Propagation Characteristics of Turbulent Methane-Air Premixed Flames at Elevated Pressure in a Constant Volume Vessel

Transcription

1 Proagation Characteristics of Turbulent Methane-Air Premixed Flames at Elevated Pressure in a Constant Volume Vessel S. Yoshida 1,*, Y. Naka 2, Y. Minamoto 1, M. Shimura 1, M. Tanahashi 1 1: Det. of Mechanical and Aerosace Engineering, Tokyo Institute of Technology, Jaan 2: Det. of Science and Engineering, Meiji University, Jaan * Corresondent author: syoshida@navier.mes.titech.ac.j Keywords: OH PLIF, PIV, Turbulent flame, Methane-air remixed flame, Constant volume vessel ABSTRACT Turbulent flame roagation characteristics of methane-air remixed combustion in a constant volume vessel are investigated using reliminary CH chemiluminescence imaging and high-seed simultaneous OH lanar laser induced fluorescence (PLIF) and article image velocimetry (PIV) measurements with well-controlled initial turbulence, ressure and equivalence ratio environment. The turbulent combustion conditions are arametrized in terms of initial ressure and turbulence level including reresentative laminar cases. Laser systems consist of a Nd:YAG laser and a dye laser for OH PLIF and a Nd:YAG laser for PIV. Laser ulses are emitted at 10 khz with 1 W at nm for OH PLIF and 50 W at 532 nm for PIV. The satial resolution of the measurements are 50.1!m 50.1!m for OH PLIF and 158!m 158!m for PIV. The obtained images show the convoluted flame structures with increased turbulence intensity. The global flame roagation seed, which is obtained from the CH images, increases accordance with turbulence level for the turbulent conditions. However, no significant difference is observed in the global flame seed between the cases with 0.1 MPa and MPa in the turbulent conditions. With the radius of the quasi-sherical flame, the global flame seed for the turbulent cases increases almost linearly due to the increase of flame surface area. The instantaneous results of simultaneous OH PLIF and PIV measurements show temoral evolution of flame shae, and there are abrut changes in flame surface area due to flame flame interactions and reactants/roduct ocket formations at a substantial rate. The PIV and OH images are then used to obtain the 2D local flame dislacement seed from the local fluid velocity and flame-normal vector using a geometrical relation of these fields. The 2D local dislacement seed for the turbulent conditions has relatively large deviation, and the increase of the mean local dislacement seed relative to the laminar case seems to be well scaled with the turbulent velocity fluctuation. 1. Introduction Various combustion devices need to be designed and oerated so as to achieve higher efficiency and lower emissions. Many combustors are oerated under high ressure conditions, and hence it is imortant to understand turbulent combustion mechanism in high ressure condition.

2 Elevated ressure could also influence turbulent flows inside the combustor, on ar with that to the local flame characteristics. The higher the ressure is, the lower the kinematic viscosity is, leading to a higher Reynolds number, and the turbulence flame interaction characteristics which are often reresented by Damköhler and Karlovitz numbers, are likely to be significantly influenced. Characteristics of flame roagation is one of interest due to its relation to modeling of turbulent combustion. Conventionally, line of sight measurement, such as schlieren hotograhy, is used for investigation of flame roagation (Bradley et al. 2003, etc.). In our revious study, CH double-ulse PLIF was established (Tanahashi et al., 2008a) and its combination with OH PLIF and stereoscoic PIV (Tanahashi et al., 2008b) enables us to investigate local turbulent dislacement seed. Recently the measurement of local turbulent dislacement seed has further develoed. Trunk et al. (2013) conducted high reetition rate dual lane OH PLIF and stereoscoic PIV, which is laced at the central osition between the dual lane for OH PLIF, and evaluated three-dimensional flame dislacement seed in an atmosheric freely flame roagating facility. Brian et al. (2015) used the high reetition rate measurement for flame roagation to investigate characteristics of early flame develoment in a sark ignition engine. In addition, PIV-based flame seed measurements have been conducted (Balusamy et al., 2011, Kerl et al., 2013), and numerical simulation have investigated validity of the measurement methods (Groot et al. 2002, Bonhomme et al. 2013). A lot of researches have been conducted for understanding characteristics of local turbulent dislacement seed, however, a consistent result has not been obtained even in atmosheric conditions and obtained results usually have unclear oints such as negative dislacement seed. The objectives of this study is to establish a otically measurable constant volume combustion vessel with turbulent control fans for a wide range of conditions of ressure, equivalence ratio and turbulence, and to clarify characteristics of flame dislacement seed by CH chemiluminescence and high seed simultaneous OH PLIF and PIV measurements. The remainder of the aer is organized as follows. The details of exerimental setu and aaratus are discussed in Secs Section 2.3 describes exerimental conditions of the resent measurements. Section 2.4 discusses the images ost-rocessing. Section 3 discusses the general flame features followed by flame dislacement analysis. Finally, conclusions are resented in Sec Exerimental setu and aaratus 2.1 Constant volume vessel

