24
300 0.6/1.2/1.8 1.0 0.0 0.0 0.0 0.0
Y H2 , %
0.0
100.0
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2075
Fig. 1. Schematic of test section.
closure model [18] is used for turbulence-chemistry interaction. The 9 species 19 step chemistry mechanism [19] for H2–air combustion is adopted. More details can be found in Ref. [18]. The width of the computational domain is reduced to 30 mm and periodic boundary condition is adopted in the spanwise direction, which saves much computing cost but does not significantly change the combustion characteristics since the fuel jet is basically confined in the central volume under the present conditions. Two meshes containing grid points of about 2.5 million and 4 million are tested for Pjet = 0.6 MPa. Since only minor differences between the mean and fluctuating results on the two meshes, all the calculations presented are carried out on the smaller grid, consisting of a 326 121 51 grid above the cavity and 151 61 51 in the cavity. 3. Results and discussion 3.1. Experimental and numerical observations The combustion characteristics under three injection conditions Pjet = 0.6, 1.2 and 1.8 MPa, resulting in equivalence ratios of 0.038, 0.076 and 0.11 based on the experiment conditions, are considered. It is found in the experiment that stable combustion cannot be obtained without a cavity, indicating the cavity acts as a flameholder and the combustion is stabilized in the cavity mode rather than in the jet-wake mode. Figure 2 shows the experimental average flame luminosity images, where a combustion iso-luminosity contour of 10 counts is used to characterize the combustion zone outline. The combustion zone spreads into the main flow at an approximately constant angle from the fixed stabilization location though the stabilization location is pushed upstream with increasing injection pressure. The spreading angles are 17°, 27° and 30° for Pjet = 0.6, 1.2 and 1.8 MPa, respectively. Not surprisingly, the jet is ignited
earlier for higher injection pressure due to an earlier combustion spreading location and a larger spreading angle. The flamebase (defined as the axial position of the combustion zone leading edge calculated for y/D = 0–0.1 based on the experimental flame luminosity imaging at 4000 Hz) oscillates considerably and may move up beyond the cavity leading edge intermittently for Pjet = 1.8 MPa, as shown in Fig. 3, which may suggest a heavy separation of the upstream boundary layer and an enlarged cavity recirculation. The experimental schlieren images are shown in Fig. 4. For Pjet = 0.6 MPa, the reflection wave of the bow shock is weak and interacts with the hydrogen jet in the very downstream region of the cavity, thus the interaction between the shock waves and cavity-stabilized combustion is weak. For Pjet = 1.8 MPa, the bow shock becomes stronger and adverse pressure gradients are larger due to the intense heat release around the cavity. As a result, the upper wall boundary layer is massively separated, and the separation shock wave is very strong and interacts with the hydrogen jet above the cavity, inducing a strong coupling with the cavity-stabilized combustion. For Pjet = 1.2 MPa, the separation shock wave interacts with the hydrogen jet around the cavity aft wall. Figures 5 and 6 show the instantaneous contours of temperature and density gradient dq=dy together with sonic line in the centerplane from the calculations. The average position of bow shock incidence on the upper wall boundary layer is shown in Fig. 7. The calculated combustion and wave structures agree well with the experimental observations, lending to confidence in the calculations. In general, the combustion is mainly confined in the subsonic regions around the cavity or the lower wall. The freestream keeps supersonic all the while, indicating the combustor operates at scramjet mode. For Pjet = 0.6 MPa, the hydrogen jet is not completely ignited until it is convected to the
2076
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
Fig. 2. Experimental average flame luminosity images. Dashed line is iso-luminosity contour and dashdotted line defines combustion spreading angle.