3 Fig. 1 Exerimental aaratus of constant volume vessel. Figure 1 shows the constant volume vessel develoed in the resent study. The combustor is cylindrical with the inner diameter of 280 mm and the height of 300 mm. The vessel has laser light transmitting and observation windows on each side. Well-controlled turbulence is generated by using two fans located on the to and bottom of the chamber, and the turbulence characteristics are controlled by the rotating seed of the fans. The durable ressure of the vessel is 6.0 MPa. The equivalence ratio of the remixed gas is recisely determined by the artial ressures of fuel and air. They are monitored by the two ressure sensors having different sensitivity (KYOWA ELECTRONIC INSTRUMENTS, PGM-02KGM97, PG-2KU-F). The remixed gas is ignited by electrodes located on the center of the vessel after a desirable flow field is achieved in the chamber. The ressure during combustion is measured by a fast resonse ressure sensor (KISTLER, Kistler 6045A, 20 khz). 2.2 Measurement system CH chemiluminescence is collected by a macro lens (Carl Zeiss, Makro PlanarT*2/100, 100mm/f2.0) with a bandass filter (Semrock, FF01-434/ 17-5-D) and imaged onto a high seed CMOS camera (Photron, SA-X2, ixels). For OH PLIF measurements, the Q 1 (7) transition of the A X Π(1,0) band at nm is 2 excited and fluorescence from the A-X(1,1), (0,0) bands between 306 and 320 nm is detected. The laser system consists of an Nd:YAG laser (Edgewave, IS120-2-L, 532 nm, 5 mj/ulse) and a dye laser (LIOP-TEC, LioStar-HQ, Rhodamine 6G, 0.1 mj/ulse). The fluorescence is collected by micro lens (Nikon, UV-Nikkor, 105 mm/f4.5) with a bandass filter (Semrock, SFF01-320/ D) and imaged onto an image intensifier (Hamamatsu Photonics, C F). The amlified

4 Fig. 2 Schematic diagram of high seed OH PLIF and PIV. image is detected by a high seed CMOS camera (Photron, SA-X2, ixels). The laser beam is shaed into about 200!m thickness vertical sheet with 35 mm height. An Nd:YAG laser (LEE LASER, LDP-100MQG, 532 nm, 5 mj/ ulse) is used for PIV measurements. The scattering light is collected by a macro lens (Nicon, Micro-nikkor, 200mm/f4.0) with a bandass filter (Semrock, FF01-525/15-50-D) and imaged onto a high seed CMOS camera (Photron, SA-Z, ixels). Particles of SiO2 (SUZUKI Oil & Fats, GODD BALL E-2C, !m) are used as tracer for PIV in this study. The laser beam is shaed into about mm thickness vertical sheet with 30 mm height. 2.3 Exerimental conditions Two initial ressures Pini, 0.1 MPa and MPa, and four equivalence ratios φ, 0, 0.90 and 0 for laminar conditions and 0 for turbulent conditions, are considered (total of 8 cases). The fans are in a stationary state in laminar conditions and the rotation seeds are 1000 rm and 3000 rm in the turbulent conditions. Table 1 summarize the turbulence characteristics reliminary measured by PIV in non-reacting field. Here Reλ is the Reynolds number based on the Taylor microscale λ and the root-mean-square of velocity fluctuation u rms, l is the integral length scale and η is the Kolmogorov length. The laminar flame thickness based on kinematic viscosity and laminar flame seed and laminar burning velocity are denoted by δf and SL. These conditions yields relatively large Damköhler numbers ranging from 100 to 200, classified into corrugated flamelets regime in the turbulent combustion diagram (Peters 1999). The measurement region of CH chemiluminescence is taken to be 155 mm 155 mm, and

5 Table 1. Exerimental conditions and turbulence characteristics. r[rm] Pini[MPa] Reλ u rms l[mm] λ[mm] η[mm] l/δf u rms/sl its central location corresonds to the ignition oint, yielding a satial resolution of 151!m/ixel. The CH chemiluminescence measurement was erformed at 1 khz. The measurement region for OH PLIF is set to 52.3 mm 52.3 mm, where its central location is located 30 mm above the ignition oint. The satial resolution is 50.1!m/ixel. The measurement region for PIV is set to 2 mm 2 mm, where the central location is 20 mm above the ignition oint. The interrogation window size is taken as ixels and 50% overlaing is chosen. The satial resolution is 157!m/ixel. OH PLIF and PIV measurements are erformed simultaneously at 10 khz. Figure 2 shows the exerimental setu of the simultaneous OH PLIF and PIV measurements. Note that the simultaneous OH PLIF and PIV measurements are erformed for laminar cases at φ = 0 and turbulence cases, while CH chemiluminescence images are obtained for all the conditions considered. 2.4 Image ost-rocessing for simultaneous OH PLIF and PIV measurements The size and resolution of OH PLIF images is different from PIV images, so obtained images are ost-rocessed. OH PLIF images ( ixels) including PIV areas are icked u from rawimages. The obtained velocity field contains velocity vectors (mesh oints) right after the PIV ostrocessing. These images are aligned with OH PLIF images. Then, 4th order Lagrange interolation is alied to PIV images and the vector oints are increased to so that each vector airs with a ixel of OH PLIF images at a corresonding location. The final measuring area is 18.0 mm 18.0 mm for simultaneous measurements. 3. Proagation characteristics of turbulent methane-air remixed flames at high ressure Figure 3 shows measured CH chemiluminescence images at different times for laminar and turbulent cases at 0.10 MPa and 0 MPa. Here, the time when the remixed gas is ignited is defined as t = 0. The flame roagates sherically in laminar conditions, while corrugated flame surfaces are observed for turbulent cases. The scale of the curve length of flame front is smaller

6 Fig. 3 CH chemiluminescence (φ= 0) for laminar condition at Pini = 0.1 MPa (a) and MPa (b), and for turbulence condition at Pini = 0.1 MPa, 1000 rm (c) and MPa, 1000 rm (d). (a) 1.6 (b) 1.2 P [MPa] P [MPa] φ = 0, laminar φ = 0.90, laminar φ = 0, laminar φ = 0, 1000 rm t [ms] φ = 0, laminar φ = 0.90, laminar φ = 0, laminar φ = 0, 1000 rm t [ms] Fig. 4 Pressure history in the constant volume vessel for Pini = 0.1 MPa (a) and MPa (b). for 0.1 MPa than that for MPa, suggesting the influence of turbulence having smaller turbulent length scales for higher ressure cases. Figure 4 shows the temoral evolution of ressure in the constant volume vessel. When φ increases from 0 to 0, the rate of ressure rise is increased and the higher ressure eak is achieved due to the enhanced laminar burning velocity and higher flame temerature. The rate of ressure increase in turbulent conditions is even higher than the laminar cases, suggesting the enhanced overall flame roagation seed by turbulence. To examine the effect of turbulence on roagation characteristics, a global flame roagation seed is obtained from the measured CH chemiluminescence images. A radius of flame r is defined as an azimuthally-averaged distance from the center of electrodes to the flame front obtained by image binarization, and then the global flame roagation seed s is comuted from the rate of