Fig. 3. Histories of flamebase oscillations from experimental flame luminosity images.
downstream region of the cavity, but large-scale structures formed around the jet edge may be ignited intermittently around the cavity aft wall. The heat release around the cavity is weak and no obvious separation is detected in the upper wall boundary layer. For Pjet = 1.2 MPa, the hydrogen jet is ignited around the cavity aft wall and an obvious separation is detected in the upper wall boundary layer. For Pjet = 1.8 MPa, the
intense heat release around the cavity shear layer lifts the jet greatly and pushes the sonic line deeply into the main stream. Furthermore, the hot products around the cavity leading edge are not confined below the corner but may intrude deeply into the main stream, which is consistent with the experimental observations, indicating a considerable extension of the recirculation around the jet centerplane.
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2077
Fig. 4. Experimental schlieren images.
Fig. 5. Calculated instantaneous contours of temperature together with sonic line in centerplane.
OH-PLIF images displaying the instantaneous combustion structures are taken for Pjet = 1.2 and 1.8 MPa, as shown in Figs. 8 and 9. The OH-PLIF results also demonstrate a large combustion spreading angle in the centerplane and OH radical
may spread to the upstream region beyond the cavity leading edge especially for Pjet = 1.8 MPa. From the end views, one can clearly see that the OH radical spreads from the cavity shear layer to the main stream and then ignites the whole
2078
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
Fig. 6. Calculated instantaneous contours of density gradientdq=dy together with sonic line in centerplane.
Fig. 7. Position of bow shock incidence on the upper wall boundary layer.
hydrogen jet. It is apparent for Pjet = 1.2 MPa that OH radical above the cavity is clustered around the centerplane at x = 10 mm and spreads to a wider region at x = 40 and 75 mm, suggesting that the combustion spreading from the cavity to the main stream primarily occurs around the centerplane under the jet. This is also a good proof that the interaction between the jet and the cavity shear layer plays an important role in the flameholding. For Pjet = 1.8 MPa, the OH distribution has already been rather wide in the spanwise direction at x = 10 mm, which suggests a wider range of combustion spreading around the centerplane, resulting from a stronger interaction between the jet and the cavity shear layer and a larger extension of the cavity recirculation, as can be seen from the numerical results below. Figure 10 shows the instantaneous and timeaveraged contours of OH distribution together with sonic line and axial velocity stagnation line from the calculations. With increasing injection pressure, the spanwise dimension of the subsonic region around the hydrogen jet increases and intense combustion occurs earlier. Also, the recirculation region becomes larger and extends
deeper into the main stream around the centerplane with increasing injection pressure, as a result of which the hot products can spread to the main stream more upstream and ignite the jet more readily. Moreover, the extension of the recirculation definitely increases the residence time of the reactants around the cavity, which is beneficial to reduce the distance for combustion to be completed. Here, the interaction between the jet and the cavity shear layer can tear a gap intermittently on the shear layer and enhance the mass and momentum exchange between the fluid in and out of the cavity. In addition, the heat release around the cavity and the counter-rotating vortex induced by the jet promote the extension of the cavity recirculation even more. This intermittently enlarged recirculation may explain the observations of experiments and calculations that hot products including OH around the cavity leading edge may intrude deeply into the freestream or even intermittently appear upstream of the cavity. The large combustion spreading angle may also be related to this recirculation extension, as will be analyzed below.
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2079
Fig. 8. OH-PLIF images of side view in centerplane and end views in three axial locations: Pjet = 1.2 MPa.
Fig. 9. OH-PLIF images of side view in centerplane and end views in three axial locations: Pjet = 1.8 MPa.
3.2. Flameholding mechanism Figure 11 shows the schematic image showing the possible flame holding and spreading. Basically, the combustion is expected to be stabilized
around the cavity shear layer via a dynamic process. First, the horseshoe vortex formed around the jet exit and the vortex interaction of the jet-with-cavity shear layer tend to transport some fuel into the shear layer, making the fluid there
2080
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
Fig. 10. Calculated oblique views of axial slices at x = 12, 7.6, 27.1, 46.7, 66.2, 85.7, 105.3, 124.8, 144.4 and 164 mm; instantaneous (left) and time-averaged (right) contours of OH distribution together with sonic line (purple) and axial velocity stagnation line (black). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Fig. 11. Schematic of combustion spreading around the cavity.
partially premixed and combustible. Second, the hot products within the cavity recirculation and the intense turbulence in the shear layer are beneficial to enhance ignition and combustion around the shear layer. Third, the shear layer supplies a favorable fluid dynamics condition for the possible flamebase to be stabilized. Due to the velocity distribution in the shear layer, there exists a transverse location where the flame speed equals the convection speed and the flamebase can be stabilized. Once a disturbance destroys this balance, the flamebase can adjust its position to achieve a new balance as long as the disturbance is a sustainable one; otherwise the flame is blown off.