7 (a) (b) s s φ = 0, laminar 0.5 φ = 0.90, laminar φ = 0, laminar 0 5 φ = 0, 1000rm r [mm] φ = 0, laminar 0.5 φ = 0.90, laminar φ = 0, laminar φ = 0, 1000 rm r [mm] Fig. 5 Global flame roagation seed and radius of flame for Pini = 0.1 MPa (a) and MPa (b). increase radius of flame. Figure 5 shows the global flame roagation seed with the radius of flame. When radius of flame is large, the global flame roagation seed becomes almost constant in the laminar flame, while it increases in the turbulent cases artly due to the increased flame surface area. When the initial ressure is increased by 0.1 MPa, there is no significant difference in the flame roagation seed ossibly since the decrease in the laminar flame seed and the increase in flame roagation seed by the enhanced Reynolds number is balanced. Figure 6 shows tyical OH PLIF images in the turbulent conditions. In Fig. 6(a), a burned region (circle area indicated by A) is surrounded by an unburned region. The burned region (area B) exands and intersects with other burned regions. Finally the unburned region is surrounded by the burned region. In Fig. 6(b) two burned regions (area C) are face to face and roagate keeing the constant distance. In the final frame, the two regions are connected. In Fig. 6(c) an unburned region (area D) surrounded by a burned region becomes small due to the consumtion of the remixed gas. A burned region (area E) is torn and surrounded by an unburned region. Figure 7 shows examles of the instantaneous velocity fields and flame front locations for a laminar and turbulent cases. Here, a local flame front is defined as the local maxima of the OH intensity gradient. The flame front is located at a relatively small fluid seed region in laminar condition, while at a larger fluid seed region in the turbulent condition. This yields that that the absolute dislacement seed in the turbulent condition is larger than in laminar condition. Figure 8 shows the normalized OH PLIF intensity and fluid velocity with the distance from flame front in the laminar condition at P ini = 0.1 MPa, φ =. This result qualitatively agrees with the numerical simulation result of Groot and De Goey (2002), suggesting that the results of the simultaneous measurements are reasonable. To examine more detail into turbulence effects on roagation characteristics, the 2D local flame dislacement velocity ud is estimated from OH PLIF and PIV images. The local flame dislacement velocity is defined by the flame velocity uflame and

8 18th International Symosium on the Alication of Laser and Imaging Techniques to Fluid Mechanics LISBON PORTUGAL JULY 4 7, 2016 (a) t = 7.4 ms t = 7.6 ms t = 7.8 ms A B (b) t = 8.2 ms t = 8.4 ms t = 8.6 ms t = 6.4 ms t = 6.6 ms t = 6.8 ms C (c) t = 6.2 ms E D Fig. 6 Tyical OH PLIF images for P = 0.1 MPa, 3000 rm (a), P = MPa, 1000 rm (b), P = ini ini ini MPa, 3000 rm (c). Fig. 7 Instantaneous fluid velocity (vectors) and flame front (red line) for P = 0.1 MPa, laminar ini (a), P = MPa, 1000 rm (b). ini

9 v fluid 1.5 fluid velocity OH PLIF signal intensity [-] x [mm] Fig. 8 Fluid velocity and normalized OH PLIF signal for Pini = 0.1 MPa, laminar. the local fluid velocity ufluid (Trunk et al 2013): u "#$%& = u ( + u "#*+(. (1) The local flame dislacement velocity is exressed by the 2D local flame dislacement seed sd and the local flame-normal vector n. u ( = n s (. (2) cross-correlation of double-ulse CH PLIF images (Tanahashi et al. 2008), but using a geometrical relation of flame and fluid velocities. In this study the local flame dislacement seed is calculated by following rocedure: (a) Calculate fluid velocity on the flame front ufluid at t = t1 from PIV results. (b) Calculate the flame-normal vector n from OH PLIF images. (c) Decide the local flame dislacement seed as the oint on the flame front at t = t1 moves to at t = t1 + dt, that is Eq(1) comleted. Figure 9 shows the robability density function (PDF) of the local flame dislacement seed. The PDF is obtained from single samle for cases of 0.1 MPa-laminar, 0.1 MPa-1000 rm and MPa-

10 (a) (b) t = 9.0 ms t = 1 ms t = 1 ms t = 7.0 ms (c) s d (d) s d t = 9.0 ms t = 1 ms t = 9.0 ms t = 1 ms s d s d Fig. 9 PDF of the local flame dislacement seed for Pini = 0.1 MPa, laminar (a), Pini = 0.1 MPa, 1000 rm (b), Pini = MPa, 1000 rm (c), (d) rm. In all cases the overall shae of PDF does not change significantly for different times, and this consistent trend that there is a little change of the flame structure. In same initial ressure the mean local dislacement seed in turbulent case (2.36 m/s for Fig. 9(b) at 8.0 ms) is larger than laminar case (1.97 m/s for Fig. 9(a) at 9.0 ms). In turbulent cases, Fig. 9(b), (c) and (d), the deviation of the local dislacement seed from its mean is larger than laminar cases, reflecting the existence of turbulent velocity across the flame fronts, which is clearly shown in Fig. 7. For the measured sets of the statistically same initial condition, Fig. 9 (c) and (d), the shae is different due to the difference in the velocity field and the structure of the flame. Figure 10 shows the PDF of the flame seed (uflame) which includes the fluid convection seed. The flame seed is evaluated just for comarison with the global roagation seed obtained by CH chemiluminescence. The mean flame seed in 0.1 MPa-turbulent case (2.97 m/s for Fig. 10(b) at 8.0 ms) almost equals to that in MPa, turbulent cases (2.59 m/s for Fig. 10(c) at 8.0 ms, 3.01 m/s for Fig. 10(d) at 8.0 ms). This result agrees with the result of CH chemiluminescence measurements. Namely, there is no