Notably, the flamebase may appear upstream of the cavity leading edge intermittently for high injection pressure due to heavy separation of the upstream boundary layer and the enlarged cavity recirculation. However, the combustion is stabilized in the cavity mode at most time and cannot be stabilized in the jet wake without the cavity flameholder. That is, the intermittent jet-wake combustion is unstable and must rely on the stable cavity-stabilized combustion. Once the combustion is stabilized around the cavity shear layer, it needs to spread to the main flow and ignite the jet so that the flameholding process can be accomplished. Flameholding in
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
2081
Fig. 12. Evaluated flow conditions in the spanwise direction.
the traditional sense (diffusive preheating and seeding reactants with combustion products) appears impossible due to low flame speeds (10 m/s) in comparison to combustor velocity (1000 m/s) under scramjet conditions [20]. The large combustion zone spreading angle observed in the present study also indicates the spreading mechanism for the cavity-stabilized supersonic combustion may be different. Take Pjet = 1.2 MPa for example. The combustion spreading angle from the cavity shear layer to the main stream is around 27° and the freestream velocity is 1340 m/s. If one supposes that a flame spreads to the main stream under the flow condition like the freestream, the corresponding flame speed is up to 600 m/s, which seems impossible. So it is believed that there must be quit different flow conditions where the combustion spreads from the cavity shear layer to the main stream. Still take Pjet = 1.2 MPa for example. In order to obtain the flow conditions in the spanwise direction, one can integrate velocities in the spanwise planes just above the cavity (x/D = 0–7 and y/D = 0– 1 are used here) based on the calculation. If one presumes the stabilization is achieved by a flame spreading, according to the observed spreading angle a ¼ 27 , the integrated velocity magnitude U , the velocity angle to the axial direction b, and the schematic in Fig. 11, the required flame speed S f at each spanwise position can be evaluated. The results are shown in Fig. 12, where z0 denotes the spanwise distance away from the centerplane. It can be seen that the flow is well decelerated (450 m/s) and turns to the main stream (b ¼ 24) around the centerplane, where the flame speed (21 m/s) required to achieve the observed spreading angle is the least and reasonable. This may be attributed to the enlarged recirculation resulting from the heat release and the interaction between the jet and the cavity shear layer. Though the effects of auto-ignition behavior on the combustion cannot be clarified, these analyses indicate the possibility of the existence of a flame. Away from the centerplane, the velocity of the main stream becomes higher and the required flame speed becomes larger and unpractical. Therefore, no matter the combustion
is dominated by auto-ignition or flame propagation, it is believed that the combustion spreading from the cavity shear layer to the main stream mostly occurs around the jet centerplane where the flowfield supplies a favorable flow condition for the combustion to spread. Meanwhile, the combustion spreading is achieved not only by the diffusion process but also by the convection process around the centerplane, which is very different from that in the traditional sense where it is mainly controlled by the mass and heat diffusion. Therefore, the flameholding is a strongly coupled process of flow and heat release around the cavity flameholder, where the extension of the cavity recirculation and the interaction between the jet and the cavity shear layer play an important role. 4. Conclusions Combustion characteristics in a supersonic combustor with hydrogen injection upstream of a cavity flameholder were investigated both experimentally and numerically. The combustion and shock wave structures were clarified based on the experimental observations using OH-PLIF, flame luminosity and schlieren. Large eddy simulation was also carried out to provide insight into the flameholding mechanism. Basically, the combustion was stabilized in the cavity mode around the shear layer via a dynamic balance controlled by the local flow conditions. For high injection pressure, however, the flamebase or combustion zone might be pushed upstream of the cavity intermittently due to the heavy separation of the upstream boundary layer and enlarged cavity recirculation, resulting from the intense heat release around the cavity and the interaction between the jet and the cavity shear layer. The combustion then spread into the main stream in the region around the jet centerplane, where the flow was decelerated and turned to the main stream, supplying a favorable condition for the combustion spreading. The combustion spreading from the cavity shear layer to the main stream seemed to be dominated not only by the traditional diffusion process but also by the
2082
H. Wang et al. / Proceedings of the Combustion Institute 34 (2013) 2073–2082
convection process associated with the extended recirculation flows resulting from the heat release and the interaction between the jet and the cavity shear layer. Therefore, the cavity-stabilized combustion appeared to be a strongly coupled process of flow and heat release around the cavity flameholder.
Acknowledgements This work is supported by the National Natural Science Foundation of China under Grant Nos. 50906098 and 91016028.
References [1] T. Mathur, M. Gruber, K. Jackson, et al., J. Propul. Power 17 (6) (2001) 1305–1312. [2] A. Ben-Yakar, R.K. Hanson, J. Propul. Power 17 (4) (2001) 869–877. [3] A. Ben-Yakar, R.K. Hanson, Aerospace Sciences Meeting and Exhibit, AIAA 99-0484, 1999. [4] S. O’Byrne, I. Stotz, A.J. Neely, et al., International Space Planes and Hypersonics Systems and Technologies. AIAA 2005-3357, 2005. [5] E. Jeong, S. O’Byrne, I.-S. Jeung, et al., J. Propul. Power 24 (6) (2008) 1258–1268. [6] D.J. Micka, J.F. Driscoll, Joint Propulsion Conference and Exhibit, AIAA 2008-5071, 2008.
[7] D.J. Micka, J.F. Driscoll, Proc. Combust. Inst. 32 (2009) 2397–2404. [8] D.J. Micka, S.M. Torrez, J.F. Driscoll, International Space Planes and Hypersonics Systems and Technologies, AIAA 2009-7362, 2009. [9] D.J. Micka, Combustion Stabilization, Structure, and Spreading in a Laboratory Dual-Mode Scramjet Combustor, Ph.D. thesis, The University of Michigan, Michigan, USA, 2010. [10] M.R. Gruber, A.S. Nejda, T.H. Chen, et al., J. Propul. Power 11 (2) (1995) 315–323. [11] C.C. Rasmussen, J.F. Driscoll, K.-Y. Hsub, Proc. Combust. Inst. 30 (2005) 2825–2834. [12] C.C. Rasmussen, S.K. Dhanuka, J.F. Driscoll, Proc. Combust. Inst. 31 (2007) 2505–2512. [13] M.B. Sun, Z.G. Wang, J.H. Laing, et al., J. Propul. Power 24 (4) (2008) 688–696. [14] M.B. Sun, H.B. Wang, X.S. Bai, et al., Joint Propulsion Conference and Exhibit, AIAA 20095187, 2009. [15] H.B. Wang, N. Qin, M.B. Sun, et al., Proc. IMechE Part G: J. Aerospace Engineering (2011). [16] P.R. Spalart, S.R. Allmaras, Aerospace Sciences Meeting and Exhibit, AIAA 92-0439, 1992. [17] A. Yoshizawa, K. Horiuti, J. Phys. Soc. Jpn. 54 (8) (1985) 2834–2839. [18] H.B. Wang, N. Qin, M.B. Sun, et al., Sci. China Ser. E: Tech. 54 (10) (2011) 2694–2707. [19] C.J. Jachimowski, An Analytical Study of the Hydrogen–Air Reaction Mechanism With Application to Scramjet Combustion, Report No. TP-2791, NASA, 1988. [20] T. Cain, C. Walton, Joint Propulsion Conference and Exhibit, AIAA 2002-3877, 2002.