11 (a) 1.2 (b) 1.2 t = 9.0 ms t = 1 ms t = 1 ms t = 7.0 ms (c) u flame 1.2 (d) u flame 1.2 t = 9.0 ms t = 1 ms t = 9.0 ms t = 1 ms u flame u flame Fig. 10 PDF of the flame seed for Pini = 0.1 MPa, laminar (a), Pini = 0.1 MPa, 1000 rm (b), Pini = MPa, 1000 rm (c), (d). significant difference between the flame seeds at 0.1 MPa and at MPa. The difference between the local dislacement seed and the flame seed in turbulent cases is larger than in laminar case due to the existence of turbulent velocity across the flame fronts. This results indicate that it seems to be difficult to evaluate the local flame dislacement seed by sorts of line-of-sight measurement, such as chemiluminescence measurements, and simultaneous measurements of OH PLIF and PIV are needed for evaluation of flame roagation characteristics in relatively comlex turbulent flames. 4. Conclusions The newly develoed otically measurable constant volume combustion vessel is tested in the aim of alying it to the investigation into the roagation characteristics of turbulent methane-air remixed flames. Simultaneous OH PLIF and PIV at 10 khz and CH chemiluminescence are

12 erformed to measure turbulent methane-air remixed flames in the constant volume vessel under different initial ressure and turbulence conditions. Flame structure and roagation seed reflect the given turbulent statistics. There is no significant difference between the global flame roagation seeds at Pini = 0.1 MPa and those at Pini = MPa. Global flame roagation seed increases accordance with turbulence level for the turbulent conditions. The 2D local flame dislacement seed is obtained from the local fluid velocity and the flame-normal vector. The local dislacement seed for the turbulent conditions has relatively large deviation and the increase of the mean local dislacement seed relative to the laminar case seems to be well scaled with turbulent velocity fluctuation. References o Balusamy S, Cessou A, Lecordier B (2011) Direct measurement of local instantaneous laminar burning velocity by a new PIV algorithm. Ex Fluids 50: o Bonhomme A, Selle L, Poinsot T (2013) Curvature and confinement effects for flame seed measurements in laminar sherical and cylindrical flames. Combust Flame 160: o Bradley D, Haq MZ, Hicks RA, Kitagawa T, Lawes M, Sheard CGW, Woolley R (2003) Turbulent burning velocity, burned gas distribution, and associated flame surface definition. Combust Flame 133: o Groot GRA, De Goey LPH (2002) A comutational study on roagating sherical and cylindrical remixed flames. Proc Combust Inst 29: o Kerl J, Lawn C, Beyrau F (2013) Three-dimensional flame dislacement seed and flame front curvature measurements using quad-lane PIV. Combust Flame 160: o Peters N (1999) The turbulent burning velocity for large-scale and small-scale turbulence. J Fluid Mech 384: o Peterson B, Baum E, Böhm B, Dreizler A (2015) Early flame roagation in a sark-ignition engine measured with quasi 4D-diagnostics. Proc Combust Inst 35: o Tanahashi M, Taka S, Shimura M, Miyauchi T (2008a) CH double-ulsed PLIF measurement in turbulent remixed flame. Ex Fluids 45: o Tanahashi M, Taka S, Hirayama T, Minamoto Y, Miyauchi T (2008b) Local Burning Velocity Measurements in Turbulent Jet Premixed Flame by Simultaneous CH DPPLIF/OH PLIF and Stereoscoic PIV. 14th Lisbon Symosium. o Trunk PJ, Boxx I, Heeger C, Meier W, Böhm B, Dreizler A (2013) Premixed flame roagation in turbulent flow by means of stereoscoic PIV and dual-lane OH-PLIF at sustained khz reetition rates. Proc Combust Inst 34: